5th Dutch Bio-Medical Engineering Conference 2015

15:00 Poster Session I

AN ACTIVE TENDON DRIVEN STEERABLE NEEDLE WITH SHAPE SENSING
Nick van de Berg, John van den Dobbelsteen
Abstract: ABSTRACT Needle insertions are common practice in the field of interventional therapy. Used needles range from 14-25 G in size, ~0.5-2.1 mm in diameter. Quite regularly, needles deflect during insertion, complicating a precise placement. This can lead to additional tissue damage due to re-puncturing [1], misdiagnosis (false negatives), poor dosimetry, and tumor seeding [2]. To overcome these placement difficulties, active needle steering has been an ongoing research topic for the past decades. This work presents a new needle steering approach, during which a conical needle tip with an apex angle of 20°, placed on top of a miniature ball-joint, can be actively rotated with two orthogonal degrees of freedom. Tendons are fixed to the tip and run along the flexible cannula to the needle base. Four servo motors, working in complementary pairs, are used to actuate the tip. A linear stage allows for a constant insertion motion at 5 mm/s. During insertion, a steered tip will cause asymmetric tip-tissue interaction forces that will induce cannula deflections. The needle stylet is equipped with optical strain sensors (fiber Bragg gratings) to measure and reconstruct these deflections. Automated experiments were performed by means of implementing a PI controller to act on the reconstructed tip error with respect to a set target. This system was evaluated in a 15 m% gelatin, with targets selected in the eight principal steering directions (every 45°). Per location, six insertions were performed. The insertion depth was 100 mm and the targets were set at a lateral distance of 30 mm with respect to the straight insertion line. A ninth target was set at the same depth on this straight line. The targeting accuracy was determined by the in-plain error between the tip and target. The error per target (mean ± S.D.) was calculated and the average of these values over all nine targets is determined. This was 6.2 ± 1.4 mm (mean ± std). The steering precision is defined as the in-plain error between the tip and average position reached. The precision was 2.6 ± 1.1 mm. This presents how closely steering actions converge to the same spot, irrespective of systematic effects. With a relatively easy control scheme, we demonstrate how this needle is able to steer to various predefined targets. The curvature and direction of steering can be altered by selecting a different tip angle, without the need for cannula rotations. REFERENCES [1] L. B. Kratchman, M. M. Rahman, J. R. Saunders, P. J. Swaney, and R. J. Webster, “Toward Robotic Needle Steering in Lung Biopsy: A Tendon-Actuated Approach,” Proc. Med. Imag. Visualization, Image-Guided Procedures, and Modeling, vol. 7964, Feb, 2011. [2] S. P. DiMaio, and S. E. Salcudean, “Needle steering and model-based trajectory planning,” Proc. Med. Image Comput. Comput. Assist. Interv., vol. 2878, pp. 33-40, 2003.

FETAL ST MONITORING: WHY IT DOES NOT WORK IN CLINICAL PRACTICE
Rik Vullings, Olenka Hulsenboom, Kim Verdurmen, Guid Oei
Abstract: Introduction Fetal ST analysis was once believed to be the holy grail in fetal monitoring. First studies on fetal ST analysis reported a reduction of fetal metabolic acidosis and unnecessary interventions [1]. Later studies could no longer confirm these initial findings [2]. One of the main problems seems to be the poor specificity of the ST alarms of the currently employed monitor (STAN®, Neoventa Medical, Sweden) [3]. Hypothesis Based on our longstanding experience with non-invasive fetal ECG recordings, we hypothesized that variations in the orientation of the fetal electrical heart axis could explain for the poor specificity of the ST alarms. In our hypothesis, a heart axis oriented such that the electrical activity of the repolarization phase of the heart is almost parallel with the aVF lead, would yield many ST alarms. Similarly, when the activity of the repolarization phase is almost perpendicular to the aVF lead, virtually no ST alarms would be generated. Methodology and results We tested our hypothesis in three steps. In the first step, using simultaneous ultrasound examination and non-invasive fetal ECG recordings, we confirmed that inter-patient variability in the orientation of the electrical heart axis is significant. In the second step, we defined a measure to represent the orientation of the electrical heart axis and showed that the number of ST alarms generated highly depends on this measure. In other words, we confirmed our hypothesis. In the third step, based on our hypothesis, we proposed a solution to reduce the incidence of false ST alarms and evaluated our solution in a small test group. This group consisted of 20 fetuses, 10 of which were born with severe metabolic acidosis and 10 of which were born healthy. Within this group, the STAN® monitor obtained a sensitivity of 90% and a specificity of 40% in terms of correct ST alarms. In the same group, our proposed solution yielded a sensitivity of 90% and a specificity of 100%. References [1] Amer-Wahlin, I., et al. Cardiotocography only versus cardiotocography plus ST analysis of fetal electrocardiogram for intrapartum fetal monitoring: a Swedish randomised controlled trial. Lancet 358 534-8 (2001). [2] Schuit, E., et al. Effectiveness of electronic fetal monitoring with additional ST analysis in vertex singleton pregnancies at >36 weeks of gestation: an individual participant data metaanalysis. Am J Obstet Gynecol 208 187e1-13 (2013). [3] Kwee, A., et al. Occurrence of ST-changes recorded with STAN® S21-monitor during normal and abnormal fetal heart rate patterns during labour. Eur J Obstet Gynaecol Reprod Biol 135 28-34 (2007).

DESIGN OF AN ANKLE FOOT ORTHOSIS FOR PEOPLE WITH PARETIC ANKLE MUSCLES
Dymphy van der Wilk, Roy Reints, Klaas Postema, Juha Hijmans, Bart Verkerke
Abstract: Patients with paretic ankle muscles are unable to actively dorsiflex and/or plantarflex which hampers walking. A dorsiflexion moment is necessary to control loading response and to provide toe clearance during swing phase of walking[1]. A plantarflexion moment is necessary to control tibial progression during midstance and to provide plantarflexion moment during push-off[1]. To compensate for impaired dorsi- and/or plantarflexion, patients often use an ankle foot orthosis (AFO). However, current AFOs also have adverse effects on intact ankle range of motion (ROM). Hampering ankle ROM causes problems during activities in which more ROM is needed than allowed for by the AFO. One example is walking stairs[2]. A novel AFO was designed to take over impaired muscle function without hampering intact ankle ROM. To take over impaired muscle function, this novel AFO was designed to control both loading response and tibial progression. During these two phases, energy can be stored. This energy will be released during push-off to provide a plantarflexion moment and during swing to provide a dorsiflexion moment. A first prototype of the novel AFO was developed according to the ‘methodical design process of biomedical products’[3]. This prototype consists of a 3D printed mechanical joint, two pressure sensors (placed on heel and forefoot), an Arduino board, a servo motor and a shank and foot shelve that connect the novel AFO to the leg of a person. The two pressure sensors register heel and/or forefoot pressure and pass this information via the Arduino board to the servo motor that is connected to the 3D printed joint. Dependent on the walking phase or type of activity (recognized by heel and/or forefoot pressure), the servo motor switches between dorsi- or plantarflexion motion. The next steps in the design process are (i) to perform a clinical trial in which the correct functioning of the novel AFO is evaluated, (ii) to evaluate functioning in a larger group of subjects, (iii) to evaluate functioning in a home-based setting. REFERENCES [1] Perry J, Burnfield JM. Gait analysis: Normal and pathological function. Slack; 1993. [2] Radtka SA, Oliveira GB, Lindstrom KE, Borders MD. The kinematic and kinetic effects of solid, hinged, and no ankle-foot orthoses on stair locomotion in healthy adults. Gait Posture. 2006;24(2):211-218. [3] Verkerke GJ, Houwen EB van der: Design of biomedical products. In: Biomaterials in modern medicine: the Groningen perspective. Ed by Rakhorst G, Ploeg R. World Scientific Publishing, Singapore; 2008:23-38.

A NOVEL ULTRASOUND TECHNIQUE FOR DETECTION OF OSTEOCHONDRAL DEFECTS IN THE ANKLE JOINT
Nazli Sarkalkan, Arjo Loeve, Koen van Dongen, Gabrielle Tuijthof, Amir Zadpoor
Abstract: (Osteo)chondral defects (OCDs) in the ankle are currently diagnosed using computed tomography (CT), magnetic resonance imaging (MRI), and arthroscopy techniques that are not convenient to use in long-term follow-up of individuals. Ultrasound (US) imaging is a cost-effective and non-invasive alternative to these diagnostic techniques, but has limited ability to discriminate OCDs. In this research, we aim to develop a new diagnostic technique based on the propagation of US waves through the entire joint space of the ankle. In this technique, the presence of OCDs is identified by measuring the deviation of a US signal from a reference signal associated with the healthy state of the joint. This new diagnostic technique intends to benefit from the advantages of US in extensive follow-ups while not suffering from the limitations of conventional US techniques. In this study, we investigate the feasibility of the proposed technique using a simplified model of the ankle joint, and use experimentally-validated 2D finite-difference time-domain (FDTD) models of the ankle joint. Effects of dominant variables relevant to the ankle joint (i.e. joint space width), US transducers (i.e. translation and rotation of US transducer acting as transmitter) and defect (i.e. width, depth, and location) on US propagation were evaluated. The results suggest that the new technique could be used for detection of OCDs provided that the effects of other parameters (i.e. parameters related to the ankle joint and US transducers) on US waves propagating through the joint space of the ankle can be reduced.

