Poster session I
Evaluation of a Continuous Ecg Quality Indicator Based on the Autocorrelation Function
Jonathan Moeyersons, Dries Testelmans, Bertien Buyse, Rik Willems, Sabine Van Huffel, Carolina Varon
Abstract: Background: Most electrocardiogram (ECG) signal quality assessment algorithms focus on a two- or multi-level classification. However, it could be argued that signal quality would more naturally occupy a continuum of quality values. Therefore, in previous work we created a continuous quality assessment algorithm based on the autocorrelation function (ACF) . This paper evaluates this algorithm on a simulated dataset with five noise levels and known signal-to-noise ratios (SNR).
Methods: The quality assessment algorithm was evaluated on a large subset of the ECG signals of a polysomnographic dataset. An in-house quality estimation algorithm, with stringent thresholds, was applied to locate the cleanest epochs. Hereafter, calibrated amounts of two types of realistic ECG noise from the MIT-BIH Noise Stress Test Database (NSTDB) were added. Both Electrode Motion (EM) and Movement Artefacts (MA) were considered. For each clean ECG epoch a noisy epoch was randomly selected and a calibrated amount of this noise epoch was added to the clean epoch. The resulting quality values were compared by a Kruskal-Wallis test.
Results: Using only three features and a binary training set, we have shown significant inter-level quality decreases for both types of added noise (p<0.01). Despite this finding, also significant intra-level quality differences were observed, indicating a change in response according to the type of noise (p<0.01).
Conclusion: It has been stated that a continuous quality annotation is hard to validate since it is difficult to create a test dataset. In this paper we tried to circumvent this issue by evaluating a previously developed ECG signal quality indication tool on a simulated dataset with five noise levels and known SNR’s. Despite the simplicity of the algorithm, only three features were used, we have shown significant quality decreases per noise level for both types of added noise.
 J. Moeyersons, E. Smets, J. Morales, A. Villa, W. De Raedt, D. Testelmans, B. Buyse, C. Van Hoof, R. Willems, S. Van Huffel, and C. Varon, “Artefact detection and quality assessment of ambulatory ECG signals,” Under Rev., pp. 1–9, 2018.
Patient-Specific Finite Element Models for Fracutre Risk Prediction in Patients with Cancer and Bone Metastases
Florieke Eggermont, Yvette van der Linden, Loes Derikx, Marianne de Jong, An Snyers, Tom Rozema, Nico Verdonschot, Esther Tanck
Abstract: Cancer patients with bone metastases have an increased risk of fracturing their femur. Treatment is based on the impending fracture risk: patients with a high fracture risk are considered for prophylactic surgery, whereas low fracture risk patients are treated conservatively with radiotherapy to decrease pain. Current clinical guidelines suggest to determine fracture risk with x-rays, but that appears to be non specific (sensitivity=86%, specificity=58%, PPV=23%, NPV=97%). Therefore, we developed a patient-specific finite element (FE) model that has shown to be able to predict fracture risk in an experimental setting. The aim of this study was to determine if these finite element models improve current fracture risk prediction in patients with bone metastases.
66 patients (76 femurs treated with radiotherapy) affected with predominantly lytic bone metastases were included from four radiotherapy institutes in the Netherlands. Most patients were low risk and received 8 Gy. Some intermediate risk patients were too fragile to undergo prophylactic surgery and received 20 Gy in 5 fractions. CT scans were made and patients were followed for six months to determine whether or not they fractured their femur. Non-linear isotropic FE models were created with the patient-specific geometry and bone density obtained from the CT scans, and in the FE simulation, an axial load was applied to the models simulating stance. Failure loads normalized for bodyweight (BW) were calculated for each femur and were compared between the fracture and non-fracture groups. Subsequently, a threshold defining high and low fracture risk was determined and diagnostic accuracy values were calculated.
Femurs of ten patients fractured during follow-up. Median failure loads significantly differed between the fracture and non-fracture groups (5.2xBW and 8.4xBW, respectively, p<0.001). The threshold resulting in the highest sensitivity and specificity was 7.5xBW. Sensitivity was 100%, specificity was 70%, and positive (PPV) and negative predictive values (NPV) were 33% and 100%, respectively.
This study showed that patient-specific FE models can aid in predicting fracture risk in patients with lytic femoral bone metastases, showing higher sensitivity and specificity compared to current clinical guidelines. We are initiating a pilot for clinical implementation of the FE model.
