Poster session I
INCREASING THE SIGNAL-TO-NOISE RATIO OF ELECTROENCEPHALOGRAPHY DURING MOTOR TASKS
Martijn Vlaar, Alfred Schouten, Alistair Vardy, Frans van der Helm
Abstract: Brain activity results in small electric potentials, which can be recorded with electrodes on the skin over the scalp, i.e. electroencephalography (EEG). Current EEG systems can have 128 or more electrodes, giving an indication of the patterns and location of brain activity. Since EEG is only measured at the skin surrounding the brain, the exact origin of the measured signals remains unknown. Source localization is a technique to localize the origin of the signal inside the brain and to reconstruct the signal at the source. One of the applications of EEG source localisation is the study of neural activity of people who have suffered from stroke and, more specifically, the brain regions involved in motor control. A better understanding of how brain activity changes following a stroke potentially leads to more effective rehabilitation programs . However EEG source localization relies on a high signal-to-noise ratio at the electrodes to accurately reconstruct a source within the brain .
Conventionally, EEG is measured during evoked potentials, where typically hundreds of events (e.g. repeated button presses or finger taps) are recorded, averaged and used for source localization . Our goal is to develop a novel approach which improves the signal-to-noise ratio and, as such, improves the localization accuracy and reduces the number of required recordings.
There are several ways to increase the signal-to-noise ratio of the EEG recordings. An initial increase is achieved by using a state-of-the-art EEG amplifier with properly shielded electrodes. Our novel approach applies perturbations using a robotic manipulator while the subject performs a motor task such as maintaining a position or a target force level. The perturbation signal will be a periodic multisine signal (a signal consisting of multiple sinusoids) which only contains power at defined frequencies, allowing for an increase in power at these frequencies and therewith the signal-to-noise ratio, while the non-excited frequencies allow to estimate the signal distortion due to nonlinearities .
EFFECT OF HEAT-INDUCED PAIN STIMULI ON PULSE TRANSIT TIME
Marit van Velzen, Minke Kortekaas, Arjo Loeve, Egbert Mik, Sjoerd Niehof, Robert Jan Stolker
Abstract: INTRODUCTION Pain is usually assessed by the interpretation of behaviour of and/or reports by patients. Both methods bear subjective elements. Having purely objective pain assessment tools would greatly benefit clinical practice. Vasoconstriction is one of the physiological responses of the body to pain. Vasoconstriction changes the arterial compliance (index of arterial elasticity). The compliance can be determined using photoplethysmography (PPG).During vasoconstriction, the blood velocity increases when the blood flow is kept constant despite the reduced arterial diameter, and the arterial compliance decreases. Consequently, the time it takes for a pressure pulse wave (PW) to travel from the heart to the periphery (called the pulse transit time, PTT) decreases. The PTT is determined by taking the time between the R-wave of the ECG and the arrival of the foot of the PW in the fingertip. The PW in the fingertip can be visualized using PPG, a widely available non-invasive technique for measuring PWs with a simple photodiode and infrared light. Theoretically, the vasoconstriction in response to a pain stimulus causes a decrease of PTT. The aim of our study is to investigate if the PTT decreases in response of heat induced pain stimuli. METHODS The PTT was measured in healthy volunteers (18-35 years of age, no cardiovascular or other vasomotor function affecting disease, no limb injuries) on both index fingers, using PPG-sensors. Baseline PTT was measured for 1 minute while the subjects were sitting relaxed without moving. To test if the PTT decreases due to a vascular response to pain, the subjects received a heat-induced pain stimulus, which was applied using a temperature control block (thermode). Each stimulus started at a temperature of 32°C and rose to 50°C with a speed of 1°C/s. The temperature of the thermode was set to drop back to the starting temperature of 32°C with a speed of 10°C/s after each stimulus. To explore the best way to measure the repeatability of PTT measurements, the stimulus was applied 3 times with a time interval of 20 or 60 seconds (randomized). Afterwards, the relative PTT change between the last heartbeat before and the first heartbeat after each the stimulus was determined. RESULTS Three subjects were included for the pilot study (2 men and 1 woman, age 29 ± 4.5 years (mean ± SD), BMI 22 ± 2 kg/m²) and a mean difference in PTT of –7.7 ± 0.71% on the stimulated side was found. CONCLUSION Preliminary results suggest that PTT decreases in response to heat-induced pain stimuli. After the approval of the METC Ethical board, expected early October 2012, 20 subjects will be measured. It is expected that PTT can be used as an objective indication of pain.
IDENTIFICATION OF THE NOCICEPTIVE FORWARD PATHWAY USING AMPLITUDE-RESPONSE PAIRS
Huan Yang, Jan Buitenweg, Hil Meijer
Abstract: Malfunctioning of the nociceptive forward pathway plays a key role in the development of chronic pain, which reduces the quality of lives of the patients. To quantitatively characterize the nociceptive forward pathway, four neurophysiological parameters can be estimated by integrating computational models and multiple perception thresholds. This model-based approach could reveal the state of the nociceptive malfunctioning for understanding the development of pain, e.g. central sensitization. With suitable psychophysical procedures, one can obtain amplitude-response pairs around a perception threshold. Combining these techniques, one can first perform logistic regression to obtain a threshold from amplitude-response pairs and use that for parameter estimation.
In this work, we directly estimate parameters using the amplitude-response pairs without intermediate transformations. We study how the number of trials included influences the estimation and compare with the earlier approach. Furthermore, considering the clinical aspect, whether the pairs using fewer combinations of the temporal settings can still enough to estimate the parameters will be addressed. This work will only consider the simulated dataset to estimate the parameters, which is an essential step to further investigations with real datasets.
The estimate of the system parameters using amplitude-response pairs directly converges faster than the estimate based on the perception threshold. Such improvement of estimation could provide more reliable information for further interpretations of the state of the nociceptive system.
APPLICATION OF VIDEO THERMOGRAPHY TO INVESTIGATE SMALL FIBER NEUROPATHY
Yusang Wu, Mariska Nieuwenhoff, Frank Huygen, Frans van der Helm, Alfred Schouten, Sjoerd Niehof
Abstract: Background: Small fiber neuropathy (SFN) is a peripheral nerve disease that preferentially or selectively affects small nerve fibers and their functions. Diabetes is known as one of the major causes of SFN. Patients with SFN suffer from a combination of symptoms, including pain, numbness and vascular dysfunction. Early diagnosis of SFN and thus early treatment are crucial to prevent the development and progress of SFN, and to reduce medial costs and usage of healthcare resources. Today, however, no good test or tool is available to identify SFN in an early stage. Our project aims to develop a novel method using non-contact heat stimuli and video thermography to assess small nerve fiber function and as such to realize quantitative and non-invasive diagnosis of SFN in an early stage. In this study, the application of the method was demonstrated in healthy subjects.
Method: Eight Caucasian female subjects participated in the study (20-30 years old). The dorsal side of the subjects’ hands was heated up to 42 ℃ with a medical infrared lamp. The thermal response of the skin in the cooling phase was evaluated based on two signals: (1) the skin temperature, measured with a video thermography camera and (2) the skin blood perfusion, measured with a laser Doppler flowmetry at the center of the dorsal side of hands.
Results: After the heating the skin blood perfusion declined, and reached a steady level when the skin temperature was approximately 38 ℃. The skin temperature declined exponentially indicating a first-order process. The estimated first-order process time constant was related to the skin blood perfusion and ranged between 50 and 80 s in most measurements.
Conclusion: Recent studies reported that thermally sensitive small fibers in the skin were important in the vasodilation response to local skin warming . In our study, the comparison between the skin temperature and the skin blood perfusion demonstrated the control mechanism of small fibers. The analysis of the video thermography provided a comprehensive description of the skin blood perfusion on the dorsal side of hands. Next studies will be carried out in patients with SFN who are supposed to have a delayed/attenuated response to thermal stimuli.
 J. Johnson and D. Kellogg, “Local thermal control of the human cutaneous circulation”, Journal of Applied Physiology, Vol. 109, pp. 1229-1238, 2010
FEATURE PREPROCESSING IMPROVES SUPPORT VECTOR MACHINE ACCURACY FOR SEIZURE DETECTION IN NEONATAL EEG
Guy Bogaarts, E. Gommer, D. Hilkman, V. van Kranen, W. Mess, J. Reulen
Abstract: On Neonatal Intensive Care Units (NICU) many vital parameters are recorded but monitoring of brain function by Electroencephalography (EEG) is rare, mainly because signal interpretation requires expert visual inspection. In 1-6 % of newborns on the NICU (sub clinical) seizures occur and even more frequent in prematures and low-birth weight children . Failure of detection and subsequent lack of treatment can result in brain damage. Automatic EEG analysis could enhance the application of NICU brain monitoring.
Recently, a new seizure detection method was introduced using Support Vector Machines (SVM) . Major aim of our project is to further optimize classification accuracy by introducing two pre-processing procedures. First, a Kalman filter was used to filter feature time series in order to reduce short false detections. Second, baseline feature correction was used to reduce inter patient differences. Data from 31 newborns aged between 0 and 6 months consisted of 46 single channel routine EEG recordings (average duration of 26 minutes) in which convulsions were annotated by visual inspection. A total of 122 features for neonatal seizure detection [3-4] are computed for 10s EEG epochs. Each epoch is represented by a feature vector that can either be used to classify the epoch or for training. Feature correction is performed using either all non-seizure data for training or a 3 minute baseline for testing. Kalman filtering is performed on test data only. This results in 4 train-test combinations: 1) no preprocessing, 2) baseline correction, 3) Kalman filtering, 4) baseline correction & Kalman filtering. By applying a threshold (T), classifier produced seizure probabilities can be transformed to binary decisions. Results show that baseline correction yields a significant increase in sensitivity from 47% to 59% (p<0.05). Sensitivity is further increased to 71% when Kalman filtering is used (p<0.05 ). Both pre-processing procedures together improved classification performance from 65% to 74% (p<0.05) correctly classified epochs. Why for some individual measurements performance decreases will be subject to future research.
PREDICTION OF ICTAL PROPAGATION USING SINGLE PULSE ELECTROCORTICAL STIMULATION EARLY RESPONSES
David Keizer, Geertjan Huiskamp, Maryse van 't Klooster, Frans Leijten, Cyrille Ferrier, Michel van Putten, Maeike Zijlmans
Single pulse cortical electrical stimulation (SPES) and its responses yield information about the epileptic brain. Research has mainly focused on delayed SPES responses which are assumed to be associated with the underlying pathology. However, early SPES responses can map effective brain connectivity networks. SPES can be rapidly applied during surgery and might help to predict seizure spread without the need of an actual seizure. Accurate predictions of seizure spread may be useful for decision making in focal resections and evaluating the effectiveness of surgery. This could be specifically useful in cases when subpial transsections are preferred over resection for example when the seizure onset zone lies in an eloquent area. We aim to use clinical SPES data to assess the feasibility of predicting ictal propagation.
In this study, data from pre-surgical evaluation of 8 patients suffering from focal epilepsy with a marked gamma onset on chronic electrocorticograpy were used. SPES (bipolar monophasic 0.2Hz stimulation with 0.1ms pulse width) was applied to electrodes in the initial seizure onset zone. An automatic detection algorithm for early responses, using Wiener filtering to correct for stimulus artifacts and response amplitude offset, was developed. Based on standardized early response amplitude thresholds we predicted which electrodes could be involved in seizure spread. The resulting electrode positions were then compared to those involved in actual seizure spread as indicated by an experienced electroencephalographer. This was used as the reference for calculating sensitivity and specificity of predicted seizure spread. Furthermore, we assessed if the delay of the early responses was correlated to the timing of seizure spread for each individual patient using Spearman’s rank correlation (p<0.05).
The 8 patients analyzed in this study had 8-39 (mean 24) electrodes marked as being involved in seizure spread. SPES early responses were able to predict seizure spread to these electrodes with an average sensitivity of 67% (range 45-88%) and specificity of 88% (range 74-98%). No significant correlations between response delay and seizure spread delay were found in all but one patient (rho = 0.96, p = 0.01).