INTEGRATED ASSESSMENT OF MOTOR (DYS)FUNCTION IN STROKE AND PARKINSON’S DISEASE PATIENTS (NEURAS-IMA)
Paulina Bank, Johan Marinus, Jurriaan de Groot, Jacobus van Hilten, Carel Meskers
Abstract: Disorders like stroke and Parkinson’s disease [PD] may affect the brain, senses and musculoskeletal system. A complex interplay between primary lesion, secondary changes, compensatory movement strategies and, in PD, drug-dependent fluctuations and progression of the disease may underlie the observed phenotype of motor performance. Interventions aiming at improvement of functional activities require proper quantification of motor, sensory and cognitive factors as well as their integrative behaviour during systematic manipulation of task conditions and environmental factors. The aim of this study is to develop a paradigm for uniform, integrated assessment of motor (dys)function of the upper limb in stroke and PD patients. A one degree-of-freedom haptic robot is used for high precision assessment of motor action and neuromechanical properties of the wrist in a variety of mechanical and visual environmental conditions. A markerless motion tracking device (based on KinectTM and ThreeGearTM) is used to record functional unrestricted movements of the arm and hand. A large LED screen is used for presenting tasks in a visual environment and/or providing real-time visual feedback on a subject’s actual motor output using D-Flow software (Motek Medical B.V., Amsterdam, The Netherlands) [1]. Reliability and validity of the developed paradigm will be assessed in 57 stroke patients, 57 PD patients, and 57 healthy subjects. Neuromechanical, kinetic, and kinematic outcome parameters will be compared to relevant clinical test scores (e.g., the Nottingham Extended Activities of Daily Living for stroke patients, and the Movement Disorder Society version of the Unified Parkinson Disease Rating Scale for PD patients). Assessment of motor performance in various contexts (i.e., using different tasks and systematic manipulations of the mechanical and/or visual environment) is essential to understand the factors contributing to motor performance (i.e., the musculoskeletal, sensory, and cognitive systems) and to predict the potential for improvement of functional activities. This is essential for developing targeted therapy. REFERENCES [1] T. Geijtenbeek, F. Steenbrink, E. Otten, and O. Even-Zohar, “D-flow: immersive virtual reality and real-time feedback for rehabilitation”, In: Proc. of the 10th ACM international conference on virtual reality continuum and its applications in industry, pp. 201-208, (2011). ACKNOWLEDGMENT This work is part of the research programme IMDI Neurocontrol (Neuras project) financed by the The Netherlands Organisation for Health Research and Development.

A REAL-TIME THREE-DIMENSIONAL ULTRASOUND USER INTERFACE FOR TIPS; PRELIMINARY RESULTS
Cecile Cuijpers, Edit Varga, Camiel Klink, Jyotirmoy Banerjee, Pieter Jan Stappers
Abstract: The intrahepatic puncture is the most challenging and critical step in a transjugular intrahepatic portosystemic shunt (TIPS) placement [1][3]. The main problem while doing so is the limited three-dimensional, real-time information about the anatomy and instruments [1]. As a result, TIPS is risky and only experienced IRs are allowed to perform it[1]. Thus, additional support is needed from the image guidance system (IG system). Real-time three-dimensional ultrasound (3D US) is a promising modality to provide appropriate guidance [2]. The aim of this research study was to test if our new 3D US UI provides the IR with sufficient support to perform the intrahepatic puncture of a TIPS placement with effectiveness and satisfaction. A UI was created in MeVisLab showing the optimal needle line in: 1) a 3D image of the veins; 2) a 2D US image visualizing the longitudinal section of the veins; 3) an axial 2D US image visualizing the cross section of the veins. By using CT to 3D US registration and electromagnetically tracking, needle movements were visualized on each images of the UI. For two puncture lines, participants were asked to use the UI as IG system and to puncture, and to tell when they thought the puncture was successful. The number of punctures needed to achieve a successful puncture was counted, the coordinates of the two successful punctures were saved. Next, participants were asked a) to grade, on a scale from 1-10, if they thought that this new visualization method could improve the intrahepatic puncture and to provide additional feedback. 19 IRs and 9 trainees, from 14 nationalities participated in our test. All punctured in the target vein. In 48 of 56 trials one puncture was needed to successfully puncture the target vein. Only four times, two puncture attempts and once three puncture attempts were required. For the others, the amount was unknown. The mean grade for the overall UI was a 7.8 out of 10. All said that they would use this technique during an intrahepatic puncture if possible. Participants appreciated the availability of visual feedback about the puncture; instead of making an estimated guess the UI allowed them to control and adjust movements. The experiment was the first of its kind and showed that the new UI provides valuable visualization support to perform the intrahepatic puncture during TIPS placement. Physicians were very impressed and satisfied with the UI. Based on these findings we expect that the UI can decrease the number of puncture attempts, procedure time and risks in the future. In future research, we intend further develop the UI based on participants’ feedback and test in a more real-life set up. Overall, our findings strongly suggest that the 3D US UI will be a promising technique to support the IR sufficiently to perform the intrahepatic puncture of a TIPS procedure with increased effectiveness and satisfaction.

ANALYIS OF THE DENSIFICATION BEHAVIOUR OF TITANIUM/CARBAMIDE POWDER MIXTURES IN THE PREPARATION OF BIOMEDICAL TITANIUM SCAFFOLDS WITH THE SPACE HOLDER METHOD
Budi Arifvianto, Marius Leeflang, Jie Zhou
Abstract: Tissue engineering is a promising approach to the reconstruction of critical size bone defects. In this approach, a porous material, namely a scaffold, is devised as a template to support and guide the formation of new bone cells and the regeneration of bone tissue in the damaged site. Titanium is considered a preferred biomedical material for bone tissue engineering scaffolds. Among a number of techniques that have so far been developed to produce porous-structured titanium, the space holder method has been recognized as a viable one owing to its ability to produce porous scaffolds with desired structural characteristics. In this technique, space holding particles are utilized as a pore former. The fabrication process for titanium scaffolds is composed of a series of processing steps, i.e., (i) mixing of a titanium matrix powder with space holding particles, (ii) compaction of the powder mixture to form a composite compact, (iii) removal of space holding particles from the composite compact and (iv) sintering of the porous titanium matrix. Despite initial success in applying this technique, a number of technological challenges are still present, such as the difficulties in controlling the geometry changes of space holding particles during the compaction process. Obviously, compacting pressure must be optimized in order to prevent space holding particles from distortion, so as to ensure pore sizes and shape as desired for the scaffold product. In addition, the correlations between compaction process parameters and porous structure characteristics must be established to facilitate through-process modeling along the whole chain of the fabrication of bone tissue engineering scaffolds in the near future. In the present research, the behavior of titanium/carbamide powder mixtures during cold compaction was characterized and optimum compacting pressures for the fabrication of titanium scaffolds using the space holder method were derived. In addition, the Heckel equation describing the densification of powder mixtures during compaction was applied to assess its validity in the case of the present powder mixtures composed of two mechanically dissimilar components. A titanium powder with spherical particles and a carbamide powder with cubical particles were used as the matrix and pore former, respectively. Titanium/carbamide powder mixtures were prepared by mixing the powders for 3 h. Granular materials were then compacted with an instrumented powder compaction press. The variation of the load during compaction with the punch displacement was registered. The load-displacement plots were analyzed using the Heckel model for powder compaction and the rule of mixtures. The results showed varied compaction behavior of titanium and carbamide powders as their relative volume fractions changed. Titanium/carbamide powder mixtures exhibited intermediate behaviors of the component powders during compaction. The initial density of the compact was found to be of critical importance, as it determined the at-pressure density of the powder mixture compact. A lower compacting pressures was required for the compaction of a powder mixture with a larger volume fraction of carbamide. In addition, the experimental data could be well fitted into the Heckel equation. Although refining is still needed, the model can be used as a guide for the selection of an optimum compacting pressure in the preparation of titanium scaffolds with the space holder method.

SPREADING OF HARMFULL BONE DUST DUE TO SAWING
Arjo Loeve, Reza Gerretsen, Jenny Dankelman
Abstract: INTRODUCTION After the tsunami in the Indian Ocean in 2004 countless victims had to be identified by Dutch anthropologists. These anthropologists discovered that living HIV virus could be present in human remains as long as after 14 days. During autopsies, bone is often cut using electric oscillating saws. It is known that during sawing of bones, miniscule bone dust particles (aerosols) are spread several meters around the sawing location and stay suspended in air for several hours, which is a potential threat to workers in forensic and medical fields. Besides using breathing protection, the risks could be reduced by optimizing sawing techniques to minimize the spreading of bone aerosols. The Delft University of Technology and the Netherlands Forensic Institute cooperate to achieve safe bone separation techniques by investigating the effects of parameters such as saw tooth dimensions, sawing speed and sawing pressure on the spreading of bone aerosols. The presented work focussed on the sawing speed. METHODS An oscillating saw was used to create cuts of constant depth in pieces of bone. A custom made, air tight setup allowed to saw with a constant contact pressure, a preset sawing depth and a preset oscillation frequency of the saw. Dry human femur, containing no bone marrow, blood or other soft tissues, was used for its quick availability and in order to distinguish the effect of the bone properties in later experiments. A particle counter, placed at a height comparable to where the user’s head would be, was used to measure the amount of particles of 0.3, 0.5, 1, 2.5, 5, and 10 μm that reached the ‘breathing height’ above the sawing location. The setup was cleaned and checked for the absence of bone aerosols after each measurement. RESULTS Results suggest that the number of particles (ranging about 3.000 to 30.000.000 particles/L) at breathing height during bone sawing exponentially decreases with increasing aerosol particle sizes. Increasing sawing speeds seemed to exponentially increase the number of particles at breathing height for all measured particle sizes. CONCLUSION Slower sawing motions seem to reduce the amount of harmful aerosols spread by reciprocal saws when sawing human femurs. Further experiments will have to confirm whether the obtained results translate to other bones and to wet bones that still contain fat and soft tissues. Additionally, it will be investigated whether changing the movement, shape and size of the saw blade and saw teeth could reduce the spreading of harmful bone aerosols.