 Van der Linden et al, J Bone Joint Surg Br 86:566-573, 2004.
 Derikx et al, J Bone Joint Surg Br 94:1135-1142, 2012.
Orthostatic Hypotension and Falls in Older Adults: A Systematic Review and Meta-Analysis
Arjen Mol, Phuong Thanh Silvie Bui Hoang, Sifat Sharmin, Esmee Reijnierse, Richard van Wezel, Carel Meskers, Andrea Maier
Abstract: Background: Orthostatic hypotension is a potential risk factor for falls in older adults, but existing evidence on this relationship is inconclusive. The aim of this study was to perform a systematic review and meta-analysis of the literature assessing the association between orthostatic hypotension and falls.
Methods: A literature search was performed on February 20th 2017 in MEDLINE (from 1946), PubMed (from 1966) and EMBASE (from 1947) using the terms “orthostatic hypotension”, “postural hypotension” and “falls”. References of included studies were screened for other eligible studies. Study selection was performed independently by two reviewers using the following inclusion criteria: published in English; mean/median age of the population ≥ 65 years; blood pressure measurement before and after postural change; assessment of the association of orthostatic hypotension with falls. The following studies were excluded: studies provoking orthostatic hypotension by exercise or medication; conference abstracts, case reports, reviews and editorials. Data extraction was performed independently by two reviewers, in accordance with the PRISMA and MOOSE guidelines. Unadjusted odds ratios were used for pooling using a random-effects model. Studies were rated as high, moderate or low quality using the Newcastle Ottawa Scale.
Results: Out of 5646 studies, 63 studies (51,800 individuals) were included in the systematic review and 50 studies (49,164 individuals) in the meta-analysis. Out of 63 studies, 39 were cross-sectional and 24 were longitudinal. Orthostatic hypotension was positively associated with falls (odds ratio 1.73, 95% confidence interval 1.50-1.99). The result was independent of study population, study design, study quality, orthostatic hypotension definition and blood pressure measurement method.
Conclusions: Orthostatic hypotension is significantly positively associated with falls in older adults, underpinning the clinical relevance to test for an orthostatic blood pressure drop and highlighting the need investigate orthostatic hypotension treatment to potentially reduce falls.
Design of a Dynamic and Adaptive Head Support
Anoek Geers, Paul Groenland, Arjen Bergsma, Bart Koopman
Abstract: For people with severe muscle weakness or paresis in the trunk and neck muscles, adequate head support is required. Although several assistive devices exist that can support a person’s head position, there is an absence of devices that are capable to support head movements in a natural and safe way. The large individual variation between users requires an individual match between user and assistive device. From initial market research it can be concluded that there is a need for assistive devices that provide dynamic adjustability by combining changes in position of the trunk and head with continuous stabilization. Within the project, the main objectives were to characterize this need for support, and to develop a first proof-of-concept of a dynamic and adaptive head support.
Position control was implemented on an actuated head support system with four degrees of freedom (e.g. three translations and one rotation in flexion-extension), using a six degree-of-freedom force sensor as a joystick interface. For the current control method, manipulation of the joystick results in the head support following part of the natural flexion-extension motion of the head, coupling multiple degrees of freedom of the actuated system. Additionally, the system can autonomously adapt the head support position according to the back seat angle of the electric wheelchair, to compensate for changes in posture relative to the wheelchair seat caused by changing seat settings.
Initial functional testing shows that the current prototype matches the majority of the requirements set in the design phase. Compared to current solutions, the presented system can steer the head support position in 3D in a more efficient and natural way. Therefore, it can be concluded that the redesigned system is a promising first step in the development of a new generation of dynamic and adaptive head supports that are intelligent enough to autonomously personalize their behavior to the user.