Our results show that SPES early responses can be a valuable tool to peroperatively map the epileptic brain. SPES early responses were able to predict seizure spread of the patients’ dominant seizure type based on stimulation of the epileptic focus. Future research should be performed to further improve this prediction and evaluate the clinical use.
CHALLENGES IN THE ANALYSIS OF VIDEOTHERMOGRAPHY TO ASSESS THE LOCAL VASOMOTOR RESPONSE
Mariska Nieuwenhoff, Yusang Wu, Alfred Schouten, Frans van der Helm, Frank Huygen, Sjoerd Niehof
Abstract: INTRODUCTION In the human body, small nerve fibers mediate information regarding pain and temperature discrimination, functions that are often impaired in diabetes. Small nerve fibers play an important role in local control of the vasomotor response. We are investigating the effect of impaired small nerve fibers on the vasomotor response. Local thermal stimuli are applied to the skin with an irradiation heat source to elicit a vasomotor response. The response to the thermal stimulus is recorded with thermal imaging, enabling evaluation of the thermodynamic response of the skin. Irrespective of the application area, analysis of dynamic thermal imaging is not without challenges. We will discuss results of our experiment and the challenges we encountered during data analysis.
METHODS We have included five healthy subjects, aged 20-30 years. In each subject the thermodynamic response of the skin on the hand was measured three times with videothermography. The thermodynamic response on different locations on the hand was evaluated. The analysis was performed using matlab.
RESULTS Four out of five subjects were female, the mean age was 24 ± 3 years. We calculated tau (τ) as an indicator of the decline of the skin temperature curve. The analysis of tau of all measurements had a mean tau of 59.6 s with a range of 51.2 – 70.2 s.
DISCUSSION During the analysis of our data we encountered several challenges, which can be divided into two categories; subject related challenges and method related challenges. The challenges we encountered include: the effect of different temperatures, the range of tau, and unequal distribution of heat due to the natural curve of the back of the hand, difficulty selecting a uniform and standardized area for analysis. The results and challenges will be discussed.
THE GLASGOW-MAASTRICHT FOOT MODEL- A 26 SEGMENT FOOT
Michiel Oosterwaal, Sylvain Carbes, Scott Telfer, Lodewijk van Rhijn, Søren Tørholm, Kenneth Meijer
Abstract: Standard gait analysis considers the foot as one rigid segment. While this approach is sufficient when focusing on hips or knees, studies of foot disease and injuries require more detailed models. Several multisegmental foot models have been proposed, up to 11 segments. However to our knowledge no rigid-body biomechanical foot model representing all the 26 foot bones has been developed to this day; neither exists an adequate marker protocol able to
provide the relevant motion capture data necessary to construct, validate and use this model.
In this study a 26 segments model is developed and used for inverse dynamic simulations. The model contains all intrinsic foot muscles and major plantar ligaments, as well as the needed joints and kinematic constraints to link properly the 26 segments. 25 subjects have been measured via a novel protocol, involving anatomical, biomechanical and clinical measurements. Kinematic and kinetic measurements have been done with 41 reflective markers, force plate and pressure plate. Via inverse dynamics, muscle function and joint reaction is computed using the AnyBody Modeling System (AnyBody Technology A/S, Aalborg, Denmark).
A 46 DoF foot and ankle model is developed, based on 1 subject. This model is scaled and validated for the other subjects. The kinematic model is compared with bone pin studies in literature and showed good comparison. Measured talonavicular plantar flexion ROM during gait is 10 degrees and model prediction is 9 degrees. Similarly for calcaneocuboid plantar flexion the measured ROM is 10.5 degrees and model prediction is 8.5 degrees. Kinetic
validation using measured EMG data is currently conducted.
This study shows the development of a new biomechanical foot and ankle model. Further validation will be performed using CT images of the subjects.
LI-ION CHARGER CIRCUIT FOR IMPLANTABLE PULSE GENERATOR
Rico van Dongen, Senad Hiseni, Wouter Serdijn
Abstract: Electrical stimulation of targets deep inside the brain has proven to be a powerful treatment for various movement disorders like Parkinson’s disease and essential tremor. Potentially even psychiatric disorders can be treated using electrical stimulation .
For electrical stimulation of the brain tissue, pulses are generated by an Implantable Pulse Generator (IPG). Due to the size of the IPGs used today, they need to be implanted inside the chest and connected to the stimulation electrode via an extension lead. To increase patient comfort and minimize the risk of lead failure the device should be miniaturised so that the complete system can be placed in the skull of the patient.
Since battery replacement requires surgery, miniaturisation should be achieved without compromising the current lifetime of the device. Although rechargeable batteries have a lower volumetric energy density than their non-rechargeable counterparts they can be recharged hundreds of times. By switching from a non-rechargeable IPG to a system that can be recharged transcutaneously, the volume can be reduced without reducing the lifetime.
Due to the sensitive nature of Lithium-Ion batteries they have to be recharged according to strict methods. If voltage or current limits are exceeded the battery can be damaged, reducing the capacity considerably. Traditionally Lithium batteries are charged in two phases via the Constant Current Constant Voltage (CCCV) method. In the first phase a Constant Current (CC) is pushed into the battery until the battery reaches its maximum voltage. After this a Constant Voltage (CV) is maintained across the battery and the current gradually decreases.
Although this method ensures safe and complete charging it has some drawbacks. First of all, the control is rather complex and requires two separate control loops: One to control the charge current in the CC phase and one to control the charge voltage in the CV phase. Furthermore, instability of the charger near the switchover from phase one to phase two may occur resulting in an increased charging time.
By utilizing the hyperbolic tangent transfer of a subthreshold transconductance amplifier the two charge phases of the CCCV method can be merged into a single control loop. The abrupt transition between the two charge phases is herby eliminated. The result is an inherently stable and simple control circuit .
A system that charges a battery through an inductive link has been designed. Switched mode operation has been used to further increase the power efficiency of the circuit. Chip area has been kept to a minimum by combining the charge current control switch with the AC to DC rectifier. Circuit simulations showed that the power transfer efficiency from the inductive link to the battery is, on average, 87.25% over a full charge cycle. A prototype using discrete components has been built to verify the correct operation of the circuit. Both simulations and measurements show that the charger efficiency is dominated by the loss in the rectifier circuit.
STABILITY OF SMALL ARTERIES DURING VASOCONSTRICTION AND REMODELLING: A MODEL STUDY
Ed van Bavel, Bilge Guvenc Tuna
Abstract: Organ perfusion is regulated by diameter control of small arteries and arterioles. These resistance vessels are sensitive to their mechanical and chemical environment, resulting in a range of cellular processes that ultimately affect vascular structure, mechanical properties and vascular tone (the degree of vasoconstriction).
The control loops regulating vascular function and structure are strongly interacting. As an example, an increase in pressure causes wall-stress driven distension followed by a myogenic constriction that again normalizes the stress. This action, at constant flow, increases endothelial shear stress, driving vasodilation. Such acute regulation interacts with structural responses (remodelling and hypertrophy) due to shared biomechanical stimuli and signalling pathways. The net effect of all such regulation is that resistance vessels are maintained in a functional and stable state, characterized by amongst others regulated wall stress, shear stress, matched active and passive biomechanics and sufficient vascular reserve.
This modelling study identifies four critical adaptation processes that act on the vascular biomechanics, effectuating integrated regulation of diameter: control of tone, smooth muscle cell length adaptation, eutrophic matrix rearrangement and trophic responses. Their combined action maintains arteries in their optimal state, ready to cope with new challenges, allowing continuous long-term vasoregulation. Excluding any of these processes results in a poorly regulated state and in some cases instability of vascular structure, and would eventually lead to failure of the vessel to play its role in cardiovascular homeostasis.
The current analysis adds to the comprehensive understanding of the pathogenesis of hypertension and other cardiovascular disorders, and generates testable hypotheses that could lead to new therapeutic approaches.
Supported by the FP7 Marie Curie Initial Training Network ‘SMART’
PRESSURE DROP ESTIMATION IN CORONARY ARTERIES BY FORWARD GEOMETRICAL REMODELING
Jelle Schrauwen, Jolanda Wentzel, Ton van der Steen, Frank Gijsen
Abstract: Fractional Flow Reserve (FFR) is an important indicator for the hemodynamic significance of a coronary stenosis. The FFR is defined as the ratio between the pressure distal of the stenosis and the aortic pressure under hyperemia. In clinical practice the FFR is measured with a pressure wire and intervention is warranted if it is below 0.8. With current coronary angiography it is possible to obtain the coronary geometry instantly. Therefore the goal of our research is to assess how well we can determine the FFR based on geometrical features derived from imaging data.
Sixteen coronary arteries of patients with coronary artery disease were imaged with MSCT and intravascular ultrasound (IVUS). Each artery was modeled in three consecutive steps, where in each step a geometrical complexity was added. This geometrical change could be correlated to the change in pressure drop it caused. First a tapered model was made with the mean inlet and outlet radius of the artery and its length. Secondly, in the stenosed model the lumen outline from IVUS data was used to replace the wall of the tapered model. Thirdly, a curved model was created by adding the centerline obtained with MSCT, which resulted in the complete 3D reconstruction of the artery. CFD was performed to compute the pressure drop in all these models with steady flows for Reynolds numbers ranging from 5 to 300. The computed pressure drop was compared to the pressure drop determined from geometrical features alone.
For flows with low Reynolds numbers the pressure drop in the tapered geometry could be predicted excellently with the tapering angle (r=0.99). The additional pressure drop in the stenosed geometry was captured best by the maximal degree of stenosis (r=0.84). The curvature had no significant contribution for low Reynolds numbers. In flows with high Reynolds numbers the pressure drop in the tapered and stenosed geometry could best be fitted with the product of the tapering angle and the maximal degree of stenosis (r=0.93). The additional pressure drop in the curved model correlated to the sum of the angular change along the centerline (r=0.51). Although the estimated pressure drop could deviate from the CFD results (majority within ±25%), we found an excellent correlation between the FFR from the CFD simulations and our FFR based on geometrical features (r=0.96).
With our approach of forward geometrical modeling the effects of the geometric features on the pressure drop could be determined. With these correlations we were able to predict the pressure drop and therefore the FFR in coronary arteries based solely on its geometry. However high FFR values were found (>0.85), meaning that the arteries were relatively healthy. To assess clinical relevance of the current findings we need to include more heavily diseased coronary arteries.
A VISCOELASTIC MODEL OF THE FIBRIN NETWORK IN BLOOD CLOTS
Thomas van Kempen, Gerrit Peters, Frans van de Vosse
Abstract: Blood clot formation serves to stop blood loss in case of a vascular injury but can also occur intravascularly, for instance after rupturing of an atherosclerotic plaque which may lead to myocardial infarction. The blood clot forms as a platelet plug and is subsequently stabilized by a network of fibrin fibers that forms within the clot. Fibrin increases the mechanical rigidity of the clot and is therefore of major importance for the mechanical properties of the clot . The mechanical properties of the fibrin network play a role in many diseases but are nevertheless poorly understood. Therefore a constitutive model is developed to study the mechanical properties of the fibrin network based on structural quantities such as the size of the fibers and their stiffness.
The fibrin network is modelled as a viscoelastic solid using the Kelvin-Voigt model, often represented as an elastic spring parallel to a viscous dashpot. By relating the shear modulus and viscosity of the model to the temporal fibrin concentration, the network formation is modelled as a transition from a viscous fluid to a viscoelastic solid. Model output is compared with rheometer experiments in which the network formation is followed in time by imposing an oscillatory shear deformation. Networks are formed with various initial fibrin concentrations under a range of frequencies.
The model is able to describe the viscoelastic properties of the developing fibrin network as measured in the experiments. Three free parameters are used to fit the model to experimental results. The kinetic rate constant, that governs the kinetics of the network formation, does not very with initial concentration or frequency. The mass-length ratio of the fibers increases with concentration indicating that the fibers become thicker if more fibrin is initially available. The friction coefficient, that represents viscous drag within the fibers, decreases with frequency. Using the mass-length ratio of the fibers, the model is used to estimate the length and diameter of the fibers which gives values within the expected range found experimentally . The results show that the fibrin network behaves as a viscoelastic solid, justifying the choice of the Kelvin-Voigt model. It is concluded that the model is able to describe the viscoelastic properties of the developing fibrin network.