DESIGN OF A 2MM DIAMETER 8 DOF STEERABLE PROTOTYPE FOR THE TREATMENT OF CHRONIC TOTAL OCCLUSIONS
Aimee Sakes, Awaz Ali, Paul Henselmans, Paul Breedveld
Abstract: ABSTRACT As of today, Chronic Total Occlusions 1 (CTO) represent the most technically challenging lesions interventionists face during Percutaneous Coronary Interventions (PCI), with considerably lower success rates (50-88%) in comparison to semi-occluded and acutely occluded arteries [1]. The main technical challenge in PCI of CTOs lies in successfully puncturing and crossing the CTO with a guidewire. Initial puncture of the CTO is difficult to achieve due to less favourable biomechanical properties, including the presence of a fibrous cap and heavily calcified regions. This often leads to buckling of the guidewire, which lacks the columnar strength for sufficient force build up. Furthermore, since the guidewire cannot be actively steered, the path of least resistance is followed through the CTO, which in turn may cause dissection of the blood vessel wall or subintimal crossing, between the intima and adventitia. Subintimal crossing is (in most cases) not preferable since it can create a false lumen inside the blood vessel wall and make re-entry into the true lumen very difficult. Blood vessel wall dissection, however, poses a more immediate treat, since it can lead to excessive bleeding and subsequent emergency surgery. A steerable crossing device could be the solution to prevent crossing difficulties in the future, as it will give the interventionist the freedom to navigate through the vascular system and CTO freely. Therefore, a steerable prototype is designed with an 8 degrees of freedom (DOF) cable actuated tip (Ø 2 mm, L = 32 mm) divided over 4 steering segments; allowing for complex S-curves. The tip contains a lumen (Ø 1 mm) to allow for the insertion of, amongst others, a balloon catheter, a guidewire, or an IntraVascular UltraSound probe (for visualization purposes). The steerable tip is connected to a rigid shaft (Ø 2 mm, L = 200 mm), which in turn is connected to the handle. The handle consists of an innovative combined locking and steering mechanism to lock the tip position in place and to precisely steer each segment separately. This construction allows for both the tip position and direction to be changed independently, allowing for a scanning movement to determine, for example, the best (central) entry point of the CTO (with for example (IVUS)). Furthermore, the steerable tip can be used to increase the columnar strength by abutting of the blood vessel wall for support, and as such it can help to prevent buckling during initial puncture. Currently, tests are being performed to determine the feasibility of the design for PCI in the future. Even though it is still a long way towards a fully applicable clinical tool, the tests will give first insights into the possibilities and advantages of having such a tool in PCI. Therefore, this prototype represents a first step in the direction towards increasing the success rate of this intervention. REFERENCES [1] G. W. Stone, et al., “Percutaneous Recanalization of Chronically Occluded Coronary Arteries, A Consensus Document Part II”, Circulation, Vol. 112, pp. 2530-2537, (2005). 1 defined as a heavily calcified total occlusion of one of the coronary arteries of at least three months old.

LOW-POWER, LOW-COST 1D HAPTIC FEEDBACK ON EMBEDDED PLATFORMS
Tom Lankhorst, Victor Sluiter, Arno Stienen
Abstract: Haptic feedback is used to assist human control tasks such as surgery[1] or driving, but also for rehabilitation of stroke patients[2]. The goal of this research is to develop an alternative for existing expensive conventional haptic feedbacksystems. A cheap hybrid stepper motor (HSM) is used instead of a normal brushed DC-motor. HSMs are especially designed for open-loop position control, and are capable of providing a relatively high torque at low speeds which makes them interesting for haptic systems. The significant cogging torque of the HSM was compensated by adding a rotational position sensor to the system, and implementing a closed-loop current control. For haptic rendering we used impedance control. In the past, attempts have been made to implement haptic feedback on low-cost stepper motors using series elastic actuation[3]. Direct control of torque by the stepper motor is preferred over series elastic actuation because it simplifies the haptic system. Our device uses Field Oriented Control to control the torque-producing current in the HSM. Using both position and force based compensation method, the cogging of the motor can be reduced by 89% to 0.18mNm RMS, which equals comparable research on PMSM motors[4]. Using both friction and cogging compensation, the impedance control was evaluated using a stiffnessdamping plot, and it was shown with a that a wide range of haptic systems could be rendered. We have demonstrated that a low-power, low-cost 1D haptic device with good haptic rendering can be made of standard low-price mass-produced components. It has been shown that conventional HSMs can be open-loop torque controlled using FOC combined with cogging and friction compensation. All compensationand calibration methods can be implemented on the system itself, eliminating the demand for an external computer or measurement set-up. REFERENCES [1] Kitagawa et. Al, Effect of sensory substitution on suture-manipulation forces for robotic surgical systems, Jounal of Thorac Cardiovascular Surgery, pp151-8 (2005) [2] Broeren et al, Virtual reality and haptics as a training device for movement rehabilitation after stroke: A single-case study”. Archives of Physical Medicine and Rehabilitation, 85(8), pp. 1247–1250, (2004) [3] Lawrence et al, Low cost actuator and sensor for hig-fidelity haptic interfaces, HAPTICS ’04 Proceedings. 12th International Symposium on Haptics, pp74-81 (2004) [4] Z.Jabbour et al, Identification and Compensation of Torque Ripples of a PMSM in a Haptic Context, IECON2010 – 36th Annual Conference on IEEE Industrial Electronics Society, DOI 0.1109/IECON.2010.5675431 (2010)

DEVELOPING A DYNAMIC BRACE IN THE TREATMENT OF CLUBFOOT
Bob Giesberts, Edsko Hekman, Patrick Maathuis, Sjoerd Bulstra, Bart Verkerke
Abstract: Clubfoot (talipes equinovarus) is a common congenital deformity and often treated with the Ponseti method [1]. To correct the adductus deformity, the orthopaedist applies manual pressure on the medial side of the first metatarsal with counter pressure on the lateral side of the talar neck, abducting the foot while aligning the talus with the calcaneus. This manipulation stretches the tissues on the medial side of the foot and is maintained for a week with a plaster cast. Most cases of clubfoot are corrected after five to six cast changes and, in many cases, a percutaneous Achilles tenotomy. After the casting period a foot abduction brace is used for four years to prevent relapse. Magnetic Resonance Imaging comparing tarsal bones before and about ten minutes after casting shows immediate shape changes to the cartilage anlagen [2, 3]. Our preliminary measurements show that the initial cast/foot interface pressure of 2 – 10N at the first metatarsal and 10 – 15N at the talar neck both drop to zero about an hour after casting [4]. These findings suggest that most of the correction is done long before the end of the week and treatment could be accelerated drastically. Despite the Ponseti method being the ‘gold standard’ [5], the toe-to-groin plaster cast hinders normal leg movement and makes bathing and maintaining child hygiene difficult [6]. Current study aims to develop a dynamic brace as a more efficient and more user-friendly alternative. To do so, extensive pressure measurements are scheduled to first objectify the Ponseti method and to find possibilities for improvement. After these measurements multiple concepts for dynamic braces will be developed and prototypes will be tested in a clinical setting.

PILOT STUDY: 5DT DATAGLOVE ASSESSMENT AND THE MEASUREMENT OF MAXIMUM ACCELERATIONS DURING FINGER FLEXIONS AND EXTENTIO
Ewout Arkenbout, Joost de Winter, Paul Breedveld
Abstract: The design of steerable instruments for Minimally Invasive Surgery is challenging, because one has to take into account (1) anatomical restrictions, (2) instrument handling capabilities, and (3) mental workload for the surgeon. Instruments having multiple Degrees of Freedom (DOF) incorporated into the shaft or tip are especially difficult to control [1]. To research control coupling strategies between hand movements at the instrument handle or master interface and these DOF, a test setup is being developed incorporating 3Gear Systems software [2] and two Kinect cameras. Using this test setup, we previously measured the position and orientation of the hand, as well as the deflections of the individual fingers. Results showed the variance of these measurements to be dependent on hand orientation [3]. We hypothesize that the addition of Dataglove sensors, and subsequent fusion of these sensor data with the 3Gear software output, will improve tracking performance and robustness. To this end, the implementation of a Kalman filter was chosen, for which some a priori knowledge is required on the reliability of the chosen sensors as well as the maximum finger flexion accelerations of the hand. The 5DT Dataglove [4], measuring fingers flexure (1 sensor per finger with 75Hz sampling frequency), was assessed by comparing its sensor readings to actual Metacarpophalangeal (MCP) and Proximal Interphalangeal (PIP) joint angles. The MCP and PIP angles were calculated through the tracking of colored markers attached to the glove and post-processing of the video recordings. Due to marker occlusion, only the thumb, index, and pink fingers were assessed. Data were gathered for 25 finger flexions per finger. Both the Dataglove sensor readings as well as the sum of the MCP and PIP angles were rescaled to the range [0, 1] where 1 equals full finger flexion and 0 full extension. Moreover, maximum finger accelerations were calculated from Dataglove measurements during flexion and extension of the fingers at maximal speed. These data were resampled to 1000 Hz, and filtered with a 2nd- order Butterworth filter with a cut-off frequency of 10 Hz. All measurements were conducted on the hand of the first author. For the thumb, index and pink the mean (SD) of the absolute error over the full measurement range between Dataglove measurements and the sum of MCP and PIP angles were 0.066 (0.066), 0.104 (0.092), and 0.125 (0.095) respectively. The corresponding MCP:PIP ratio, indicating how much the MCP and PIP joints contributed to the Dataglove sensor reading, for the respective fingers were 37:63, 52:48 and 42:58. The mean (SD) of the maximum accelerations for the fingers from thumb to pink were 2.1 (0.5), 5.1 (0.6), 6.0 (1.0), 5.0 (1.2) and 5.5 (1.7) ·104 deg/s2, occurring about 30 to 50 ms after movement initiation. Maximum extension accelerations were approximately 40% lower than their counterpart flexions, except for the thumb where they were almost equal. Uncertainty in the performed measurements include possible shifting and misalignment of both the Dataglove sensors and/or the markers due to elasticity of the glove. However the data clearly showed there being a difference between fingers when attempting to estimate MCP, PIP and full finger flexion and extension.