A Novel Setup and Protocol to Measure the Range of Motion of the Wrist and the Hand
Kostas Nizamis, Noortje H.M. Rijken, Ana Mendes, Mariska M.H.P. Janssen, Arjen Bergsma, Bart F.J.M. Koopman
Abstract: The human hand is important for the performance of activities of daily living which are directly related to quality of life. Various conditions, such as Duchenne muscular dystrophy (DMD) can affect the function of the human hand and wrist. The ability to assess the impairment in the hand and the wrist by measuring the range of motion (ROM), is essential for the development of effective rehabilitation protocols. Currently the clinical standard is the goniometer. In this study we explore the feasibility and reliability of an optical sensor (Leap motion sensor) in measuring active hand/wrist ROM. We measured the hand/wrist ROM of 20 healthy adults with the goniometer and the Leap motion sensor, in order to check the agreement between the two methods and additionally, we performed a test-retest of the Leap motion sensor with 12 of them, to assess its reliability. The results suggest low agreement between the goniometer and the leap motion sensor, yet showing a large decrease in measurement time and high reliability when using the later. Despite the low agreement between the two methods, we believe that the Leap motion sensor shows potential to contribute to the development of hand rehabilitation protocols and be used with patients in a clinical setting.
Nonlinear System Identification of the Human Auditory Pathway
Elisabeth Noordanus, Lei Wang, John van Opstal
Abstract: Auditory steady-state response (ASSR) is the sustained response on a periodic auditory stimulus, e.g., an amplitude-modulated sine. The modulation frequency(ies) of these tones can be measured using EEG. The clinical use of ASSR is the assessment of auditory function and hearing loss in newborns and young children. In this study, we used ASSR to explore nonlinearity in the human auditory system. Specific nonlinearities can putatively be used as the hallmark of function or dysfunction of a processing level in the auditory pathway (including the cochlea and the auditory nerve).
Our main objective is to develop a reliable method to characterize the monaural and binaural auditory performance of hearing-impaired listeners, to be used for automatic fitting of hearing aids and cochlear implants. The secondary objective is to localize generators of significant ASSR associated with different nonlinearities of the auditory pathway.
Methods: We performed two experiments to evaluate 2nd and 3rd-order nonlinearities of the auditory system. In both experiments (employing different frequency combinations) we presented stimuli as a superposition of four pure tones, two in the left and two in the right ear. The frequencies were chosen such that each interaction component between two or more sinuses would yield a unique frequency. Ten normal-hearing subjects were measured on two different days to check response repeatability. We used spectral analysis to identify frequencies with a significant response.
Results: Major frequencies generated by 2nd (e.g., f2-f1) and 3rd-order (e.g., 2f2-f1) nonlinearities could be reliably detected in the EEG signals, which indicates that both 2nd and 3rd-order nonlinearities exist in the human auditory system. In general, the interactions between frequencies presented in the same ear were stronger than the interactions of frequencies across ears; moreover, 2nd-order responses were typically stronger than 3rd-order responses. The binaural response resulted to be larger when stimuli generated no 2nd-order monaural response. We also found that EEG electrodes above the temporal lobe and the central-frontal area yielded the highest signal-noise ratios. Finally, within-subject variability is smaller than between-subject variance.
Conclusion: ASSR can be used to assess nonlinearities in the human auditory systems, both on a generic and an individual level.
Detecting Sleep Apnea Using Pulse Photoplethysmography
Margot Deviaene, Jesús Lázaro, Dorien Huysmans, Dries Testelmans, Bertien Buyse, Sabine Van Huffel, Carolina Varon
Abstract: Aim: Obstructive sleep apnea (OSA) often remains undiagnosed. Therefore, this study investigates the use of pulse photoplethysmography (PPG) for the detection of sleep apnea and its added value to oxygen saturation (SpO2).
Methods: A dataset of 102 subjects, suspected of having OSA was recorded, the data was split in 1 minute segments to be classified as apneic or normal breathing. The PPG signals were preprocessed and 6 PPG series were extracted: the pulse rate, amplitude and width variabilities, slope transit time, maximal pulse upslope and the area under the PPG pulse . Moreover, the instantaneous powers in the high and low frequency bands of the pulse rate were estimated using a point-process model . For all these series, 5 features were computed over a 1 minute interval: the mean, the minimum and the maximum value, the standard deviation and the gradient. Feature selection resulted in the 6 most discriminative features for PPG-based detection of apneic intervals. These features were used as input for a least-squares support vector machine classifier. A second classifier was trained including SpO2 features extracted as described in .
Results: A classification accuracy of 68.7 % was achieved for PPG-based apnea detection. If only the most severe events with complete breathing cessations are considered, this value rises to 74.8 %. When SpO2 features were added to the classifier the accuracy increased to 83.4 %, which is only slightly higher than the 82.2 % obtained using only SpO2. These results suggest that the studied PPG features have potential for sleep apnea detection, and that, however, their added value to SpO2 is limited.