EFFECTS OF PLAQUE MORPHOLOGY ON PEAK CAP STRESS IN HUMAN CORONARY ARTERIES
Ali Akyildiz, Lambert Speelman, Harm Nieuwstadt, Ton van der Steen, Jolanda Wentzel, Frank Gijsen
Abstract: Background: More than 70% of fatal heart attacks are caused by rupture of vulnerable plaque caps in atherosclerotic coronary arteries. From a mechanical point of view, a plaque cap ruptures when stress exceeds strength at a certain location in the cap. Plaque morphology is an important determinant of cap stresses . Previously developed models based on idealized geometries have provided very valuable insight to geometrical risk factors in atherosclerotic plaques; however, they were not able to incorporate the complex geometry of plaques completely. The aim of this study is to investigate the effects of plaque morphology on cap stresses using realistic geometries and to search for a statistical model based on geometric plaque features to predict cap stresses in human coronary plaques.
Methods: We included 77 atherosclerotic lesions from 13 human coronary arteries. Histology was applied to identify the relevant plaque components: adventicia, media, and the intima containing the lesions with the necrotic core and fibrous cap. Peak cap stresses were computed by using finite element method, including initial stress computations  and anisotropic material models. Measures of geometric plaque features for all lesions were determined and their relations to peak cap stress were examined.
Results and discussion: The study showed that peak cap stress is highly heterogeneous and is mainly affected by minimum cap thickness and maximum lumen radius. Additionally, large lumen curvature increased peak cap stresses locally. A statistical model (R=0.79), employing minimum cap thickness and maximum lumen radius, was employed to estimate peak cap stress in coronary plaques based on geometrical plaque features. Furthermore, the ratio of maximum lumen radius to minimum cap thickness enabled us to stratify coronary plaques based on peak cap stress: a ratio of 2.5 correctly identified all plaques with peak cap stress below 140 kPa, which is generally assumed to be the lower threshold value for cap strength. This ratio can be determined by intravascular imaging techniques and can possibly improve clinical rupture risk assessment for human coronary plaques.
 Akyildiz et al.,Biomedical Engineering Online, doi: 10.1186/1475-925X-10-25, (2011).
 Speelman et al., Journal of Biomechannics, vol. 44(13), pp. 2376-582, (2011).
MOTION ARTEFACTS REDUCTION IN CAPACITIVE ECG RECORDINGS USING AN INJECTION CURRENT
Aline Serteyn, Rik Vullings, Jan Bergmans
Abstract: Capacitive electrodes are an interesting alternative to conventional adhesive gel electrodes in the sense that they allow biopotential recordings through different layers of cloths. Capacitive electrodes can also be invisibly embedded in everyday objects like chairs and beds. These electrodes are however very sensitive to motion artefacts which are caused by distance variations between the electrode and the body surface. These variations induce a change in the capacitive coupling of the electrode, which affects the transfer function of the recording system. The dynamic and unknown variations of the transfer function corrupt the biopotential recordings, making them sometimes hardly usable.
The goal in this study is to dynamically track the variations of the transfer function due to motion and correct the biopotential recordings accordingly. As a first step, a model for the body/electrode interface  is adapted to our specific measurement setup, yielding a model with four parameters (i.e. coupling capacitance, coupling resistance, input capacitance and bias resistance). As a second step, a known sum of two sinusoids is injected through the system and recorded jointly with the biopotentials. The real and imaginary parts of these injected sinusoids are used to dynamically estimate the four model parameters. Once the behaviour of the transfer function is known over time, the biopotential recordings can be corrected for motion artefacts.
A MATLAB Simulink model was developed to simulate the capacitive sensor and the body/electrode interface in presence of motion artefacts. An electrocardiogram recorded with capacitive electrodes was used as an input to the system. Two sinusoids of respectively 110 and 120 Hz were injected to the system to estimate the four parameters of the transfer function. Since our model-based correction for motion artefacts shows promising results in this simulated environment, the method is currently tested in a lab setup.
Motion artefacts are not the only interferences corrupting capacitive biopotential recordings. However, a proper cancellation of motion artefacts is a first step towards a robust capacitive system enabling unobtrusive ECG or EEG recordings on dressed patients or in ubiquitous systems .
A MODIFIED CUSUM ALGORITHM FOR DETECTION OF TACHYCARDIA IN PATIENTS WITH EPILEPTIC SEIZURES
Constantin Ungureanu, Martien van Bussel, Francis Tan, Ronald Aarts, Johan Arends
Abstract: Currently there are various techniques that can be used for real-time detection of epileptic seizures. Among them, the ones based on accelerometer, heart rate changes or muscle activity show high potential to be transferred in wearable devices. Each one of these systems have its own advantages and disadvantages.
Rapid heart rate increase known as tachycardia was observed to occur in 86.9 % of focal epilepsy . These changes can precede and/or follow an epileptic discharge. Additionally, ECG sensors may provide information to detect medical complications that can lead to SUDEP (Sudden Death in Epilepsy).
At Kempenhaeghe an algorithm based on classical CUSUM method  was developed to detect rapid increases in heart rate caused by an epileptic seizure. This algorithm was evaluated off-line on heart rate data acquired from 30 patients with more than 40 seizures. Three different sensor configurations were used: one (ShimmerTM Ireland) where ECG is streamed wirelessly to computer, one where ECG data was saved on a SD card inside the sensor node (Holst Center (The Netherlands) and another one where data is saved on a portable recorder (TMSiTM).
The results of the analysis show that all nocturnal seizures were detected close to electrographical onset. The principal factors affecting the sensitivity of the algorithm were: lost packages during wireless transmission of ECG data, motion artefacts, time synchronization and electrode lose contact with the skin.
The developed algorithm is patient adaptable and relative simple to implement on sensor nodes. Based on these results, we are currently developing a real-time system for detection of epileptic seizures and alarming composed of a portable ECG sensor node and a smart-phone to be used by patients with epilepsy on a daily basis.
 F. Leutmezer , C. Schernthaner, S. Lurger , K. Pötzelberger , C. Baumgartner “Electrocardiographic changes at the onset of epileptic seizures”. Epilepsia 44: 348-54 (2003)
 E. S. Page "Continuous Inspection Scheme". Biometrika 41 (1/2): 100–115. (1954).
SIMULATING MUSCLE ACTIVITY USING THE ANYBODY MODELING SYSTEM (AMS): COMPARISON OF MEASURED EMG AND PREDICTED MUSCLE ACTIVITY PATTERNS DURING NORMAL WALKING
Adhi Wibawa, Nico Verdonschot, J.P.K. Halbertsma, M.S. Andersen, R.L. Diercks, Bart Verkerke
Abstract: This study focused on validating muscle activities predicted by the AnyBody Modeling System (AMS) against measured muscle activity (EMG) from ten healthy subjects who performed a normal walking task. The GaitLoweExtremity Model (GLEM) from AMS was used in this study. Eight EMG electrodes measured the activity of eight different muscles of the right leg: Vastus Medialis (VM), Vastus Lateralis (VL), Rectus Femoris (RF), Semitendinosus (ST), Biceps Femoris (BF), Gastrocnemius Medialis (GM) and Lateralis (GL) and Tibialis Anterior (TA). Four different thresholds were applied on both curves (predicted and measured muscle activity): 10%, 25%, 35% and 45% of the mean of the RMS envelope threshold (MRET) before they were compared quantitatively in the same threshold level. Number of onset, offset, hills and duration of muscle activity were used to quantify the level of agreement. For the parameters number of onset, offset and hills, the weighted kappa method was used. Concordance correlation coefficient analysis was used for parameter duration of muscle activity.
Visual inspection showed good agreement between EMG and predicted muscle activity. Quantifying the muscle activity by using number of onset/offset, number of hills and duration of muscle activation showed that, in general, for all parameters, the 45 % MRET showed the best agreement compared to the other MRET. For the number of onset and offset, two muscles (TA and VL) showed a fair agreement (0.20 < kappa value < 0.40) and four muscles showed a slight agreement (0 < kappa value < 0.20), the other two muscles (VL and ST) showed a poor agreement (kappa value < 0). For the number of hills, two muscles (GM and TA) showed a fair agreement and five muscles showed a slight agreement while only one other muscle (VM) showed a poor agreement. For the duration of muscle activity, all muscles showed poor agreement (concordance correlation value < 0.90). This first attempt in a quantitative point of view showed that the parameter number of hills was the best result. The differences between AMS and EMG patterns can be attributed to the nature of the AnyBody modeling process, the choice of parameters and the absence of a gold standard to compare the results with.
VALIDATION OF A MULTI-BODY MODEL OF THE BICYCLE AND RIDER TO ASSES STABILITY
Vera Bulsink, Marc Beusenberg, Bart Koopman
Abstract: Elderly people experience problems of stability while cycling. The exact nature of these stability problems is not well understood, but is believed to be related to a reduced motion feedback ability of elderly, health issues that reduce power or force, and insecurity due to slower information processing in heavy traffic or situations where multiple tasks are required.
With the use of an advanced multi-body computer model we are able to simulate the behavior of elderly on the bicycle in different problem scenario’s. This model covers the bicycle dynamics, the tire-road contact, the biomechanics and balance control of the rider, and influences from the environment. The model is fully parameterized, to be able to test the influence of various design variables of the bicycle and parameters of the rider on the behavior of the system. The final goal is to develop countermeasures to decrease problems during bicycling, like instability.
As a first validation, the self-stability of the bicycle model and passive rider model are compared to linearized models in literature [1,2]. The software ‘JBike’ calculates the eigenvalues of the system and shows the self-stability of the bicycle at certain forward speeds . The forward speed where the bicycle becomes stable is comparable between both models (around 4 m/s). The system becomes unstable at low speeds when a passive rider is added to the bicycle model, also shown by Schwab et al. . At higher speeds the system of a bicycle with a passive rider can be stable depending on the impedance of the rider’s arms on the handlebars.
Further validation of the model is needed, using experimental data. First experiments include normal cycling on a treadmill, while measuring the lean and steer angles and rates of the bicycle and the tire contact forces on the ground. Also the center of mass of the whole system can be calculated and compared to the results of the numerical simulations. The center of mass with respect to the heading of the bicycle is expected to be a way of expressing the stability of the system .
First results of the validation experiments are shown and further steps for validation of the rider movements are discussed.
3D KINEMATIC AND EMG STUDY OF THE ARM FUNCTION IN BOYS WITH DMD: A PILOT STUDY
Mariska Janssen, Jaap Harlaar, Imelda de Groot
Abstract: Introduction: Duchenne muscular dystrophy (DMD) is characterized by progressive muscle wasting and weakness, resulting in loss of functional abilities. With increasing life expectancy, the preservation of functional abilities of the upper limb becomes increasingly important. Knowledge on the way arm function is lost in DMD is needed to monitor disease progression and to design new arm supports. Therefore feasibility and validity of a new measurement protocol, using 3D kinematics and surface electromyography (sEMG) for measuring arm function in boys with DMD was examined in a pilot study.
Methods: Five boys with DMD and 6 age matched controls participated in this pilot study. Standardized (StM) and non standardized (NStM) single joint movements and ADL activities were examined by means of 3D motion analysis in combination with sEMG. Outcome measures were: normalized EMG amplitude, passive range of motion (pROM), active range of motion (aROM) and the absolute difference between pROM and aROM.
Results: All boys with DMD and controls were able to perform the measurement protocol. Boys with DMD used significantly more of their maximal muscle capacity to conduct movements compared to controls. pROM and aROM of some movements were impaired in boys with DMD.
Discussion/Conclusion: The measurement protocol was found feasible for measuring arm function in boys with DMD. The tool was able to discriminate between patients and controls and also between different stages of the disease. To gain more specific insight in the deterioration of arm function in DMD more patients should be measured. For future measurement we have made adjustments in the measurement protocol. A different kinematic model will be used which can measure scapula kinematics. More muscles are measured to gain insight in specific muscle activation patterns while the number of movement directions is reduced, and the activities in the protocol should be adaptable for different disease severities.