SYMBIONIC HAND ORTHOSIS FOR PEOPLE WITH DUCHENNE MUSCULAR DYSTROPHY
Kostas Nizamis, Ronald Bos, Claudia Haarman, Bart Koopman, Imelda de Groot, Micha Paalman, Jaap Haarlar, Peter Veltink, Arno Stienen, Just Herder, Dick Plettenburg
Abstract: Unrestricted functioning of the human arm and hand is essential for autonomy and personal quality of life, but functions can be lost completely as a result of a disease or trauma. Day by day, it becomes more apparent that wearable assistive devices represent the future in terms of assisting people in improving or regaining basic functional capabilities. Duchenne Muscular Dystrophy (DMD), is a recessive X-linked form of muscular dystrophy, affecting around 1 in 4000 boys, causing muscle degeneration and eventually death. Rapid pharmaceutical advances have increased the life expectancy of people with neuromuscular diseases. The need of an active and adaptive support for people with DMD is now imminent in order to improve their quality of life and increase their independence. In our view, devices should continuously adapt to the user according to a therapy plan or to compensate for user changes, changing environment, or changing tasks. The objective of the Symbionics program is to create systems that co-adapt automatically, either intrinsically by design, by control, or a combination of both. Moreover, it is of great importance that any wearable orthosis should be inconspicuous, in order enhance social acceptance. There is a strong connection to the Flextension A-gear (STW-OTP) project, which aims at the development of a wearable and adaptive arm orthosis for boys with DMD. We intend to integrate our intuitive hand orthosis to the Flextension A-gear arm support. The design and development of the aforementioned hand orthosis is separated in four stages:  Investigate intuitive control strategies for a high-tech hand orthoses with many degrees of freedom.  Develop mechanisms that allow for transparent (predictable) man-machine interaction, a comfortable fit and inconspicuous design.  Develop functional hand orthoses that can be used throughout the stages of the disease.  Integrate the hand orthoses into the Flextension arm device. Our work will be conducted in close collaboration with partners including Hankamp Rehab, TMSI, Festo, Pontes Medical and Spieren voor spieren. This research is supported by the Dutch Technology Foundation STW (grants 13524, 13525).

HIGHER HARMONIC GENERATION MICROENDOSCOPY OF HUMAN BRAIN TUMORS
Nikolay Kuzmin, Giovanni Galgano, Philip de Witt Hamer, Pieter Wesseling, Johannes Baaijen, Huibert Mansvelder, Marie Louise Groot
Abstract: Surgical resection of diffuse infiltrative gliomas to the maximal feasible extent is increasingly used in treatment for these tumors. The major challenge of this type of neurosurgery is the detection of tumor margins, assisted nowadays by different pre-operative imaging techniques and intraoperative histopathological analysis. The development of a noninvasive in-situ optical biopsy probe of glioma margins is highly desirable for increase of surgical precision. Higher harmonic generation microscopy (HHGM) is a novel imaging technique in brain research [1,2]. It provides label-free images of living tissues with sub-cellular resolution and intrinsic depth sectioning. HHGM contrast relies on tissue discontinuities, noncentrosymmetric molecules and autofluorescent proteins generating higher optical harmonics and multiphoton autofluorescence. HHGM of healthy human brain tissues reveals lipid-rich axonal networks, neuronal and glial cell somata with lipofuscin deposits and nuclei/nucleoli outlines, erythrocytes and blood vessels [1,2]. With a stationary microscope we show that HHGM has a great potential as an optical biopsy probe of glioma margins. HHGM images of fresh unfixed brain samples reveal areas with increased cellularity, malignant blood vessel architecture, multi-nucleated giant cells and abnormally clustered cells, bilobed and multinucleated cells – the key "histological" hallmarks of brain tumors. We aim to translate the laboratory HHGM into a clinical microendoscope – a rapid intraoperative tool for in-situ assessment of glioma margins. We will present several HHGM-microendoscope designs including selection of a powerful and compact near-infrared femtosecond laser source, optical fibers for excitation/emission light delivery, distal/proximal laser beam scanners, high-NA micro-objectives as well as preliminary images of test samples and brain tumor tissues. REFERENCES [1] S. Witte, A. Negrean, J.C. Lodder, C.P.J. de Kock, G.T. Silva, H.D. Mansvelder, M.L. Groot, Label-free live brain imaging and targeted patching with third-harmonic generation microscopy. Proceedings of the National Academy of Sciences of the United States of America, 2011. 108(15): p.5970–5975. [2] N.V. Kuzmin, M.J.T. ter Veer, D.A. Kozareva, J.C. Baayen, H.D. Mansvelder, M.L. Groot, Apoptotic/necrotic dynamical processes in deep-tissue mouse and human brain resolved with third harmonic generation microscopy. To be submitted for publication.

A REVIEW OF NEEDLE RELATED COMPLICATIONS
Elise Buiter, Gabrielle Tuijthof, Jenny Dankelman
Abstract: Needle interventions are routinely performed. An estimated 20 billion injections alone are administered yearly [1]. Needle interventions are associated with reduced surgery time and reduced trauma compared to open surgery. However decreased visibility of the instrument tip leads to complications not previously encountered. A literature search in Scopus was performed to highlight complications resulting from needle interventions and based upon the results guidelines were created for needle design to prevent the found complications. The Clavien-Dindo classification was applied to define the complication severity; grade 0 for no patient discomfort to grade V for patient death [2]. Complications in the integumentary system are common (up to 90% prevalence), but have low severity (grade 0-II). Complications in the musculoskeletal system have low severity (grade 0-II) and are relatively uncommon (up to 1.8% prevalence). Complications occurring in deeper laying tissues, the nervous system, veins and organs, are more severe (grade 0-V) and have a prevalence of up to 7%. The design guidelines follow from the influence of needle geometry on the complication rate. A smaller outer diameter of the needle leads to less complications [3], [4]. For example, a 12 fold decrease in eye surgery for a 25- vs 20-gauge needle [3]. In clinical practice, a minimal needle diameter is required to effectively deliver or extract substances. For instance, a 27-gauge needle is the smallest needle recommended for delivering local anaesthetics [5]. Sharp needle tips cause less trauma during insertion, but have an increased chance of damaging surrounding tissues [5]. The tip shape and the location of the needle orifice must be optimised for specific interventions. For local anaesthetic the orifice must be as close to the tip as possible to prevent Dural puncture: Whitacre needles result in less post dural puncture headache than Quincke needles, 0.4% and 2.7% respectively [6], [7]. For ethanol injections, substance distribution is important [8] and an improved design might have multiple orifices. Invisibility of the needle tip on imaging systems can cause misplacement of the needle tip which increases the risk of damaging surrounding healthy tissues, nerves and organs [9]. To conclude, the complication rate is influenced by several aspects of needle geometry that can be taken into account during needle design. More research is necessary for improvement of needle tip visibility on imaging systems, such as different needle material or tip coatings. This will decrease the complications related to needle misplacement and accidental trauma to surrounding tissues. Application of a standard grading system for complication severity will improve the quality of future research into decreasing instrument related complications. REFERENCES [1] World Health Organisation, “Background information on injection safety,” 2014. [Online]. Available: who.int/injection_safety. [2] P. a Clavien, J. Barkun, M. L. de Oliveira, J. N. Vauthey, D. Dindo, R. D. Schulick, E. de Santibañes, J. Pekolj, K. Slankamenac, C. Bassi, R. Graf, R. Vonlanthen, R. Padbury, J. L. Cameron, and M. Makuuchi, “The Clavien-Dindo classification of surgical complications: five-year experience.,” Ann. Surg., vol. 250, no. 2, pp. 187–96, Aug. 2009. [3] D. Y. Kunimoto and R. S. Kaiser, “Incidence of endophthalmitis after 20- and 25-gauge vitrectomy.,” Ophthalmology, vol. 114, no. 12, pp. 2133–7, Dec. 2007. [4] P. Geraghty, S. Kee, G. McFarlane, M. k Razavi, D. Y. Sze, and M. D. Dake, “CT-guided Transthoracic Needle Aspiration Biopsy of Pulmonary Nodules: Needle Size and Pneumothorax Rate 1,” Radiology, pp. 475–481, 2003. [5] D. K. Turnbull, “Post-dural puncture headache: pathogenesis, prevention and treatment,” Br. J. Anaesth., vol. 91, no. 5, pp. 718–729, Nov. 2003. [6] F. Bano, S. Haider, S. Aftab, and S. Sultan, “Comparison of 25-gauge, Quincke and Whitacre needles for postdural puncture headache in obstetric patients.,” J. Coll. …, vol. 14, no. 11, pp. 647–650, 2004. [7] U. Santanen, P. Rautoma, H. Luurila, O. Erkola, and P. Pere, “Comparison of 27‐gauge (0.41‐mm) Whitacre and Quincke spinal needles with respect to post‐dural puncture headache and non‐dural puncture headache,” Acta Anaesthesiol. Scand., vol. 48, pp. 474–479, 2004. [8] S. Shiina, T. Teratani, S. Obi, K. Hamamura, Y. Koike, and M. Omata, “Percutaneous ethanol injection therapy for liver tumors.,” Eur. J. Ultrasound, vol. 13, no. 2, pp. 95–106, Jun. 2001. [9] A. Perlas, G. Lobo, N. Lo, R. Brull, V. W. S. Chan, and R. Karkhanis, “Ultrasound-guided supraclavicular block: outcome of 510 consecutive cases.,” Reg. Anesth. Pain Med., vol. 34, no. 2, pp. 171–6, 2009.