 Lázaro J. et al. Pulse photoplethysmography derived respiration for obstructive sleep apnea detection. Computing in cardiology 44 2017;1.
 Barbieri R. et al. A point-process model of human heartbeat intervals: new definitions of heart rate and heart rate variability. Am J Physiol Heart Circ Physiol 2005;288.1;H424-H435.
 Deviaene M. et al. Automatic screening of sleep apnea patients based on the SpO2 signal. IEEE JBHI 2018; Early Access; DOI:10.1109/JBHI.2018.2817368.
Automated 3D Ultrasound for Quantitative Assessment of Mechanical Properties of Skeletal Muscle
Leon de Jong, Anton Nikolaev, Anna Greco, Jurgen Fütterer, Chris de Korte
Abstract: Purpose: This study focuses on providing new parameters within the diagnostic workup for patients having (a suspicion of) neuromuscular disease. Even though many of these patients show muscle weakness, the underlying mechanism of the disorder varies heavily. Over 600 different types of disorders are known in literature, indicating the need for adequate diagnostics. Myopathy patients form a group of neuromuscular disorders that suffers from mechanical changes in skeletal muscle tissue. In this study, we assess if automated 3D ultrasound can be used to quantify parameters related to these mechanical properties in a healthy population.
Study population: 124 healthy volunteers with 11 different nationalities participated in this study; 65 males and 59 females, age 38 ± 16, BMI 21.4 ± 4.3, 12 lefthanded, 110 righthanded 2 ambidexter. Volunteers declared to have no prior medical conditions relevant to this measurement.
Method: The overall parameters that were measured prior to the ultrasound examination are; weight, bone mass and water fraction. Segmental parameters (arms/legs/core) are; fat fraction and muscle mass. Medical Research Council (MRC)-scores were acquired by physical examination by a physician and maximum hand grip strength was measured for both hands. Automated 3D ultrasound scans of the tibialis anterior muscle were performed whilst maintaining isometric contractions at 0, 20% and 40% of maximum voluntary contraction (MVC) force. Acquisition planes were in transversal as well as sagittal direction.
Results: Initial analysis of 10 subjects show an increasing pennation angle with respect to the initial angle from 1.1° (0-20% MVC) via 1.8° (20-40% MVC) to 3.0° (0-40% MVC), where positive values represent increasing angles.
Future implications: This study provides a non-invasive method to assess quantitative parameters that are of interest in the diagnostic track of neuromuscular disorders, especially myopathies. We suggest that the acquired parameters (echogenicity, muscle volume and architecture, fascicle length and pennation angles) of this dataset can be used to investigate disease gradation, progression and treatment effect.
Characterization of Glaucomatous Rat Eye Optic Nerve Head Biomechanics Through Individual-Specific Computational Modeling
Su Guvenir, Stephen A. Schwaner, Amir A. Zadpoor, C. Ross Ethier
Abstract: Gradual vision loss in glaucoma is caused by the death of retinal ganglion cells (RGCs) primarily in the optic nerve head (ONH). Elevated intraocular pressure (IOP) is a well-known risk factor. Rats are widely used to better understand the biomechanics-driven cell processes between the IOP and the RGC apoptosis. The effect of early stage remodeling is important to understand, and glaucomatous rat eye ONH biomechanics has never been characterized. This research , for the first time, characterizes glaucomatous rat eye ONH biomechanics by using individual-specific finite element (FE) modelling.
Tissue delineation marks from three-dimensional (3-D) histomorphometric reconstructions  were used to build two individual-specific glaucomatous rat ONH FE models (FEMs), which were embedded into a generic posterior eye model. Since the IOP during tissue fixation was reported to be 10 mmHg, FEMs were solved for the IOP change from 10 to 30 mmHg . Moreover, as an upper IOP boundary, experimentally measured IOP values of 32.5 mmHg and 34.6 mmHg  were also applied to solve FEMs. As the IOP cause deformations, the mean and the 95th percentile of the 1st (maximum tension) and the mean and the 5th percentile of the 3rd (maximum compression) principal strains on the anterior ON were calculated.
The mean of the 1st principal strain ranged from 4.90% to 5.30% and the 95th percentile was 9.48% for both eyes. The mean and the 5th percentile of the 3rd principal strain magnitudes ranged from -4.74% to -4.76% and -7.52% to -8.86%, respectively. Larger strains were located at the inferior side of the anterior ON and strain magnitudes were higher compared to healthy rat ONH FEMs . Relative differences in strains between the healthy and glaucomatous eyes increased when experimentally measured IOP values were applied.