AN AUTOMATIC CAMERA FOLLOWING SYSTEM FOR CLINICAL GAIT ANALYSIS
Bart Klaassen, Leendert Schaake, Jaap Buurke, Martijn van Eenennaam, Bart Koopman, Hans Rietman
Abstract: Video images, combined with additional EMG and ground reaction force data, are currently widely used in daily clinical practice for clinical gait assessment. Gait event detection, like identification of initial contacts, is usually done manually from these video recordings. Video capturing with a static camera on a tripod along the side of the walking track causes the angle between the camera and the patient to change over time by panning and tilting the video camera. This complicates automatic gait event detection and gait feature extraction. A possible solution to this problem is a following camera that is usually operated manually. However, the manual handling of the camera may introduce irregularities in video recordings.
To overcome these problems, an automatic camera following system has been developed that is able to follow patients without changing the angle between camera and patient. A dolly car has been motorized and mounted on a truss rail which is parallel to the walking path of the patient (7 meters). Two cameras are mounted on the dolly car. One camera is used for the actual patient video recordings and the second camera is used for tracking the subject. A fast and reliable method for tracking the subject is by use of a selective color marker. This marker is placed on the subject. A dedicated algorithm was developed, using Matlab, to control the speed of the dolly car in a smooth and safe way. As a first validation of the system, controlled gait pattern scenarios have been carried out with three subjects and compared with Vicon motion capture system measurement data for quantifying the tracking performance. These scenarios included different walking speeds as well as irregular walking patterns commonly seen in patients at the Roessingh Research and Development centre.
Tracking validation results show that the system is able to track subjects performing slow (< 1 m*s^-1), normal (about 1.4 m*s^-1) walking speeds and irregular motion patterns with a maximum offset (difference between the centre of patient and video camera/dolly car) of 585±59 mm and a minimum offset of 20±85 mm. The maximum offset produces an angular deviation of 13±1.4 degrees between the two centers (patient at 2.5 meter distance from the camera) and still makes video images eligible for clinical observations and automatic feature extractions. Higher walking speeds (> 2 m*s^-1) have different requirements for the tracking algorithm and hardware, and need further development.
ESTIMATION OF MUSCLE ATTACHMENT SITES IN THE LOWER EXTREMITY FOR SUBJECT-SPECIFIC MUSCULO-SKELETAL MODELING
Pim Pellikaan, Vincenzo Carbone, René Fluit, Marjolein van der Krogt, Nico Verdonschot, Bart Koopman
Abstract: Introduction: To generate subject-specific musculo-skeletal models of the lower extremity, muscle attachment sites needs to be estimated with high accuracy. Therefore, the aim of this study was to develop a morphing-based method to automatically predict the location of muscle attachment sites in the lower extremity, based on the assumption that their position is in direct relation with the geometry of the bone.
Methods: Two cadaver dissections were performed to measure the contour of all muscle attachment sites on the bones of the lower extremity. CT scans of both cadavers were used to segment the bones using Mimics software (www.materialise.com). STL-files containing the geometry of the bones and the measured muscle attachment sites were morphed from one cadaver to the other and vice versa. Muscle attachment sites were divided into three types: point, line and surface area, following the Twente Lower Extremity Model (TLEM) . To determine the accuracy of the morphed attachment sites, a distance error was calculated as the mean Euclidean distance between: the mean of the morphed and un-morphed points in case of attachment points, the pairs of 6 equidistance points on a third order polynomial line in case of attachment lines, the pairs of 6 equi-area points on the projected surface area in case of attachment areas.
Preliminary results: The mean distance error for all attachment sites was 13.31 mm ± 8.08 mm for cadaver 1 and 12.69 mm ± 7.95 mm for cadaver 2. The smallest mean distance error was 1.33 mm for the origin area of the medial part of the Gluteus Maximus and the largest mean distance error was 43.40 mm for the origin area of the Soleus Lateralis.
Discussion: The proposed method shows reasonable average error (<15mm) between measured and morphed attachment sites. Large distance errors can be partially explained by differences in shape of attachment sites between the two cadavers. In particular, differences in length along the long axis of the bone had a large influence on the calculated distance error. The potential effect of these errors on subject-specific musculo-skeletal model prediction should be quantified in the future. In conclusion, the proposed method showed promising results in estimating muscle attachment sites, and further improvements may help to reach the accuracy needed to obtain reliable subject-specific musculo-skeletal models.
Significance: Automatic estimation of the location of muscle attachment sites, based on medical imaging techniques, represents an important step in order to generate reliable subject-specific musculo-skeletal models.
SELF-PACED VERSUS FIXED SPEED TREADMILL WALKING
Lizeth Sloot, Marjolein van der Krogt, Jaap Harlaar
Abstract: Gait research increasingly involve instrumented treadmills, since they offer new experimental possibilities and simplify continues measurement. However, walking on a treadmill is known to effect gait performance, presumably due to the imposed walking speed. A feedback-controlled treadmill that follows so called self-paced walking would be a good alternative to allow for small, natural variation in walking speed. The purpose of this study was to compare the effect of different self-paced (SP) walking algorithms on spatiotemporal parameters, joint kinematics and kinetics, and compare the effect of SP versus fixed speed (FS) treadmill walking.
Eighteen healthy subjects (12 male, age 29±4) walked on a dual-belt instrumented treadmill in a speed-matched virtual environment (Gait Real-time Analysis Interactive Lab (GRAIL), Motek Medical B.V., the Netherlands). During SP, the treadmill speed was regulated by a PD controller, with velocity dependent gains with additional correction for sudden slowing down (SP1), a standard PD controller with the same correction (SP2), and a PD controller with position dependent differential gain and improved correction for deliberate de- and acceleration (3). After 10 minutes of habituation, subjects walked for three minutes at SP1 and speed-matched FS, followed by SP walking in each mode in random order. A Vicon system measured 25 markers and joint kinematics were calculated following the HBM model. For each subject, both the mean stride and stride variation were calculated. The difference in joint kinematics or kinetics were expressed as the average offset or gain and the offset or gain corrected root mean square (RMS) value, thus indicating a difference in pattern. The effect of SP-mode was evaluated using an ANOVA, whereas SP and FS were compared with paired t-tests.
Variance of walking speed as well as stride length were slightly decreased in the SP2 mode (p=0.03 and p=0.01), whereas kinematics and kinetics were comparable between modes. During SP, walking speed varied considerably over time (p<0.001), while mean stance percentage was slightly reduced (p<0.001). There were no relevant differences in kinematics or kinetics between SP and FS treadmill walking.
The results show that gait performance was relative insensitive for the specific SP mode, although it seems that the altered PD controllers (SP1 and SP3) allow for more variability in walking speed. This walking variability was also increased compared to constraint FS walking, whereas other gait measures did not show relevant significant differences. Therefore, SP walking seems a good experimental alternative to simulate over ground walking in clinical gait analysis. Additionally, the application of SP allows for studying fatiguing and pathological gait variability. Further research should focus on whether SP better resembles over ground walking than FS treadmill walking.
MAXIMAL HAND FORCE ESTIMATION WITH AN MRI-BASED SUBJECT-SPECIFIC MUSCULOSKELETAL MODEL
Bart Bolsterlee, Frans van der Helm, DirkJan Veeger
Abstract: Introduction One important error source of musculoskeletal models is a mismatch between the anatomy of the model (mostly derived from cadaver studies) and the subject or patient that is analysed. This hampers the applicability of these models above the level of general applications, or “what if” questions. A subject-specific model can in principle be used on more specific, patient- or subject-related questions, but requires the model’s anatomy to be adapted to fit the subject. Personalization of the muscle’s physiological cross-sectional area (PCSA) is expected to decrease differences between model predictions and experimental data.
Methods Of five subjects with strongly different builds, the maximal isometric force they could exert in six different directions on a handle that was gripped by the right hand with the elbow 90° flexed, was measured. The right shoulders of the same five subjects were MRI-scanned. By manually outlining muscle boundaries and summing the volume of the selected voxels, muscle volumes were obtained. Default PCSA values (measured on a single cadaver) that are used in the Delft Shoulder and Elbow Model (DSEM, ) were scaled with the ratio of muscle volume between subject and default model, divided by the ratio of muscle lengths. Two subject-specific PCSA-sets were made, namely one that uses the same volume scale factor for all muscles, and one that uses different scale factors for different muscles. For all five subjects and six force directions and with both the default and two subject-specific PCSA sets, the lowest value of maximum muscle stress (σmax, maximum force a muscle can produce per area of cross-section) for which the DSEM could reproduce the recorded force, was calculated. σmax is generally considered to be constant across muscles and individuals.
Results With the default DSEM, a value of σmax=94.9 N/cm2 ±32.2 was required to reproduce the measured forces with the model. A lower and more consistent value across subjects resulted after uniform scaling, σmax=62.9 N/ cm2±4.9. Muscle-specific scaling did not lead to a significant difference with respect to uniform scaling, σmax=63.4 N/cm2 ±4.3.
Discussion σmax varied substantially across subjects when using the default model (standard deviation 32.2 N/cm2). After uniform scaling, σmax estimation was much more consistent (62.9 N/cm2 ±4.9), indicating that PCSA scaling improves the fit of model predictions on experimental data. Because the relative strength per direction is dependent on relative PCSA values between muscles, muscle-specific scaling was expected to lead to a more consistent σmax across force directions but this was not the case. Other factors such as inter-individual differences in muscle moment arms might be of more influence. In conclusion, a musculoskeletal model with for each muscle individualized PCSA values can predict the maximal hand force consistently across different subjects, but does not perform better than a model of which the PCSA values are scaled by a single factor.
 A.A. Nikooyan et al., Development of a comprehensive musculoskeletal model of the shoulder and elbow, Med Biol Eng Comput 49, pp. 1425-35, (2011).
RESPONSE OF A BIOACTIVE AND BIODEGRADABLE COMPOSITE
Zhiguang Huan, Sander Leeflang, Jie Zhou, Jurek Duszczyk
Abstract: In recent years, research and development of advanced biomedical materials have been directed toward biodegradable materials, namely metallic, polymeric or ceramic biodegradable materials in order to avoid long-term medical complication and removal surgery. Among metallic biodegradable materials, a lot of attention has been paid to magnesium as a potential biodegradable material, especially for orthopaedic applications, because of its biodegradability in the bioenvironment as well as its relatively low Young’s modulus . However, magnesium degrades too fast in physiological solutions. Alloying may effectively reduce its degradation rate, but it may also introduce cytotoxic elements. Moreover, magnesium itself lacks bioactivity.
As an alternative to adding alloying elements to magnesium, adding bioactive agents to magnesium to form magensium matrix composites may offer a way to reduce the degradation rate of magnesium and enhance its bioactivity . It was hypothesized that suitable bioactive agents added to magnesium might induce the formation of a surface mineral layer that could protect it from fast degradation and enhance its surface bioactivity. In this study, a composite was prepared from pure magnesium and bioactive glass powders through the powder metallurgy route. Pure magnesium prepared under the same conditions was used for comparison purposes. Fig. 1 shows a comparison in microstructure between pure magnesium and the composite, as well as the distribution of bioactive glass particles in the composite.
Fig. 1 Optical micrographs of extruded Mg (a) and Mg-BG (b) rods on the transverse section; SEM micrograph of the Mg-BG composite on the transverse section of extruded rod (c); the inset shows the topography at a higher magnification.
Immersion tests in the E-MEM solution were performed to determine the response of composite samples to the simulated physiological solution over a period of six days in terms of weight loss percentage and hydrogen evolution rate. Results are shown in Fig. 2.
Fig. 2 Weight loss percentage (a) and hydrogen evolution (b) from Mg and Mg-BG composite samples as a function of time during the immersion tests in the E-MEM solution.