NEUROMODULATION OF THE SPINAL CIRCUITS BY TSDCS FOR THE REHABILITATION OF SPINAL CORD INJURY
Alexander Kuck, Edwin van Asseldonk, Dick Stegeman, Herman van der Kooij
Abstract: Rationale: Spinal Cord Injury (SCI) is a severe injury to the pathways of the central nervous system (CNS). Despite a heavy post injury physical rehabilitation regime, SCI patients are often bound to a wheelchair or left with other impairments diminishing their quality of life. Trans-spinal direct current stimulation (tsDCS) is a promising new technique for the treatment of SCI. During tsDCS a small direct current is applied to the spinal cord via two or more stimulation electrodes, placed on the back of the subject. The technique thereby aims to alter the response of the neural pathways in the spinal cord, which is hypothesized to have a positive effect on the recovery of the damaged spinal cord neurons. Objective: The objective of this study, is to assess how tsDCS modulates the excitability of the spinal cord and whether this modulation is dependent on the electrode placement configurations. The primary goal is to compare a new electrode placement configuration with one that is commonly used in previous tsDCS studies. This will be assessed using the Hoffman Reflex, whereby the novel configuration is hypothesized to have a significantly larger modulatory effect on the H reflex recruitment curve. Methods: TsDCS is applied on the lumbar spinal cord for a period of 15 minutes with a current of 2,5mA. The two different configurations are: 1) cathode and anode placed on the T11 vertebra and the left shoulder blade respectively (commonly used) and 2) the two electrodes placed over the spinal cord, 8 centimeters apart and centered around the 11 thoracic vertebra. The ascending part of the H-reflex recruitment curve is measured before, during and after tsDCS. This is assessed by stimulating the tibial nerve with a biphasic pulse, at increasing amplitudes and measuring the response in the m. Soleus by using surface EMG. Results: Preliminary results show a significant difference between the two stimulation configurations. The proposed novel placement induces a visible left shift of the H reflex recruitment curve, whereas no significant effect can be observed when using the previously used electrode configuration. Conclusion: The results are promising to be able to contribute to the effectivity tsDCS, which could possibly be applied in the rehabilitation of spinal cord injured subjects in the future.

ROBOTIC SIMULATION OF ADVANCED HAND ORTHOSES
Niek Beckers, Ronald Bos, Teun Stortelder, Hans Rietman, Herman van der Kooij, Dick Plettenburg, Arno Stienen
Abstract: Everyday we use our hands to manipulate objects in our environment, ranging from fine motor tasks such as writing to gross motor tasks such as carrying a full shopping bag. Even the simplest of tasks, such as using cutlery, requires complex coordination and precision of our multi-fingered extremities. Thus, when disease or injury impairs hand function, the ability to execute activities of daily life is diminished considerably. More than ever people suffering from such impairments rely for their independence on wearable assistive devices that recover functional capabilities. We believe that these devices should continuously adapt to the user according to a therapy plan or to compensate for user changes, changing environment, or changing tasks. Therefore, the objective of the Symbionics program (STW, NL) is to create systems that are able to adapt to the user or therapeutic demands through design, control, or both. As one of two projects within the Symbionics program, we will focus on robot-aided rehabilitation of stroke, which includes developing novel adaptive hand orthoses and exoskeletons. More specifically, in a collaborative effort between the University of Twente and Delft University of Technology, we aim to develop A) a framework that connects clinical impairments to conceptual solutions for dynamic hand orthoses, B) a range of state-of-the-art implementations of device components that span the range of these solutions, C) a high- performance exoskeleton that can simulate current and proposed orthoses, and D) a realistic implementation of a new hand orthosis targeted at post-stroke impairments. A key element is to balance the crossover from function recovery (rehabilitation) to compensation (permanent assistance) based on the user and therapeutic needs. Furthermore, we are determined to make our technology available to individuals with impaired motor function by closely collaborating with several clinical partners such as the Roessingh Research and Development centre, Medisch Spectrum Twente, and Radboud University Nijmegen, as well as commercial partners including Hankamp, Hocoma, MOOG, and Festo. This work is supported by the Dutch Technology Foundation STW (grants 13524, 13525).

ROBOT-ASSISTED LEARNING OF COMPLEX UPPER EXTREMITY MOVEMENTS
Niek Beckers, Arno Stienen
Abstract: Being able to play the piano the way Rachmaninoff did epitomizes the amazing capabilities of our motor control system – it requires perfect timing, precision, and complex coordination of the arms, hands, and fingers. However, when motor function is impaired, for instance due to stroke, one’s ability to perform complex functional movements such as tying shoelaces or eating with knife and fork diminishes considerably. Individuals after stroke often struggle with relearning these everyday tasks or acquiring new skills. Robot-assisted rehabilitation has been shown to facilitate stroke recovery, however consistent improvements in functional abilities are marginal at best. We therefore aim to advance rehabilitation of functional movements of the arm, hand, and fingers. The first step is to investigate motor learning of complex functional movements in healthy people. To study the human ability to learn existing and new functional arm movements, we take a multidisciplinary approach including state-of-the-art robotics and novel motor human learning paradigms. Our first goal is to develop and validate a novel exoskeleton robot that allows for unimpeded assessment and natural modulation of multi-joint movements of the arm, hand, and fingers. Second, using this robot, we intend to explore and study (re)learning of functional movements. We aim to develop novel motor learning paradigms that bridge the extensive knowledge on human motor learning in abstract movement tasks with learning of functional arm movements. We propose playing the piano as a translational-research model for learning complex functional movements for healthy people. First, piano playing is an excellent case study for validating the robot on unimpeded assessment and natural modulation of upper extremity movements. Second, it is not unreasonable to say that the dexterity required for many everyday activities is also necessary for playing the piano and that mastering the piano may be as difficult as relearning everyday movements is for patients. Note that we do not posit that learning to play the piano in healthy people is the same as learning activities of daily life in stroke patients, nor do we aim to teach patients how to play the piano. We do postulate that both state-of-the-art robotics and novel learning paradigms for complex movements for healthy people would greatly benefit motor recovery paradigms. This work is supported by the Dutch Technology Foundation STW (grants 12479, 13524, 13525).

THE RELATION BETWEEN THE NUMBER OF A-MODE ULTRASOUND SAMPLED POINTS AND THE REGISTRATION ACCURACY AND ROBUSTNESS ON COMPUTER AIDED ORTHOPEDIC SURGERY
Kenan Niu, Jasper Homminga, Victor Sluiter, André Sprengers, Nico Verdonschot
Abstract: BACKGROUND: In Computer Aided Orthopaedic Surgery, intraoperative registration is a crucial step. Here the coordinate system of the preoperatively scanned image and the coordinate system of the actual patient are linked, so the surgeon can follow the pre-planned surgical procedure. This registration is typically done using a mechanical probe to palpate and acquire the set of points by exposing the surface of bones. But recently, A-mode ultrasound had been proposed and validated in experimental and clinical environment, showing advantages, such as the non-invasiveness and lower time requirements[1]. However, currently it is still unknown how many points should be acquired by A-mode ultrasound probe and whether there are more or less efficient regions for these points on the femur surface[2]. METHODS: In this study, we used simulations for exploring the relation between the number of sample points and the accuracy of the registration. After the sample points, (between 2 and 20), were randomly picked and moved to a new location, they were registered to the original positions on the bone surface. This registration used a pre-registration (based on automatically segmented anatomical landmarks on the bone surface), followed by an Iterative Closest Points (ICP) algorithm. The quality of the registration was measured by the Root Mean Square Error (RMSE) of the distances between original and registered points. The robustness was analysed by repeating this procedure 100 times for each set of points. RESULTS: The addition of a pre-registration to the ICP algorithm improved the registration quite remarkably. Without the pre-registration a good registration required more than 20 sample points, with the addition of the pre-registration a good registration required only 11 sample points. Overall the average RMSE was 1.2 to 3.1 mm. The study of the distribution of the sample points over the surface of the bone showed no clear correlation between the locations of the sample points and the quality of the registration (the RMSE). CONCLUSION: There is a nonlinear but inverse relation between the number of sample point and target registration error, and 11 sample points seems to be enough for a good registration of pre-made images to an actual patient. The fact that there was no clear correlation between the location of the sample points and the quality of the registration was unexpected. We feel that this may have been caused by the uniqueness of the surface of a real bone: every point on the surface is unique and therefore useful for the registration. REFERENCES: [1] A. Mozes, T.C. Chang, L. Arata et al. “Three-dimensional A-mode ultrasound calibration and registration for robotic orthopaedic knee surgery”, Int J Med robotics Compit Assist Surg 2010; 6: 91-101 [2] C. Amstutz, M. Caversaccio et al, “A-Mode Ultrasound-Based Registration in Computer-Aided Surgery of the Skull, Arch Otolaryngol Head Neck Surg. 2003 Dec;129(12):1310-6.