To conclude, this research is the first to characterize glaucomatous rat eye ONH biomechanics and suggests that alterations in the ONH geometry due to remodeling and damage in early stages of glaucoma caused higher deformation and larger strains on the inferior side of the anterior ON. These models will also be favorable to better interpret results obtained from rat experimental studies.
Imaging the Contraction of Mechanically Supported Ex Vivo Beating Hearts
Louis Fixsen, Niels Petterson, Frans van de Vosse, Marcel Ruttern, Richard Lopata
Abstract: In mechanical circulatory support (MCS) patients, clinicians lack tools to monitor the condition of the heart due to metals in the body. Factors such as LV dP/dtmax and contraction pattern are valuable in assessing cardiac function, but have yet to be quantified in MCS patients. Ultrasound (US) strain imaging could enable the study of these parameters. However, due to a high level of illness it is difficult to validate new methods in MCS patients. Therefore, in this study, 2-D US strain imaging was used to investigate changes in heart mechanics in left ventricular assist device (LVAD) supported hearts, with an isolated beating porcine heart platform.
Ex vivo porcine hearts were implanted with Thoratec HeartMate II (n=2) and MicroMed DeBakey (n=2) LVADs. The hearts were attached to a mock-loop, re-perfused with oxygenated blood, resuscitated and paced at 120 bpm (PhysioHeart, LifeTec). Measurements were performed whilst the LVAD-supported hearts deteriorated from baseline condition (based on cardiac output). Radio-frequency US data were acquired (MyLab70, Esaote) at pump speeds from 0 (no outflow) to 10.5 krpm. Data were manually segmented and local radial (erad) and circumferential strains (ecir) were estimated over each heart cycle. The unloaded 0 krpm initial geometry was used as an initial condition for further pump speeds per heart condition.
In each heart, as the pump speed was increased and the hearts degraded, measured dP/dtmax reduced. An increase was seen in ecir magnitude relative to the reduction in pressure, showing that ecir estimated using US strain imaging can be related to dP/dtmax. Time-to-peak strain around the left ventricle was measured. As the pump speed was increased and the heart condition degraded, regions of early and late contraction grew in size and magnitude. This is a first step in unravelling the interplay between pump-action and remaining cardiac function.
This work was funded by the EU-MSCA. GA No 642612 (www.vph-case.eu).
Time-to-Boundary Measures as Measures of Balance in Chronic Stroke Patients Compared to Healthy Individuals
Anne L. van Daalen, Elise J. Padmos, Sarah B. Zandvliet, Carel G.M. Meskers, Gert Kwakkel, Erwin E.H. van Wegen
Abstract: An important consequence of stroke is postural instability, associated balance problems and falling. Adequate standing balance performance is related to perception of spatiotemporal margins relative to stability boundaries at the support surface. Boundary relevant measures are altered in neurologic patients compared to healthy individuals, however they have never been assessed in patients with a stroke.
Objectives: Primary objective was to examine differences in boundary relevant measures between stroke patients and healthy controls. Secondary we studied the relationship between boundary relevant and traditional posturographic measures.
Methods: In 24 chronic patients and 10 healthy age-matched individuals ground reaction forces were measured during quiet stance. Posturographic time-to-boundary (TtB) measures (mean TtB, TtB variability, TtB temporal distance (TD) and TD variability and CoP measurements (range, sway, path) in two conditions (eyes open, eyes closed) were calculated.
Results: Stroke patients had a significantly lower mean TtB compared to healthy individuals in medio-lateral (ML) direction (mean(m)=7.89s (standard deviation (sd)=1.06) vs. m=16.49s, (sd=1.64), resp. p<0.01), and a significantly lower ML TtB variability (m=1.31s (sd=0.16) vs. m=2.52s (sd=0.24), resp. p<0.01). While no difference was found in anterior-posterior (AP) direction for mean TtB (stroke m=2.52 (sd=0.41) vs. healthy m=3.43s (sd=0.63), p=0.23), and TtB variability (stroke m=0.45s (sd=0.07) vs. healthy m=0.57s (sd=0.11), p=0.38). No significant differences were found for mean TD or TD variability for both directions. CoP sway and range in both directions and CoP path were significantly higher for stroke patients compared to healthy individuals. A significant negative correlation was found between mean TtB and CoP sway in ML direction (t=-0.66, p<0.01) and between mean TtB and CoP range in ML direction (t=-0.67, p<0.01) in stroke patients.