Post-immersion surface morphology of Mg-BG sample in comparison with that of Mg was characterized by means of SEM and EDX (Figs. 3 and 4). For pure magnesium, the surface was covered with small particles (Fig. 3a), and exhibited a cracked corrosion layer beneath these particles. For the Mg-BG composite, however, SEM image at a higher magnification (Fig. 3b inset) did not show the presence of a cracked corrosion layer. The corrosion layer on the Mg-BG composite appeared to be more compact than that on pure magnesium. The elemental compositions of the corrosion layers on magnesium and Mg-BG samples, determined by EDX, are shown in Fig. 4. The EDX results suggested that Mg(OH)2 was the main constituent of the corrosion layer on pure magnesium (Fig. 4a), in addition to a small amount of Mg3(PO4)2 . The presence of the element Ca was not confirmed. By contrast, the EDX analysis showed that apart from magnesium, oxygen and phosphorus, a small amount of calcium was present in the corrosion layer on Mg-BG sample surface after 1-day immersion (Fig. 4b). This confirmed that the surface layer of the composite contained a higher concentration of Ca than that of Mg sample.
Fig. 3 SEM micrographs showing the surfaces of Mg (a) and Mg-BG (b) samples after immersion in the E-MEM solution for one day. The insets show the corrosion products at a higher magnification. The arrows point to the cracks in the oxide layer.
Fig. 4 EDX analysis of the surfaces of Mg (a) and Mg-BG composite (b) samples after immersion in the E-MEM solution for one day.
In summary, immersion tests of a magnesium matrix composite containing bioactive glass particles in the E-MEM solution showed that the Mg-BG composite had a lower weight loss and hydrogen evolution rate than pure magnesium, which could be attributed to Ca and P deposition on composite sample surface, induced by partial dissolution of BG into the immersion solution. The study demonstrated the feasibility of decreasing the degradation rate of magnesium and enhancing its bioactivity by adding bioactive glass particles.
1. Staiger M.P., Pietak A.M., Huadmai J., Dias G., Magnesium and its alloys as orthopaedic biomaterials: A review, Biomater. 2006;27:1728-34.
2. Witte F., Feyerabend F., Maier P., Fisher J., et al., Biodegradable magnesium-hydroxyapatite metal matrix composites, Biomater. 2007;28:2163-74.
3. Bender S., Goellner J., Atrens A., Corrosion of AZ91 in 1N NaCl and the mechanism of magnesium corrosion, Adv. Eng. Mater. 2008;10:583-587.
SENSITIVITY AND SPECIFICITY OF ULTRASOUND IN DETECTING (OSTEO)CHONDRAL DEFECTS: A CADAVERIC STUDY
Aimee Kok, Maaike Terra, Juri Aaftink, Geert Streekstra, Niek van Dijk, Gino Kerkhoffs
Abstract: A chondral or an osteochondral defect (OCD) is described as the separation of a fragment of cartilage without or with damage of the underlying subchondral bone, respectively 1. The long term prognosis of small cartilage lesions evolving into an eroding subchondral bone defect is not known. Longitudinal monitoring of patients at short intervals could assist in overall understanding. As ultrasound imaging offers the preconditions for short interval monitoring (noninvasive, fast and affordable) 2,3, the aim is to determine the feasibility of ultrasound in diagnosing the presence or absence of particularly small chondral and osteochondral defects.
In the anterior talar surface of ten human cadaveric ankles, a maximum of four conditions were arthroscopically created, consisting of a randomized combination of: no defect, pure chondral defect (size Ø 3 mm or Ø 1.5 mm), and osteochondral defect (size Ø 3 mm or Ø 1.5 mm). All ankles were examined with ultrasound by two blinded observers who indicated the presence, location, classify and size of the defects. Ultrasound observations were validated using CT scans and photographs of the dissected tali.
The overall sensitivity was 96% for Observer 1 and 92% for Observer 2 with a specificity of 100% for both. Overall correct grading of the defects was 92% for Observer 1 and 79% for Observer 2. Both observers only located one defect incorrectly resulting in 96% correct localization. 68% of all measured defect sizes by Observer 1were within the clinically relevant limits of agreement (-0.2 ± 1.0 mm) and 79% of those measured by Observer 2 4.
Ultrasound has the potential to facilitate accurate identification and localization of small (osteo)chondral defects on the anterior surface area of the talus.
ARM PROSTHESIS CONTROL: PERCEPTION OF CABLE ACTIVATION FORCES AND DISPLACEMENTS
Mona Hichert, Dick Plettenburg
Body powered arm prostheses require too high operating forces. Prosthetic use is found tiresome or even painful. The required operating forces need to be lowered. The ideal prosthesis should be powered by cable operation forces and displacements which can be invariably perceived by the user and do not lead to pain or fatigue. Earlier research showed good perception in a force range between 20 and 30 N at fixed cable displacement. The question remains: With which cable force and displacement should a prosthesis be operated when also taking into account cable displacements?
A prosthesis simulator was fitted to 30 subjects without arm defect. Instead of a prehensor an interchangeable spring was placed at the end of the control cable. The cable forces were measured with a force sensor located close to the shoulder harness. Cable displacement was calculated though the known spring constant and the measured cable forces close to the spring. Cable force and displacement were fed back to a laptop running a LABVIEW programme. The subject was requested to reproduce a given force and hold it constant for 2 seconds. Visual feedback was enabled every second repetition. Nine different combinations of forces and displacements were measured.
The smallest replication error (reproduced minus reference force) was found between 24 and 33 N. For every spring an inverse relationship between cable displacement and replication error was found.
Discussion & Conclusion
The smaller the replication error the better the perception. Since in this experiment the smallest replication error (and therefore the preferred force level for prosthesis control) was found for cable forces between 24 and 33 N, the finding of the prior research seems to be confirmed. Perception of cable displacement seems to be dependent on spring constants and is better at larger spring deformations.
A CHEAP AND NOVEL TECHNIQUE OF WRIST AND HAND MEASUREMENTS FOR HAND AND WRIST ORTHOSES
Serdar Ates, Israel Mora Moreno, Piet Lammertse, Arno Stienen, Herman van der Kooij
Abstract: Hand and wrist orthoses (also known as exoskeletons) have been used for several years for the rehabilitation of stroke patients. Recent development in robot-mediated rehabilitation has shown the potential of robotic devices that are delivering repetitive training thus allowing for a large number of repetitions to be delivered during acute and chronic phases of stroke rehabilitation. There is growing evidence that such technologies are beneficial to patient’s recovery of functional and motor outcome. It is crucial to measure some values such as angular position/velocity, force/torque applied on the links. This information is used as feedback in control algorithms of orthosis. It is also a quantitative information for the physiotherapist to monitor patients’ recovery progress. It is very difficult to provide a cheap and affordable solution which does not impede the natural movements of human hand in order to measure the angular position of fingers and thumb. The number of degree of freedom of the hand excluding wrist is 20. Besides, the space is very limited to put both mechanical and electronic hardware for the measurement. All these facts make human hand anatomy very complex to be integrated with robotic devices. In order to overcome these difficulties, simple sensors such as flex sensors which are simply variable resistors according to angle displacements are used. These measurements are used in an approximation algorithm in order to estimate required angular positions and force values. Lower price and relatively less complexity allow them to be used as a cheap and affordable solution.
EVALUATION OF INPUT DEVICES FOR THE CONTROL OF ACTIVE MOVEMENT-ASSISTIVE DEVICES
Joan Lobo-Prat, Peter Kooren, Arno Stienen, Bart Koopman
Abstract: Active movement-assistive devices can increase independence and quality of life of patients with severe neuromusculoskeletal disorders . These systems should not operate autonomously  and therefore require interaction between the patient and the device for its control. A large variety of physiological signals (control inputs) can be measured from the human body, which can be used for the control of the active movement-assistive device: neural signals from the central nervous system (e.g. electroencephalography (EEG)), neural activation of the muscle (e.g. electromyography (EMG)), muscle interaction forces, body movements, neural signals from the physiological sensors (e.g. electroneuronography (ENG)) among others. In addition, signals can be measured from systems parallel to the assisted motor system (e.g. eye, head or tongue), which can also provide valuable input for the control of the active movement-assistive device. The selection of the body signal to control the assistive device according to each user needs and capabilities (which may change over time) is crucial for the usability and proper function of the assistive system. A large variety of control strategies have been proposed and it is unknown which strategy is the most suitable for each type of impairment and task.
This study presents a method based on the early work of Duane T. McRuer and co-workers  to evaluate and compare the performance of healthy subjects during a screen based one-dimensional tracking task using several control inputs. The subject is asked to follow with a cursor a reference point (target) that moves according to an unpredictable multi-sine signal. The reference signal is composed of 10 sinusoidal signals between 0.1 and 3Hz with amplitudes that decrease logarithmically with frequency. In addition, the cursor has dynamic properties that resemble the ones of the human arm. The measured input signals from the different devices/sensors are normalized and implemented using three different control strategies (i.e. position, speed and force) to control the cursor.
With the proposed method we will be able to carry out an objective and quantitative evaluation and comparison of the performance of healthy subjects during a screen based one-dimensional tracking task using several control inputs. The final objective of this evaluation is the selection of the most suitable control inputs (or combination of them) to operate an active upper extremity orthosis that is being developed within the Flextension project.
 Dollar, A.M., and H. Herr (2008) IEEE Transactions on Robotics 24-1, pp. 144 –158.
 Veltink, P.H., et al. (2001) Archives of Physiology and Biochemistry 109-1, pp. 1–9.
 McRuer, Duane T. (1965) Air Force Flight Dynamics Laboratory, Research and Technology Division, Air Force Systems Command.
A HYDRAULICCALLY ACTUATED PROSTHETIC HAND
Gerwin Smit, Dick Plettenburg
Abstract: Problem: Rejection rates of body-powered hand prostheses are high (26-45%).1 Current body powered hands are inefficient. They require an uncomfortable high activation force, and produce a relatively low pinch force in return (<15 N).2 They have stiff fingers, which do not adapt to the shape of the grasped object. Despite all these drawbacks, the design of body powered hand prostheses hardly has improved since the 1950’s.3 The activation force has not been reduced and the pinch force is still low. Another major problem is the mass of the devices, which is still high (≥350 gram). There have been attempts to increase the efficiency of body powered hand prostheses, by using hydraulics. However, these studies have not resulted in the commercial application of hydraulics in body powered arm prostheses.
Goal: The goal of this study was to design a new body-powered, voluntary closing, hand prosthesis, which has articulating fingers. This hand should have a much lower mass than current prosthetic hands, be energy efficient and be able to provide sufficient pinch force.
Results: A low mass (152 gram) articulating hydraulic hand was constructed and tested. The hand has 7 active DoF’s which are activated by miniature hydraulic cylinders inside the hand. The hand is controlled by a shoulder strap, connected to a miniature master cylinder. The hydraulic system enables an efficient energy transmission from the master cylinder to the slave cylinders at the finger joints (n=8). This enables a high pinch force (>30 N) and a low user effort. The hand successfully passed the mechanical and functional tests.
Discussion and conclusion: The new developed hand prototype with articulating fingers is anthropomorphic, slender, fast, efficient and silent. The hand mass is much lower than the lightest commercially available prosthetic hand. The hand therefore meets one of the most important user demands in upper limb prosthetics, which is a low hand mass.
Significance: Its low mass, high efficiency and easy operation makes the new prosthetic hand useful for a large group of upper limb amputees. The light and compact design enables fitting on a very long or very short residual arm. The new prototype might therefore also be an option for people that cannot be fitted with current available prosthetic hands.
MODELING AND CONTROL OF A POWERED UPPER EXTREMITY EXOSKELETON
Carsten Voort, Alexander Otten, Arno Stienen, Ronald Aarts, Herman van der Kooij
Abstract: Diagnostic robots provide objective information by measuring joint torques, velocities, positions and intrinsic muscle properties. This information can be used to modify training programs or to diagnose patients.
This paper focuses on the modeling and control of the Limpact exoskeleton, which will be used for diagnostics. A non-linear rigid dynamic model of the Limpact exoskeleton is derived using 20-sim. By means of state reduction the model is reduced to its four independent degrees of freedom. The resulting model is used to design a CTCL (computed-torque control law) which linearizes, stabilizes and controls the non-linear dynamic system. Simulations and experiments are shown for three different control structures: PD control, PD control with gravity compensation, and CTCL.
The simulations and experiments show the performance of the three control structures on path tracking and steady state errors. PD control with gravity compensation clearly outperforms the other two control structures.