STATE-OF-THE-ART OVERVIEW OF HAND ASSISTIVE DEVICES
Ronald Bos, Niek Beckers, Kostas Nizamis, Arno Stienen, Just Herder, Dick Plettenburg
Abstract: ABSTRACT Human hands are complex and versatile organs that provide for an essential role in manipulating objects and overall interaction with the outside world. Many stroke survivors that suffer from motor impairments or people with Duchenne Muscular Dystrophy (DMD) are limited in hand dexterity, preventing them from fully utilizing these capabilities. This might have a negative effect on the person’s independency and social interaction. Assistive devices for the hand aim to recover or enhance hand functionality, ultimately improving the quality of life for individuals suffering from motor function impairments. A multitude of assistive devices have already emerged that together focus on a range of applications, each having their own points of particular interest [1-3]. Although they may differ in their final purpose, they have a common challenge in trying to assist and/or measure the hand’s functional movements with minimal interference to the already present hand capacities. A state-of-the-art overview of existing hand assistive devices is acquired and can be classified according to their intended functional complexity: as a research tool, a clinical tool, a home rehabilitation tool or daily assistive tool. More importantly, the identification of fundamental solution strategies in their mechatronic components produces a framework that is able to review design characteristics and understand design choices. This also allows for the inclusion of supplementary technologies from prosthetics, robotics and additional novel mechanical structures. These results, combined with further endeavors from the Symbionics program that couple clinical impairments to device implementations, provide as an essential tool for the development of novel solutions. The Symbionics projects on the hand are supported by STW (NL, #12479, #13524 and #13525), Hankamp Rehabilitation (Enschede, NL), Hocoma (Volketswil, CH), TMSi (Oldenzaal, NL), Moog (Nieuw Vennep, NL), FESTO (Delft, NL), and multiple Duchenne foundations (NL & USA). REFERENCES [1] S. Balasubramanian, J. Klein and E. Burdet, “Robot-assisted rehabilitation of hand function”, Curr Opin Neurol, Vol. 23, pp. 661670, (2010). [2] P. Heo, G.M. Gu, S. Lee, K. Rhee and J. Kim, “Current hand exoskeleton technologies for rehabilitation and assistive engineering”, Int J Precis Eng Man, Vol. 13, No. 5, pp. 807824, (2012). [3] P. Maciejasz, J. Eschweiler, K. Gerlach-Hahn, A. Jansen-Troy and S. Leonhardt, “A survey on robotic devices for upper limb rehabilitation”, J Neuroeng Rehabil, Vol. 11, No. 1, art. no. 3, (2014).

INCREASING MOTIVATION IN EHEALTH THROUGH GAMIFICATION
Frederiek de Vette, Monique Tabak, Miriam Vollenbroek-Hutten
Abstract: ICT solutions can help to alleviate the increasing demand for elderly care by e.g. enabling medical professionals to remotely provide care, and activating a healthier lifestyle which extends autonomy and independence of elderly [1]. However, these solutions are often not sufficiently effective as adherence is low and decreases over time [e.g. 2]. Engagement in technology is important as studies have shown that use of an application significantly relates to improvement in health outcomes [5]. Gamification – the use of game design elements in non-game contexts [3] – offers great potential regarding the engagement and motivation of the elderly [4]. Recently, the number of gamification initiatives in healthcare is rapidly growing. However, the majority are not based on funded theory and focus on short-term engagement through extrinsic rewards. A common pitfall for gamification is to lose sight of the core experience of games and merely add visual components of games, such as points and rewards, to another application, resulting in discarding of the application after a short amount of time. We would therefore like to adopt the definition: gamification is the application of knowledge from game design to non-game fields. Especially in eHealth it is important to look further than the rather simplistic approaches of current gamification practices as long-term motivation and adherence is needed to achieve lifestyle behaviour change. Gamification, specifically for elderly, is an unexplored area and it is unknown what motivational concepts can be used to achieve this. This research works towards the realisation of a framework for creating motivating and engaging eHealth applications using gamification, for the elderly population. We present an overview providing insights in current motivational approaches, addressing underlying theory from psychology, behavioural science and game design. In order to create sustained engagement, we should aim to address the intrinsic motivation of users and satisfy their psychological needs [6]. However, psychological and motivational aspects cannot be generically applied to individuals, and classifications of the user and its preferences are needed. Preliminary results indicate a link between personality and the preference for certain game elements.

THE STABLE PLAQUE PARADIGM: RELIABLY IDENTIFYING THICK-CAP, LOW-STRESS, STABLE CAROTID FIBROATHEROMAS
Harm Nieuwstadt, Anton van der Steen, Jolanda Wentzel, Aad van der Lugt, Frank Gijsen
Abstract: Introduction: Carotid atherosclerotic plaque rupture can result in ischemic stroke, a major cause of death and disability worldwide. The number of costly and risky surgical interventions on stenosis-inducing, yet potentially stable, carotid plaques remains high, which calls for improved vulnerability assessment. The peak cap stress is a promising biomechanical marker for rupture risk, and current focus lies on identifying high-stress plaques. However, our recent study showed that MRI-based finite element analysis (FEA) stress computations are highly unreliable for thin-cap, high-stress carotid plaques [2]. The MRI voxel size (~0.6 mm) is too large with respect to fibrous cap thickness in most high-stress plaques, while cap thickness is biomechanically crucial. Thick-cap, low-stress plaques yielded a reliable segmentation and stress computation. Can we better identify low-stress, histologically-stable carotid lesions to minimize unnecessary surgeries? Methods: We obtained 32 carotid histological plaque cross-sections from 12 patients with >70% stenosis and classified the lesions according to the histological classification scheme introduced by Virmani [1]. We indentified 8 plaques as pathologic intimal thickening (Type 2), 14 as fibrous cap atheromas (Type 3), and 10 as thin fibrous cap atheromas (TCFA) (Type 4) with a cap thickness <200 microns. Type 2 and 3 plaques are generally considered stable while TCFAs are acknowledged rupture-prone, vulnerable plaques. The peak cap stress in the cross-sections was computed with Abaqus 6.11. To mimic in vivo carotid MRI, we subjected all plaques to numerical MRI simulations (JEMRIS). This provided a fully-controlled setting enabling a submillimeter-scale comparison with the underlying morphology/stress [2]. MRI segmentations were subjected to FEA for the MRI-based plaque model stress. Results: One example and all results are shown in the Figure. We found that the peak cap stress in Type 2 plaques was on average the lowest, followed by Type 3 plaques. Type 4 plaques (TCFAs) had the highest average stress values. Importantly, not all TCFAs (Type 4) had a peak cap stress higher than a fibrous cap atheroma (Type 3). This disagreement between histological classification and FEA outcome suggests the likely added value of biomechanical stress analysis over merely histology-based classification. As shown in the table, only the selection of the lowest stress plaques (25% quartile) guaranteed agreement with a histologically-classified stable plaque. MRI-based plaque FEA resulted in an inaccurate, underestimated peak cap stress, yet it still yielded consistent agreement with histological classification for only the lowest-stress (25% quartile), stable plaques. Conclusion: With current MRI technology, a large fraction of high-stress, vulnerable carotid plaques cannot be reliably identified due to stress underestimation and disagreement with histological classification. Low-stress plaques based on MRI FEA were always histologically classified as stable. Therefore, identifying solely the lowest-stress carotid plaques could be a more effective approach to reduce unnecessary surgeries. [1] Virmani et al., Neurosurgery 2006. [2] Nieuwstadt et al., Journal of Biomechanical Engineering 2014. This research was supported by CTMM and the Netherlands Heart Foundation (PARISk).