NATURAL ARM SUPPORT THAT MEETS THE GROWING NEEDS OF DUCHENNE PATIENTS
Gerard Dunning, Peter Kooren, Just Herder
Abstract: People with Duchenne Muscular Dystrophy (DMD), affecting approximately 1 in every 3500 live male births , gradually lose the ability to use their muscles. During the first years of the childhood, the larger muscles deteriorate. In a later stage, the smaller muscles deteriorate also. Most of the patients will be confined to a wheelchair before they are 10 years old, because the large upper leg muscles are affected. When the upper arm muscles deteriorate, the patients lose the ability to use their arms during activities of daily life (ADL). Consequently, they become highly dependent on caregivers. Furthermore, most of them encounter psychological struggles, due to restricted participation in social activities [2, 3].
With increasing life expectancy, the preservation of functional abilities for people with DMD becomes increasingly important. To compensate for the loss of arm function, an arm support can be used. This helps the patient to lift their arm, so the patient does not have to generate the muscle strength to overcome gravity. With such an external device, they become more independent, and are able to participate in social activities. In order to encourage the usage in daily life, it should be inconspicuous and give a natural support .
The aim of this project is to develop a wearable assistive device that supports the arm function during ADL and is inconspicuous: in the ideal case it fits underneath clothing.
Assistive devices that balanced the arm have been reviewed . There are some devices available that support the arm function, but these devices are not always adequate. None of the devices support the pronation/supination of the forearm. Most devices are highly stigmatizing, due to their large volume, which is not close to the body. Almost all of them are mounted to the wheelchair. While a larger device is acceptable for training activities, a wearable device is desired in ADL . Opportunities for a future design are to use the combined center of gravity of the whole arm as interface point between the body and the device. Another interesting feature is to balance the forearm and upper arm independent of each other.
DEVELOPMENT OF A LIFTING AID WITH HAPTIC SHARED CONTROL FOR DEXTEROUS HEAVY WEIGHT LIFTING IN INDUSTRY
Arvid Keemink, Arno Stienen, J.F. Schorsch, D.A. Abbink, Frans van der Helm, Herman van der Kooij
Abstract: In industrial settings there is the challenge to lift heavy objects from machine to machine. The negative aspects of current industrial weight support systems - lack of flexibility and operation speed - makes employees ignore the systems and perform the tasks manually. Some of these employees start experiencing physical problems in the elbow, shoulder and lower back after years of work. This motivates development of new flexible and easy lifting supports.
The goal of this project is to realize a breakthrough in physically support humans with heavy lifting, by designing a shared-control assistive human-machine interface that is fast to get into, intuitive and user-friendly. The aim is to show that by using haptic shared control, the force to the operator can be designed based on his/her motion control properties, leading to more intuitive assistance. Using the philosophy of force feedback from shared control to avoid objects in the surroundings, inherently provides stability, damping and smooth movements.
The full system is still under development. Lifting will be facilitated by using a two-part system; a lifting system to generate the supporting force and a dexterous manipulation device for intuitive dexterous manipulation of the load. Force feedback signals from a slave gripper will give proper grasp sensation and slippage information to the operator. A master device designed for the human fingers and hand will display these forces to the operator to increase awareness and increase lifting functionality and stability. Designs will be such that the system will be of minimal necessary complexity and degrees of freedom (i.e. in the slave fingers and master device) to be robust and powerful in the lifting task and the handling of the load.
Actuation of the lifting system is an important challenge, often exoskeletons become bulky because of the actuators required to generate the power needed to lift heavy objects. To tackle this, an additional innovation is envisioned: to use balanced springs to mechanically compensate for some of the load and to offer a highly dynamic and flexible load support.
THE INFLUENCE OF AN ARM SUPPORT MECHANISM ON UPPER LIMB JOINT MOMENTS: AN INVERSE DYNAMIC ANALYSIS
Alessio Murgia, Hans Essers, Paul Verstegen, Kenneth Meijer
Abstract: Objective. To investigate the moments at the glenohumeral and elbow joints while performing daily tasks with and without using an arm support mechanism.
Background. Arm support mechanisms are used to assist the movements of subjects with diverse impairments. However, the joint moments generated when using these devices during daily tasks have scarcely been investigated. Understanding how the moments change can lead to better designs by considering factors such as power, fatigue and joint loading.
Methods. One healthy male participant (25 ys) performed a cyclic task consisting of reaching from the midline to a target located on his ipsilateral side at shoulder height and one shoulder width. The movement was performed with and without the SLING arm support mechanism (Focal Meditech BV). Movements were captured with an 8-camera Vicon system at 200 Hz. The SLING support force was set to give the impression of moving in a gravity-free space. Joint angles were calculated using Vicon’s inverse kinematic pipeline. An inverse dynamic analysis was also performed with the AnyBody Modeling System (AnyBody Technology). Output parameters were the net joint moments at the glenohumeral and elbow joints.
Results. During the unassisted movement all the glenohumeral moments peaked when reaching the target, abduction was the highest at about 7 Nm. The elbow flexion moment varied between 1.5 and 3 Nm, reflecting the forearm’s cyclic movement. When the SLING was used, the glenohumeral abduction moment was reduced by 1 Nm, its pattern almost unchanged, while the external rotation and flexion moments were reduced by about 1.5 Nm and their patterns also changed as a result of the upward force applied by the SLING. The pattern of the elbow flexion moment remained unaltered but was reduced by about 0.5 Nm.
Discussion. The moments calculated during unassisted movement were within the ranges reported in the literature for healthy subjects . Using an arm support mechanism alters the moment patterns and has implications for fatigue and joint loading. At present the SLING is used to assist subjects with muscular dystrophy; future research will be focussed on using the approach described here to design better support mechanisms for this target group.
 M. Nordin, M. H. Pope, and G. Andersson, Musculoskeletal Disorders in the Workplace: Principles and Practice, Mosby Inc., 2007.
SUBJECT SPECTIFIC FES BASED ASSISTANCE FOR REHABILITATION OF REACH, GRASP AND RELEASE AFTER STROKE
Ard Westerveld, Alexander Kuck, Alfred Schouten, Peter Veltink, Herman van der Kooij
Abstract: Stroke often has a disabling effect on the ability to functionally use the hand. Control of finger and thumb positioning is necessary for many activities in daily life. Functional Electrical Stimulation (FES) can assist patients in relearning movements after stroke [1,2]. Recently, possibilities for activating individual fingers by FES have been explored, showing the need for an individualized approach . We evaluated the feasibility of subject specific approaches for positioning the thumb and fingers for grasp and release of differently sized objects. Assistance based on these approaches may be used in rehabilitation after stroke.
A model predictive controller (MPC) was compared with a proportional (P) feedback controller . Both methods were compared on their performance in tracking reference trajectories and in the capability of grasping, holding and releasing objects. Both methods were able to selectively activate the fingers such that differently sized objects, selected from the Action Research Arm test, could be grasped in healthy subjects. The MPC method gave better results and was easier to use in practice, as this method was based on a single identification of a model of the biological system. The P-controller had more parameters which need to be set correctly, and therefore needed more time to initialize. In more severely affected stroke patients, the controller performance in selective activation of fingers was limited, but was still applicable for grasp and release of larger objects.
The results of the subject specific control approaches are promising, especially for the MPC. Future research will include optimizing the method for applicability in rehabilitation, by reducing the time needed for set up and initialization and improving the performance even further. In addition, combination with robotic assistance of reaching is planned. This hybrid approach will lead to a novel functional training environment for reach, grasp and release after stroke which is able to adapt its assistance to the needs of the individual patient.
VELOCITY-DEPENDENT REFERENCE TRAJECTORY GENERATION FOR ROBOTIC GAIT SUPPORT
Bram Koopman, Edwin van Asseldonk, Herman van der Kooij
Abstract: Introduction: During the last decade robotic gait training devices have been increasingly used as a clinical tool to provide neurological patients with task oriented, high intensity and repetitive training. These traditional robotic gait trainers are now expanding to mobile systems, that can be used outside the clinic and that can be used for several applications, such as rehabilitation, personal assistance or even human augmentation. Although a broad variety of control strategies exist for these devises, for most applications some sort of reference pattern, in order to determine the amount of assistance, is still required. These patterns are often based on pre-recorded trajectories from unimpaired volunteers. The major limitation of these patterns is that they are not publically available. Additionally, most patterns are recorded at a limited number of speeds, while the progress of the patients’ preferred walking speed can be as small as 0.1 km/h. In this study we will present a new method of constructing normative angular trajectories at different speeds.
Methods: 15 elderly subjects walked on a treadmill at 7 different speeds, ranging from 0.5 to 5 km/h. Their angular trajectories (abduction/adduction of the hip and flexion/extension of hip and knee) were parameterized by defining different key events, which consisted of a selection of extreme values in position and velocity data. For each joint 6 key events were selected. Each key event was parameterized by an index, representing the percentage of the gait cycle at which the key event occurred, and its position, velocity and acceleration. Finally, the walking speed and body height dependency, of the parameters, were determined by regression models. To create the required subject- and speed dependent reference pattern, spines are fitted between the predicted key events.
Results: For most of the key events, the index, position, velocity and acceleration were dependent on the walking speed. The body height contributed to the predictability of the regression models to a lesser extent. The results showed that the reconstructed reference patterns fitted the measured data well. The root mean square error (RMSE) between the reconstructed trajectories and the actual joint trajectories (averaged across subjects, joints and different walking speeds) was 2.6 degrees. As a reference; the RMSE between the right- and left actual joint trajectories was 2.1 degrees, indicating that the prediction error is close to the natural variation between left and right leg.
Conclusion: In this study we derived and provided regression models that can be used to reconstruct patient-specific joint angle trajectories, based the subjects body height and walking speed. This will enable therapists, patients or users of robotic gait applications to easily change their gait speed, without the need to manually adjust the reference patterns.
LESS IS MORE: EFFICIENT IDENTIFICATION OF HUMAN STANCE CONTROL WITH A PARAMETRIC SUBSPACE METHOD
Denise Engelhart, Alfred Schouten, Ronald Aarts, Herman van der Kooij
Abstract: The human balance control system is a complex mechanism with many underlying neural feedback loops. In order to identify the stabilizing mechanisms generated by the central nervous system during upright stance, perturbations and closed-loop system identification techniques (CLSIT) are required . Many CLSIT used to investigate human balance control are based on non-parametric methods in the frequency domain, relating the signals with the external perturbations. Although, these non-parametric CLSIT give insight about specific dynamic behavior of a human, they do not directly quantify the underlying physiological parameters of the feedback loops and do not take advantage of the common structure between the signals and the perturbations.
As an alternative identification technique, we have adopted the Prediction Based Subspace Identification (PBSID) method . This method has several theoretical advantages for use in closed loop data, as with the central nervous system controlling the human body. With subspace techniques a parametric model structure is obtained, relating to physiological parameters, without a priori assumptions about the underlying model structure. In addition subspace identification is well suited for Multiple-Input-Multiple-Output (MIMO) systems; in case of the human stance model the contribution of both ankle and hip joints can easily be incorporated for a more realistic representation of the human body.
To test the PBSID subspace identification method for balance control, simulations were performed in Matlab. The human body is modeled as a double inverted pendulum, incorporating an ankle and hip joint, with a stabilizing mechanism (controller), activating dynamics and neural time delays. In the simulations two independent continuous perturbations are applied at hip and shoulder level and multiple realistic levels of measurement noise were simulated.
The simulations demonstrate that subspace identification estimates the stabilizing mechanism at least as good as a non-parametric CLSIT. The PBSID algorithm provides consistent estimates, also in case of short sample lengths. Furthermore, the method is robust against measurement noise, which can reduce experimental measurement time in humans. Reduction of experimental time is an advantage especially in pathological stance. A drawback of the method is that consistency depends on proper model order selection, which can be difficult in case of high noise. In short, subspace is a good alternative over nonparametric methods, implicitly handles MIMO systems, and can deal with short measurement time.
 H. v.d. Kooij, Comparison of different methods to identify and quantify balance control, Journal of NeuroScience methods, 2005, 145(1); p, 175-203
 J.W. v. Wingerden, Control of Wind Turbines with ‘Smart’ Rotors: Proof of
Concept & LPV Subspace Identification, Delft University of Technology, 2008
NOVEL INSTRUMENT FOR VIBRATION EXERCISE STUDIES
Lin Xu, Chiara Rabotti, Massimo Mischi
Abstract: Vibration exercise (VE) has been suggested as an effective option to improve muscle strength and power performance. However, the underlying mechanisms are still unclear and information for the most appropriate VE protocols is limited. In this study, a new VE system is realized and characterized that will enable electromyography (EMG) studies on neuromuscular activation and reflex in the upper limbs.
QUANTIFICATION AND LEARNING EFFECTS OF MEDIOLATERAL BALANCE PERFORMANCE USING A VISUAL TRACKING TASK.
Luis Eduardo Cofré, Mirjam Pijnappels, Jaap van Dieën
Abstract: It is thought that mediolateral balance control impairment may affect performance of important daily life activities such as walking, especially in the elderly [1, 2]. It has also been proposed that balance training in the frontal plane may reduce the incidence of falls in the elderly population . Quantification and training of ones capacity to dynamically control mediolateral balance might therefore be a powerful tool in fall prevention. For this purpose, a zero order visual tracking task (VTT), using the centre of pressure (CoP), which demands the integration of multiple sensory inputs, is proposed. Balance performance during either a predictable or a pseudorandom VTT can be described in the frequency domain in terms of phase shift, and gain. The aim of this study was to determine whether learning effects occur in these tasks. .
For this experiment, 20 healthy young subjects (28±3 yrs) stood barefoot on a force plate (Kistler) and performed a series of 4 VTTs with a predictable target and 4 with a pseudorandom target, preceded by 2 practice trials for each task. The VTT consisted of tracking a target signal on a screen in front of the subject, by a projection of the ML displacement of the CoP. The frequency of the target increased from 0.3 to 2.0 Hz. For the predictable target, the frequency increased with steps of 0.1 Hz every 5 seconds. For the pseudorandom target, multisines with a bandwidth of 0.6 Hz and lowest frequency increasing from 0.3 to 1.5 Hz at 0.1 Hz were used. D-Flow (Motek Medical, The Netherlands) was used to produce target signals and project CoP feedback (spheroids). Performance was expressed as the gain and phase-shift between the target and CoP projection. Repeated measures ANOVAs were performed to assess learning effects at frequencies from 0.3 To 2.0 Hz, with an alpha of 0.01.
For the predictable task, significant learning effects were found at 1.5 and 1.7 Hz in phase shift, and at 0.3-0.4 and 1.1-1.2 Hz in gain (all p<0.01). For the pseudorandom target, significant learning effects were observed for phase shift at all frequencies analyzed (p<0.01), except at 0.3 and 0.4 Hz, while gain exhibited a significant learning effect only at 0.4 and 0.5 Hz.
Overall, close to optimal values and no major learning effects were found over the range of frequencies for the predictable task, indicating that physical constraints do not limit performance in CoP tracking up to 2 Hz. When comparing both tasks, larger phase shifts and lower gains in pseudorandom tracking indicate that information processing may be a limiting factor. For the pseudorandom task, the phase-shift decreased substantially over sessions. This may indicate the information processing required in the pseudorandom task can be improved with a single session and the effect can be maintained up to one week
1. O'Connor, S.M. and A.D. Kuo, Direction-Dependent Control of Balance During Walking and Standing. Journal of Neurophysiology, 2009. 102(3): p. 1411-1419.
2. Maki, B.E., P.J. Holliday, and A.K. Topper, A prospective-study of postural balance and risk of falling in an ambulatory and independent elderly population. Journals of Gerontology, 1994. 49(2): p. M72-M84.
3. Waddell, C.L., R.B. Mellifont, and B.J. Burkett, Improving balance in community-dwelling older people through a targeted mediolateral postural stability program. Journal of the American Geriatrics Society, 2009. 57(12): p. 2380-2382.
LOW-BACK STABILIZATION DURING TRANSLATIONAL AND SWING-LIKE PERTURBATIONS
Paul van Drunen, Yorick Koumans, Riender Happee
Abstract: Low-back pain (LBP) is a common disorder, which affects 40-60% of the adult population annually in Western Europe and North America . Motor control deficits (e.g., delayed reflex responses and increased co-contraction) have been suggested as potential cause and/or effect of a-specific chronic low-back pain and its recurrent behaviour [2, 3]. In a previous study, motor control identification was achieved for low-back stabilization . However, preload biased the results due to the reduced involvement of the abdominal muscles.
In this study, six healthy subjects were seated on a movable platform while restrained at the pelvis. Upper body sway was evoked by moving the platform in the sagittal plane, while low-back & trunk kinematics and back & abdominal muscle activity were recorded. Subjects were asked to perform a relax task or a stay upright task. Two perturbations types were applied: anterior-posterior horizontal translations and swing-like rotations around the virtual rotation point in the lumbar spine. For the stay upright task, we hypothesized a modulation in the neuromuscular control due to perturbation type; the intrinsic contributions are dominant during the translational perturbations, while reflexes contribute more during the swing-like perturbations. During the relax task, no big differences are expected for both perturbations.
Coherences were high for the EMG measurements of both the back and abdominal muscles, indicating involvement of both muscle groups. In the translational task, increased co-contraction levels were found for the stay upright task leading to a decreased compliance.
In future studies, this method will be applied to determine the neuromuscular control of LBP-patients, so potential modulations in reflexive and intrinsic contributions could be identified.
“This research is supported by the Dutch Technology Foundation STW, which is the applied science division of NWO, and the Technology Programme of the Ministry of Economic Affairs.” See www.neurosipe.nl - Project 10732: QDISC
 P. L. Loney and P. W. Stratford, “The Prevalence of Low Back Pain in Adults: A Methodological Review of the Literature”, Phys.Ther., Vol. 79, pp. 384-396, (1999).
 J. Cholewicki, A. P. D. Simons and A. Radebold, “Effects of external trunk loads on lumbar spine stability”, J.Biomech., Vol. 33, pp. 1377-1385, (2000).
 A. Radebold, J. Cholewicki, G. K. Polzhofer and H. S. Greene, “Impaired Postural Control of the Lumbar Spine Is Associated With Delayed Muscle Response Times in Patients With Chronic Idiopathic Low Back Pain”, Spine, Vol. 26, pp. 724-730, (2001).
 P. van Drunen, E. Maaswinkel, F.C.T. van der Helm, J.H. van Dieën and R. Happee, “Identifying Intrinsic and Reflexive Contributions to Low-Back Stabilization”, Under Review in J.Biomech.
PRESSURES REQUIRED FOR WATER JET DISSECTION IN MINIMALLY INVASIVE HIP REFIXATION TESTED IN INTERFACE TISSUE SUBSTITUTES
Gert Kraaij, Steven den Dunnen, Jenny Dankelman, Rob Nelissen, Edward Valstar
Abstract: Introduction: Worldwide about 1.5 million hip prostheses are implanted annually and this number is growing due to the longer life expectancy of people . Within the first ten years approximately 10% of the hip prostheses fail because of aseptic loosening . An alternative for the current invasive revision surgery, which requires complete removal of the loosened prosthesis and insertion of a new prosthesis, is a new Minimally Invasive Hip Refixation (MIHR) procedure. An important step in this procedure is the removal of the soft peri-prosthetic interface tissue. In a previous study laser and coblation were evaluated , however heat generation was an issue. This can be solved by using a coolant. But why not using the coolant itself, for water jet cutting, to cut the tissue? In medical field, the current applications of water jet are limited, but it is already used for liver resections. In experimental examinations Rau et al  found a pressure of 3-4 MPa (30-40 bar) and a nozzle diameter of 0.1 mm to be very effective to dissect normal liver tissue. The effect of a water jet on a material is influenced by the pressure and nozzle diameter combination. The aim of this pilot study is to evaluate the effect of the nozzle diameter on the required minimum water jet pressure to perforate soft elastic materials with different thicknesses.
Materials and Methods: Water jets with nozzle diameters of 0.1, 0.2 and 0.6 mm were applied to specimens with thicknesses of 3, 6 and 9 mm and consisting of interface tissue substitutes: ballistic gel and silicone (Dragon Skin Pro). A specimen was placed under the water jet cutting head and by trial and error the minimum pressure was found to perforate the specimen. A low starting pressure was chosen and the water jet was activated for 2 seconds. At a new spot the water jet was activated again while the pressure was increased with 0.5 MPa. This procedure was repeated until the water jet pressure was high enough to perforate the specimen. To be sure the right minimum pressure was found, we performed three additional trials respectively with the minimum pressure and minimum pressure minus 0.5 MPa bar.
Results: In Figure 1 the resulting minimum pressures for ballistic gel and silicone are shown. Minimum pressures for silicone were higher compared to those for ballistic gel. Pressure was (linear) increasing with increasing specimen thickness. A larger nozzle diameter resulted in a lower minimum water jet pressure. The largest drop in minimum pressure was found between a 0.1 and 0.2 mm nozzle.
Discussion/Conclusion: This pilot study shows that using larger nozzles results in reduced minimum water jet pressures to perforate soft elastic material. However, research with human interface tissue is necessary to find the optimal combination of nozzle diameter and associated minimum water jet pressure for interface dissection. Currently we are collecting human interface tissue to test water jet dissection.
 R.J. Looney et al., Arthritis Research and Therapy, 4:59–63, 2002.
 H. Malchau et al., The wellcemented total hip arthroplasty: Theory and practice, chapter 11, pages 291–299. Springer Berlin Heidelberg, 2005.
 Kraaij et al., Medical Engineering & Physics, 34:370-377, 2012
 Rau et al., HPB , 10: 275-280, 2008
THREE-DIMENSIONAL ASSESSMENT OF BILATERAL SYMMETRY OF THE RADIUS AND ULNA FOR PLANNING CORRECTIVE SURGERIES
Joy Vroemen, Iwan Dobbe, Remmet Jonges, Simon Strackee, Geert Streekstra
Abstract: The contralateral unaffected side is often used as a reference in planning a corrective osteotomy of a malunited distal radius . Two-dimensional radiographs have proven unreliable in assessing bilateral symmetry, so we assessed 3-dimensional configurations to assess bilateral symmetry .
We investigated bilateral symmetry using 3-dimensional imaging techniques. A total of 20 healthy volunteers without previous wrist injury underwent a volumetric computed tomography of both forearms. The left radius and ulna were segmented to create virtual 3 dimensional models of these bones. We selected a distal part and a larger proximal part from these bones and matched them with a mirrored computed tomographic image of the contralateral side. This allowed us to calculate the relative displacements (∆x, ∆y, ∆z) and rotations (∆x, ∆y, ∆z) for aligning the left bone with the right bone segments. We investigated the relation between longitudinal length differences in radiuses and ulnas.
Relative differences of the radiuses were (∆x, ∆y, ∆z): -0.81±1.22 mm, -0.01±0.64 mm, and 2.63±2.03 mm; and (∆x, ∆y, ∆z): 0.13°±1.00°, -0.60°±1.35°, and 0.53°±5.00°. The same parameters for the ulna were (∆x, ∆y, ∆z): -0.22±0.82 mm, 0.52±0.99 mm, 2.08±2.33 mm; and (∆x, ∆y, ∆z): -0.56°±0.96°, -0.71°±1.51°, and -2.61°±5.58°. There is a strong relation between absolute length differences (∆z) between the radiuses and ulnas of individuals.
We observed substantial length and rotational differences around the longitudinal bone axis in healthy individuals. Surgical planning using the unaffected side as a reference may not be as useful as previously assumed. However, including the length difference of the adjacent forearm bones can be useful in improving length correction in computer-assisted planning of radius or ulna osteotomies and in other reconstructive surgery procedures.
Awareness of te level of bilateral symmetry is important in reconstructive surgery procedures where the contralateral unaffected side is often used as a reference for planning and evaluation.