ESTIMATION OF ATHEROSCLEROTIC PLAQUE MATERIAL PROPERTIES FROM ULTRASOUND BY INVERSE FINITE ELEMENT ANALYSIS
Ali Akyildiz, Rik Hansen, Lambert Speelman, Harm Nieuwstadt, Anton van der Steen, Jolanda Wentzel, Chris de Korte, Frank Gijsen
Abstract: Atherosclerotic plaque rupture is the main cause of myocardial infarction and stroke. Biomechanical modelling has great potential for future plaque rupture risk assessment. However, data on material properties of plaque components are scarce and present an enormous variation due to large methodological differences and variation in loading conditions and plaque integrity. We therefore propose a method that combines ultrasound imaging with inverse finite element (FE) modelling of intact specimen under physiological conditions to obtain patient specific plaque material properties. Ex-vivo ultrasound measurements were performed on 6 excised atherosclerotic porcine iliac arteries. Intravascular pressure steps (10 to 120 mmHg) were applied and radiofrequency (RF) ultrasound data were recorded at different pressures. Displacement fields between the pressure steps were determined using advanced cross-correlation techniques on the 2D RF data. Thereafter, histological analysis was done on the arteries, from which FE models were created and displacement fields were computed. By varying the Young’s Moduli of the intima and outer wall in the FE models and mimizing the difference between the computed and ultrasound displacement fields, the optimum Young’s Moduli of the intima and wall were determined. As plaque material is generally non-linear, a piecewise linear optimization was done for pressure intervals of 10-80 mmHg, 80-100 mmHg, and 100-120 mmHg. To evaluate the effect of plaque-specific properties, cap stress was calculated at 120 mmHg with FE analysis using average and plaque-specific Young’s Moduli. Based on a standardized displacement bullseye plot with 2 rings and 16 segments (see figure), correlations between the ultrasound and computed axial displacements showed to be strong for the 6 specimen (R2 = 0.83-0.91). Although considerable variation between the plaques was found, the intima and wall showed increasing stiffness for higher pressure intervals (10-80; 80-100; 100-120 mmHg, mean (range)): intima 14(0.6-36); 106(3.6-186); 227(1.8-534) kPa, wall 212(114-330); 1060(318-2200); 1150(320-2360) kPa (see figure). When cap stresses were calculated with FE analysis using the average and the plaque specific Young’s Modulus, differences in peak von Mises cap stress were found between -37% and +176%. Determining material properties on a plaque-specific basis is crucial for future biomechanical plaque analysis. The proposed innovative approach combines plaque imaging with biomechanical modeling to estimate material properties of intact plaques under physiological loading conditions and provides possibilities for measuring these plaque-specific properties in vivo. Using these plaque-specific properties in biomechanical plaque analysis improves the prediction of cap stresses, and therefore plaque risk assessment.

AUTOMATIC DETECTION OF TREMOR AND NON-TREMOR WINDOWS IN MOVEMENT DISORDERS
Frauke Luft, Sarvi Sharifi, Winfred Mugge, Alfred Schouten, Lo Bour, Fleur van Rootselaar, Peter Veltink, Ciska Heida
Abstract: Tremor is the most common symptom in movement disorders. It is characterized as a repetitive, oscillatory movement in one or more body parts. Two of the most common movement disorders presenting with tremor are Essential Tremor (ET) and Parkinson’s disease (PD). The objective analysis of tremor commonly include parameters such as frequency, burst duration and time of occurrence. Typically, ET presents with an action tremor while PD presents with a resting tremor. However, studies showed that up to 60% of the PD patients present with a postural tremor and in some cases ET patients have a resting tremor. Furthermore, the tremor frequency in ET is usually higher compared to PD. A characteristic of tremor, often not considered, is its discontinuity. This means that periods of tremor are disrupted by periods of silence or by periods during which the intensity of the tremor is not disabling. Salarian et al. [1] used an autoregressive (AR) model to split gyroscope data of PD patients into tremor (TW) and non-tremor windows (NTW). Heida et al [2] used the same method to determine the effect of DBS on tremor during rest and movement in PD. It was shown that during a TW almost all power was concentrated around the tremor frequency and in a NTW power was mainly concentrated in the movement (0.25 – 3.5 Hz) instead of the physiological (7 – 30 Hz) bands. Whether this method reliably detects TW in different movement disorders and during different tasks has not been tested. Therefore, the objective of this study is to compare two mathematical approaches to detect TW and NTW in hand movement signals recorded from PD and ET patients and healthy, age-matched controls (HC). A reliable method must be able to distinguish between pathological and physiological tremor during different tasks and be able to detect different types of tremor. For this purpose 14 HC, 3 PD and 15 ET patients were included in the study. Subjects performed three tasks: 1) a resting, 2) a postural and 3) an elbow flexion and extension task. Movement was recorded using 3D-accelerometers attached to the hands of the subject. Data was analyzed offline in Matlab using 1) an AR-model and 2) the periodogram function. First results show that both methods reliably distinguish between physiological and pathological tremor during rest and the postural task. During movement the AR-model overestimated the number of TW in ET and PD, and detected false positive TW in healthy subjects. Furthermore, the number of TW during rest was larger in the PD group compared to the other two groups, as expected. It can be concluded that short recordings can be used to reliably detect TW and NTW in movement disorders. Furthermore, these results suggest that this might be a new approach to differentiate between PD and ET.

PHOTOACOUSTIC PERFUSION MEASUREMENTS: A COMPARISON WITH POWER DOPPLER IN PHANTOMS
Maarten Heres, Ümit Arabul, Benjamin Tchang, Marcel Rutten, Richard Lopata
Abstract: Ultrasound-based measurements using Doppler, contrast, and more recently photoacoustics (PA) have emerged as techniques for tissue perfusion measurements. In this study, the feasibility of perfusion measurements with a fully integrated, hand-held, photoacoustic probe was investigated and compared to Power Doppler (PD). Three cylindrical polyvinyl alcohol phantoms were made (diameter = 15 mm) containing 100, 200 and 400 parallel polysulfone tubes (diameter = 0.2 mm), resulting in a perfused cross-sectional area of 1.8, 3.6 and 7.1% respectively. Each phantom was perfused with porcine blood (15 mL/min). Cross-sectional PA images (λ = 808nm, frame rate = 10Hz) and PD images (PRF = 750Hz) were acquired with a MyLab One and MyLab 70 scanner (Esaote, NL), respectively. Data were averaged over 70 frames. The average PA signal intensity was calculated in a region-of-interest of 4 by 6 mm. The percentage of colored PD pixels was measured in the entire phantom. The average signal intensity of the PA images increased linearly (R2 =0.99) with perfusion density, being 0.54 (± 0.01), 0.56 (± 0.01), 0.58 (± 0.01) with an average background signal of 0.53 in the three phantoms, respectively. For PD, the percentage of colored pixels in the phantom area (1.6% (± 0.1%), 3.8% (± 0.3%), 10.9% (± 0.3%)) also increased linearly (R2 = 0.99). These preliminary results show that PA imaging can detect an increase in perfused area with an integrated PA probe and is not inferior to PD. In the future, in vivo measurements will be explored, although validation will be more complex.

ESTIMATING 3D GROUND REACTION FORCES AND L5/S1 MOMENTS DURING ASYMMETRIC TRUNK BENDING USING AN INERTIAL/MAGNETIC SENSOR SUIT
Gert Faber, Chien-Chi (Max) Chang, Idsart Kingma, Jack Dennerlein
Abstract: Background: Wearable inertial/magnetic motion capture (IMC) systems have become increasingly popular for ambulatory 3D kinematics assessment. However, few studies have attempted to use these measurement techniques to estimate kinetic variables. It could be useful for example to be able to estimate ground reaction forces (GRFs) and L5/S1 moments during ergonomic evaluations at the workplace. The aim of the present study was to investigate the performance of a full-body ambulatory IMC system (Xsens) for estimating 3D GRFs and L5/S1 moments during asymmetric trunk bending (n=10). Methods: Using the ambulatory IMC system, GRF were estimated based on the full-body segment kinematics (using Newton’s second law: F=ma), and L5/S1 moments were estimated based on the upper-body segment kinematics using a top-down invers dynamics analysis. As a gold standard reference, a laboratory measurement system was utilized: GRFs were measured with Kistler force plates (FPs), and L5/S1 moments were calculated using a bottom-up inverse dynamics model based on FP data and lower-body kinematics measured with an optical motion capture system (Optotrak). Correspondence between the laboratory and ambulatory measurement systems was quantified by calculating the RMS errors between the outcomes of both measurement systems. Results: Averaged over subjects, GRF RMS errors remained below 8 N (1.5% of Body weight) and L5/S1 moment RMS errors remained below 9 Nm (6% of the peak extension moment). Discussion: In conclusion, the ambulatory IMC system was able to estimate GRFs and L5/S1 moments with reasonable accuracy. A limitation of the current method is that the estimated L5/S1 moments only considers the moments due to the body segments, while in practice there might be external forces acting at the hands (for example during manual materials handling). These hand forces could potentially be estimated by ambulatory GRF measurement using instrumented ForceShoes [1].

USING DAMPING FORCES TO INCREASE THE PERFORMANCE OF QUICK FITTS-LIKE 1-D REACHING TASKS
Arvid Keemink, Arno Stienen, Nienke Schimmel, Jack Schorsch, David Abbink, Herman van der Kooij
Abstract: In the H-Haptics project we are developing a passive assistive lifting device to be used in industry. This device, besides supplying gravity compensation, can display forces to the user. In this project the device will be used to place gears inside a machining device. A Shared Controller is used that acts like a virtual operator of the device. In this way the device can increase situational awareness, increase system stability or improve task performance (reduce task time, reduce vibrations/oscillations, reduce positioning overshoot) for the human operator with minimal increase in cognitive load. Since we want the device to only actively compensate gravity, we can only display passive (energy conserving or dissipating) forces to the operator in the other degrees of freedom. We would like to know whether smart use of a dissipative haptic rendered mechanical force (i.e. damping-like) can reduce the task completion time (increase speed) and targeting overshoot (increase accuracy) in a reaching task, since weightless gear-placement approaches a reaching task. We performed a repeated measures experiment with 12 subjects using an active 1-D admittance controlled device to simulate a passive virtual mass and a dissipative force during a Fitts like 1-D reaching tasks with the right arm. The reaching motion was over a distance of 30 centimeters into a target of 1.5 cm wide and the cursor, target and home positions were displayed on a computer screen. We compared three dampers to base-line (no damping), while doing 30 successful repetitions per condition. If a motion had overshoot or undershoot, the subject had to redo the trial. The three damper conditions were all viscous dampers with varying damping coefficients: 1) exponentially increasing, 2) linearly increasing, 3) exponentially increasing with a different starting point and slope than in condition 1. The damping values around the end point were around 250 Ns/m, starting halfway through the motion. We saw a significant decrease in mean target reaching time of more than 10% (p<0.05) with one of the exponential dampers. The accuracy, measured in the amount of missed trials, improved drastically by a factor of 7 (p<0.01) for all dampers. The subjects tended to apply 3 times higher forces and achieve 5 times higher maximum velocities. Nonlinear optimization models hint towards a human control objective of reducing target error together with muscle activation. In conclusion variable damping seems to be able to reduce the time and increase the accuracy of 1-D reaching tasks. Future work includes looking into the influence of imperfect knowledge of the target location by the device, how humans perform when the moveable inertia is high and using dissipative forces based on more physical phenomena such as friction, dash-pot damping and fluid drag.

AN ONTOLOGY-BASED RECOMMENDER SYSTEM TO PROMOTE PHYSICAL ACTIVITY IN PRE-FRAIL ELDERLY
Mohammad Hossein Nassabi, Harm op den Akker, Miriam Vollenbroek-Hutten, Hermie Hermens
Abstract: Frailty is a prevalent condition in which the older adult is at an increased risk of developing disabilities and adverse events such as dementia, falls, hospitalization and mortality. Physical activity has been shown to be a preventative factor for frailty and its progression [1]. The focus of this research is to keep seniors physically active by designing an intelligent system that recommends exercises tailored to an individual’s health status, goals and preferences. The research is done in the context of PERSSILAA European project. For developing the recommender system, an ontology-based approach is taken as relevant data can arise from heterogeneous sources ranging from multiple software systems like CoCo Physical Trainer [2] and Guttmann Neuro Personal Trainer [3] to activity trackers. The recommender system integrates and reasons about the available data in order to provide the most clinically relevant and motivating exercises to the older adults. To achieve this, the recommendation process is performed 3 steps: 1) Initially, a set of exercises is filtered from a physical activity Knowledge-Base which actually represents the professional’s choice of exercises based on the health status of the older adults 2) The exercises are then personalised based on user preferences, prior exercise history and user feedback. This step also includes the determination of the difficulty level of exercises e.g., number of repetitions 3) At the last step, exercises are presented to the older adult including an explanation of the reason they have been suggested in order to increase the end-user health literacy and self-awareness. At the current stage of the project, the mentioned physical activity Knowledge-Base is under development by interviewing healthcare professionals. The recommendations will be first evaluated with the domain experts e.g., physiotherapist. At the later stages, the effectiveness of the envisaged system will be validated with a group of older adults (65+) in terms of health improvements and adherence to the exercises.

THE EFFECT OF MOTION BLUR ON THE ACCURACY OF WRIST JOINT KINEMATICS DETECTION FROM 4D-CT SCANS
Geert Streekstra, Jim Visschers, Guido de Haan, Iwan Dobbe
Abstract: In any 4D imaging method the representation of a moving object in the image is degraded by motion blur. Although not systematically investigated, motion blur will also be present in 4D imaging of the wrist joint with CT. In literature, the use of dual source high end scanners with superb time resolution has been reported for imaging wrist joint motion [1]. The question is if the time resolution of regular CT scanners would also be sufficient for adequate 4D imaging of wrist joint motion. Therefore the purpose of this study is to investigate the effect of wrist motion on wrist joint kinematics estimation from 4D CT images with a regular 64 slice CT scanner. To determine the kinematic parameters of carpal bone motion, a time series of 3D scans (i.e. 4D-CT image) is acquired and analyzed. This happens in three steps: Image acquisition, image segmentation and image registration [2]. Via this process kinematical parameters of each individual carpal bone can be estimated and represented in a 3D movie showing how the bones move in time. By doing kinematical measurements in 4D CT images of a rotating 3D printed model of a wrist, we establish what the relation is between the position determination errors and the rotational speed in the wrist. The rotational speed is varied in a physiological range to mimic the motion of the carpal bones when a real wrist is moved. The experiments reveal that, within a physiological range of carpal bone velocities, the accuracy of joint kinematics estimation are in the order of 1 degree for rotational and 0.2 mm for translational parameters. Systematic errors in kinematical parameters increase slightly with rotational speed. In general we can conclude that motion of the bones in the wrist influence the estimates of rotational and translational parameters of carpal bones. However, the influence of wrist joint motion on kinematic parameters is limited compared to expected values of kinematical parameters in patients. Therefore we are confident that sufficiently accurate estimation of carpal bone kinematics with regular 64 slice 4D CT scanners in patients is feasible.

HOW THE PARAMETERS AFFECT THE DYNAMIC BEHAVIOUR OF AN INVERTED PENDULUM BALANCE CONTROL MODEL
Vasiliki Spyropoulou, Tjitske Boonstra, Alfred Schouten
Abstract: ABSTRACT During bipedal stance, humans generate corrective joint torques to counteract the destabilizing gravitational torque in order to maintain an upright posture. To investigate how this is achieved and to identify possible causes of balance disorders, neuromuscular modelling is employed. The developed balance control models consist of various parameters able to describe balance behaviour. Theoretically, differences in the values of these parameters could provide insight into pathophysiological changes in certain patient groups. However, even after fitting a model to experimental data, it is hard to recognize the relative contribution of each parameter in specific frequency points. Therefore, a sensitivity analysis was performed in an inverted pendulum balance control model [1],[2], with the goal to identify the unique contribution and the relative importance of the model parameters. In the model the dynamic balance behaviour was expressed with frequency response functions (FRFs), representing the relationship between external perturbations and the response (i.e. body sway), in terms of amplitude and timing. The analysed FRF included the following parameters: the human body mass (m), moment of inertia (J) and centre-of-mass height (h); the intrinsic muscle stiffness (K) and damping (B); the sensory weighting factor (W); the neural controller’s proportional and derivative feedback gains (KP, KD) and time delay (τd); the muscle activation dynamics’ eigen-frequency (ω0) and relative damping (β) and the force feedback’s gain and time constant (KF, τF). The sensitivity of each model parameter was determined for both the FRF magnitude and phase through the analytically obtained partial derivative, using the Matlab symbolic toolbox (Mathworks). The results for the main feedback parameters showed that: a) both K and KP shape the FRF in the lower frequencies (0.1-1 Hz); b) KD has a distinct influence on the FRF, by shaping the peak and slope in the 0.5-0.9 Hz range; c) W influences the overall magnitude of the FRF, without any apparent frequency dependency, and does not have any effect on the phase; d) the effect of τd is difficult to detect in the FRF magnitude, but becomes apparent in the phase above 0.6 Hz. In conclusion, it is possible to uniquely determine the value of the sensory weighting factor, the time delay, the derivative gain, and the lumped sum of the passive and active stiffness, making the used balance control model suitable to determine differences in these parameters in different patient groups. REFERENCES [1] R. J. Peterka, “Sensorimotor integration in human postural control,” J. Neurophysiol., vol. 88, no. 3, pp. 1097–1118, Sep. 2002. [2] M. Cenciarini and R. J. Peterka, “Stimulus-dependent changes in the vestibular contribution to human postural control,” J. Neurophysiol., vol. 95, no. 5, pp. 2733–2750, May 2006.

AUTOMATIC CELL DETECTION AND QUANTIFICATION OF 3D THIRD HARMONIC MICROCOPY IMAGES
Zhiqing Zhang
Abstract: Third-Harmonic Generation (THG) Microscopy is a novel technique to make high contrast microscopic images of samples of tissue. Because no external contrast agents are required it is particularly promising for the study of living brain tissue [1]. THG images are three dimensional images with a resolution ranging from 0.1 to 0.5 µm per pixel in the x and y directions and 1 µm in the z direction, and with depth range up to 350 µm. Visual inspection of THG images has demonstrated rich morphological information like neurons, glial cells, blood cells, fibers – axons and dendrites, blood vessels and other tissue features. Statistical analysis of these images will reveal important characteristics of the tissue. Furthermore, clinical application of THG imaging in tumor diagnosis requires automated methods for quantification and characterization of cells and other structures in THG images. Developing such tools is hampered by the heterogeneity, huge dimensions and complexity of the images. To address this problem we combined several classical image processing techniques and developed our own C++ software library. A smoothing filter was applied for noise reduction. Both background and foreground seeds were detected and fed to a seeded watershed algorithm. A combination of connected component labeling and morphological filters was used to detect cell fragments. Finally, these fragments were merged using an algorithm based on an empirical gradient rule to build the final cell detections and overcome the over-segmentation problem suffered in the watershed algorithm [2]. We applied our algorithm to five different images of health mouse brain tissue and we were able to detect neuronal cells using the same filter parameter settings for each image. Results were verified by visual inspection. Typical computer time was nearly five minutes for images of 1000 x 1000 x 100 pixels. To achieve clinical application, more images need to be tested, and more advanced methods are needed to verify the results. Furthermore, we will consider THG images acquired from human brain tissue. Machine learning techniques will be applied to automatically detected features in order to be able to classify different cell types and states.