A NON-FUSION SCOLIOSIS CORRECTION SYSTEM
Martijn Wessels, Edsko Hekman, Bart Verkerke
Abstract: The rigid nature of the majority of current implantable scoliosis correction systems leads to several negative side-effect. One effect is vertebral fusion, resulting in a rigid section of the spine. Another negative result is that once the implant is fixed to the spine further growth is arrested. We designed two flexible implants with the specific objective to correct idiopathic scoliosis without introducing fusion. The XS-LAT is a system made of NiTinol, which essentially provides a lateral bending force. The XS-TOR, made of a high grade Titanium alloy, essentially provides axial torsion. The correction process is force-driven: in fact, the spine is “guided” into growing straight again. Another characteristic feature of both implants is that they can slide in the anchors which connect them to the spine. Since fusion is prevented, functional bending of the spine and growth are largely unaffected.
The systems have been tested by scoliosis induction in a porcine animal model. Objective was to determine the degree of deformation which the implants could induce, and to affirm that no fusion would occur. Six animals received the XS-LAT system while six other animals received the XS-TOR system. The implants were placed posteriorly and fixed to the T12, T15 and L2 vertebrae. Care was taken to avoid periosteal damage during surgery. Radiographs were taken preoperatively, peroperatively and 1, 4 and 8 weeks post-operatively. After 8 weeks the animals where sacrificed and analyses were performed investigating fusion, growth, tissue response, actual induced deformation and status of the implant.
The XS-TOR induced a mean torsion of 12° between T12 and T15 and of 7.1° between T15 and L2, but only limited lateral deformation (1°). The mean Cobb-angle induced by XS-LAT was 19°, whereas induced torsion was 5°. There were no signs of obstruction of growth or decrease of kyphosis. Mean length progression of the implants in 8 weeks was 19.5%. In all pigs, some heterotopic ossification at different levels was observed, as well as cartilage degeneration in the facet joints of the instrumented vertebrae T12, T15 and L2. However, fusion of joints was not observed anywhere, and intervertebral discs showed no ossification. Tissue response to the implant was characterized as normal.
We tested the implants is a scoliosis induction model, the only animal model available. A study by Meijer [Meijer, 2010] indicates that correction results may be different but not necessarily worse. We found that the system is capable of inducing significant deformities in two different directions without inhibiting spinal growth. The implants effect the intended directions, suggesting that splitting up the functionality of the implants in bending and torsion enables patient specific tuning of the corrective actions. With respect to fusion, this test can be considered as worst-case since bone formation response in young animals is always more severe than in humans [Eitel, 1981]. Long-term behaviour of the implant, including osseo-induction remains unknown.
 G.J.M. Meijer: Development of a non-fusion scoliosis correction device: numerical modelling of scoliosis correction, PhD-thesis, Enschede 2011
 F. Eitel, F. Klapp, W. Jacobson, L. Schweiberer., “Bone regeneration in animals and in man.” Arch Orthop Trauma Surg. 1981;99(1):59-64.
ADVANCED IMAGING METHODS FOR BREAST ULTRASOUND
N. Özmen-Eryilmaz, Nicole Ruiter, Koen van Dongen
Abstract: Breast cancer is the most frequently diagnosed cancer and the leading cause of death for women worldwide. Fortunately, death rates are decreasing as a result of early detection via screening large populations and improved treatment. Today, mammography is the most common technique for breast cancer examination. However, it can miss cancers for women with dense breasts. The main reason is that both dense tissue and cancerous lumps show up white on mammograms, making it difficult to differentiate between them.
Breast ultrasound is gaining interest as an alternative to mammography due to its potential to detect cancer in dense breasts. In addition, it is safe, fast and cost-effective. Recently, researchers have been working on building fully automated three-dimensional breast ultrasound scanning systems  and new ultrasound imaging methods. In principle, breast imaging is a nonlinear inverse problem. However, the problem is often linearized by applying approximations such as the Born approximation. Unfortunately, as a result of these approximations, breast images get blurred and show incorrect reconstructions.
Our research project aims to develop advanced ultrasound imaging techniques that allow for accurate three-dimensional reconstruction of the breast in terms of acoustic medium parameters, i.e., speed of sound and attenuation. We overcome the problems coming along with the linearization by going beyond the Born approximation, and solve the actual nonlinear inverse problem using a Contrast Source Inversion (CSI) method . In this conjugate gradient based numerical method, a cost functional is minimized iteratively and contrast sources and contrast functions are updated, simultaneously. Results using synthetic measured data show that our method indeed yields sharper reconstructions as compared to the common applied methods. Moreover, it provides accurate speed of sound profiles indicating the location and dimension of a tumour.
 N.V. Ruiter, G.F. Schwarzenberg, M. Zapf, R. Liu, R. Stotzka and H. Gemmeke, “3D Ultrasound Computer Tomography: Results with a Clinical Breast Phantom”, IEEE Ultrasonics Symposium, (2006).
 P.M. van den Berg and R.E. Kleinman, “A contrast source inversion method”, Inverse Problems, Vol. 13, pp. 1607–1620, (1997).
SENSITIVITY OF HEAD AND NECK HYPERTHERMIA TREATMENT PLANNING TO OBSERVER VARIATIONS IN 3D PATIENT MODEL GENERATION
Rene Verhaart, Valerio Fortunati, Jifke Veenland, Theo van Walsum, Wiro Niessen, Gerard van Rhoon, Maarten Paulides
In the hyperthermia (HT) unit of the Daniel den Hoed cancer center in Rotterdam an applicator for heating deeply seated head and neck (H&N) tumors has been developed and clinically integrated . The treatment is given to stimulate the effectiveness of the concurrent chemo or radiotherapy treatment. Accurate hyperthermia treatment planning (HTP) is needed to optimize thermal dose delivery. The treatment plan is made patient specific by delineating the tissues in the H&N area by hand. Up to now, the sensitivity of HTP to (inter and intra) observer variations in deliniation was unknown.
Three trained medical radiation technologist delineated 17 different tissues on computed tomography (CT) images of 7 patients. The patient group was chosen such that it represents the tumor types normally treated with H&N HT. Inter and intra observer variations were quantified using the Dice similarity coefficient (DC) and the mean distance (MD). The sensitivity of HTP was quantified by determining the two sd error of the average specific absorption rate in the tumor (SARIEEE-1g, tumor) at 1 W input power.
The inter observer variation (DC: 0.7, MD: 0.8 mm) was higher compared to the intra observer variation (DC: 0.8, MD: 0.5 mm). The two sd error of the average SARIEEE-1g, tumor was 0.015 W/kg at 1 W input power.
In this sensitivity study, we showed that a variation in tissue delineation leads to a small change in the SARIEEE-1g, tumor. To minimize difference in HTP it is advised to use a well defined delineation protocol. Furthermore, the development of automatic delineation algorithms will minimize HTP differences.
 Paulides, M.M. et al., Int. J. Hyperthermia, 2007, 23(7):567-576
AN ADDITIVE INSTANTANEOUSLY COMPANDING NEURAL READOUT SYSTEM FOR COCHLEAR IMPLANTS
Cees-Jeroen Bes, Wouter Serdijn, Jeroen Briaire, Johan Frijns
Abstract: Major Cochlear Implant manufacturers have included the possibility of recording neural responses. However, the possibilities are severely restricted due to the occurrence of saturation in the single channel amplifier and analog to digital converter (ADC), and the relative high noise levels. This is most clearly illustrated by the fact that objective neural thresholds are mostly found at the upper end of the subjective electrical dynamic range (Hughes, Brown, Lopez and Abbas, 1999). Recording on these relative high levels has as major drawback that different neural waveforms originating from different fibre populations are combined (Briaire and Frijns, 2005). Potentially the neural response data, thresholds, but also the spread of excitation and neural recovery functions, could provide insight in what the optimal stimulation strategy should be, and how to program the current levels of the implant for individual patients. Especially in very young children this should lead to increased performance. Researchers are now confronted with the limitations of existing neural response readout systems needed for reading out the evoked compound action potential (eCAP). These limitations urge the need for a new neural response readout system having a dynamic range of 126dB, that is small, low noise, power efficient and can handle input signals exceeding the supply voltage. Existing techniques do not offer solutions to meet the above specifications. An overall readout system design is proposed containing an additive instantaneous companding input system, multiplexer, compensation circuit, amplifier and an ADC in order to record the eCAPs from the stimulated auditory nerve.
ANTIBACTERIAL TiO2-Ag SURFACES AND THEIR CYTOTOXICITY TO HUMAN OSTEOBLAST CELLS
Bogdan Necula, Lidy Fratila-Apachitei, J.P.T.M. van Leeuwen, S.A.J. Zaat, Iulian Apachitei, Jurek Duszczyk
Abstract: The most devastating complications encountered in total joint arthroplasty are implant-associated infections. Next to patient trauma, the treatment of these infections accounts for high healthcare costs. Therefore, implants that hold an antibacterial function to prevent bacteria colonization to the implant surface might be a suitable solution to this clinical problem. The aim of this study was to test the in vitro bone cell viability and antibacterial efficacy of Ag-bearing oxidized surfaces prepared by Plasma Electrolytic Oxidation (PEO) process on the medical grade Ti6Al7Nb alloy.
The oxidized surfaces were produced in electrolytes containing calcium acetate/calcium glycerophosphate salts and two different concentrations of Ag nano-sized particles (0.3 and 3.0 g/L) as bactericidal agent. The morphology of the oxidized surfaces was investigated by Scanning Electron Microscopy. Simian Virus Human Fetal Osteoblast (SV-HFO) cells viability was quantitatively determined using the Alamar Blue assay after 2, 5 and 7 days of culture. The in vitro antibacterial activity of the TiO2-Ag surfaces against methicillin-resistant Staphylococcus aureus (MRSA) was assessed using the direct contact assay, specifically developed to mimic the conditions of an infection of a primary total joint replacement .
The surfaces produced by PEO revealed a porous structure consisting of both amorphous and crystalline TiO2 phases. Ag nanoparticles were found both on the surface and inside of the pores. In vitro viability tests showed SV-HFO cells proliferation after 2, 5 and 7 days on TiO2-0.3Ag and TiO2 (Ag-free) surfaces without notable differences between them. However, the TiO2-3Ag surfaces inhibited cell proliferation at all time-points. In vitro antibacterial testing proved excellent MRSA killing rates after 24 hours for both TiO2-0.3Ag and TiO2-3Ag samples with efficiencies of 98.25% and >99.75%, respectively. On the TiO2 control surface the number of CFU increased 1000-fold.
From the two different surfaces tested, the one produced with low concentration of Ag nanoparticles showed both bone cell viability and bactericidal activity. The findings suggest that the PEO process may be an appropriate surface modification technology for bone implants allowing synthesis of cytocompatible TiO2 surfaces with antibacterial function.
 B.S. Necula, L.E. Fratila-Apachitei, S.A.J. Zaat, I. Apachitei, J. Duszczyk, “In vitro antibacterial activity of porous TiO2-Ag composite layers against methicillin-resistant Staphylococcus aureus”, Acta Biomater., Vol. 5, pp. 3573−3580, (2009).
ENERGY HARVESTERS: IDENTIFYING POWERFUL FREQUENCIES OF HUMAN MOTION
D.I. Beekers, V.J.M. Gast, E.J. Peuscher, J.Q. Rusman, N. Tolou
Abstract: There is a hidden source of energy in our everyday motion, yet not fully explored. A small amount of the mechanical energy from human motion can be captured and converted into electric energy, using energy harvesters. This offers the possibility of lifelong power supply for implantable medical devices, such as hearing-aids, pacemakers and deep-brain-stimulation. However, in most cases, energy harvesters only perform efficiently if the resonance frequency of the harvester is adjusted to the frequency of the human body motion. Therefore the goal of this research is to identify the frequencies and body locations at which most of the human motion occurs and the largest power can be harvested. The measurement equipment consisted of several smart phones to measure the acceleration in three orthogonal axes, defined with respect to the human body. The data acquisition was performed on four individuals, subject to different daily activities and at different body locations. The activities were classified into specific activities, such as walking and cycling, and short day-to-day activities, such as having a coffee break or doing groceries. The smart phones were located at the wrist, upper arm, waist and ankle for all activities. The resulting data was processed in MATLAB using the Welch function to identify the frequencies and the related power density. We have shown that the highest power at which the body moves can be harvested between 0 and 11 Hz. The highest power density was measured during cycling in the frequency band of 0 to 1 Hz, in the direction from inferior to superior, with the device located at the ankle. This research contributes to further development of the use of energy harvesters in medical devices.