Conclusion: Boundary relevant measures are altered in the chronic phase after stroke. Separate AP and ML TtB measures provide additional directional information. Lower mean ML TtB in stroke patients indicate lower spatio-temporal stability margins, likely related to the hemiparesis or balance asymmetry. Boundary relevant measures could contain critical information concerning the postural control system.
A Heart-Rate Variability Model for the Maturation of Premature Infants
Mario Lavanga, Elisabeth Heremans, Alexander Caicedo, Katrien Jansen, Els Ortibus, Gunnar Naulaers, Sabine Van Huffel
Abstract: Multiple studies have shown that the analysis of heart-rate variability (HRV) can be used to assess the development of premature infants . However, due to the fact that these patients experience sudden bradycardias, the HRV time course can be disturbed . The objective of this study is to propose an autonomic growth model for newborns and assess the effect of the bradycardias on its output.
ECG was recorded from 25 premature patients with a post-menstrual age ranging from 27 to 42 weeks. After extracting HRV, each recording was segmented in 10 minutes windows, which were grouped in segments with bradycardias and in between bradycardias. Standard temporal, spectral and fractal features, such as the RR standard deviation ( ), the ratio between VLF and LF power bands, the regularity exponent (H) and its variation (C2), were computed. The time, spectral and fractal features were used in three linear mixed-effect models to predict infants’ PMA. Each model was assessed via the mean absolute error (MAE) for both the segments with and without bradycardias.
Overall, the HRV parameters were significantly correlated with age (Pearson correlation = 64%, H = -43%, C2 = 53%, VLF/LF = -40%) if bradycardias were excluded, but these correlations were lost in case of bradycardia inclusion ( = 4%, H = -36%). These results are supported by the time, spectral and fractal maturation models: the errors obtained were 2.57, 2.38, 2.25 weeks in case of bradycardias inclusion, but the MAEs shifted to respectively to 1.78, 2.02, 2.31 weeks if those events were excluded. In conclusion, bradycardias should be filtered in the HRV analysis to track preterm development via autonomic age model, whose results are in line with more common EEG based models.
Laparoscopic Sentinel Node Biopsy Using Differential Magnetometry
Melissa van de Loosdrecht, Sebastiaan Waanders, Erik Krooshoop, Bennie ten Haken
The aim of this study is to develop a novel laparoscopic probe for magnetic sentinel node biopsy. The latter is a procedure to determine if a tumor has metastasized via the lymphatic system , enabling personalized patient care.
Superparamagnetic iron oxide nanoparticles (SPIONs) are used as a tracer to find sentinel nodes in vivo. The principle that we use to locate them is Differential Magnetometry (DiffMag) . In DiffMag, the nonlinear magnetic properties of SPIONs are used, enabling selective detection in the diamagnetic patient.
We propose a setup with mechanically separated excitation and detection coils. As a result, the size of the excitation coils can be increased and placed outside the body. The detection coils can be made small enough to be used in laparoscopic surgery. However, the main challenge of this setup is movement of detection coils with respect to excitation coils. Therefore, the detector signal is hindered by the excitation field, requiring continuous active compensation.
We developed and implemented active compensation and tested it in a static setup. It was possible to measure small amounts of SPIONs, down to 25 μg Fe. Furthermore, it was possible to measure SPIONs at a distance up to 20 cm from the top of the excitation coils. Surgical steel and diamagnetism of water – and thus of tissue – have minor influence on DiffMag measurements.
Separation of excitation and detection coils is unique and not possible without DiffMag. These first results are promising for laparoscopic sentinel node biopsy with magnetically marked nodes, which helps to improve cancer patient care. Moving the detector is a challenge that we will solve by using faster electronics, enabling real time compensation of the excitation signal.
 Giuliano and Gangi, 2015.  Waanders et al., 2016.
Culturing Human Osteoarthritic Osteochondral Explants in a Double Chamber Culture Platform
Meike W.A. Kleuskens, René van Donkelaar, Keita Ito
Abstract: Current knee osteoarthritis (OA) treatments are sub-optimal or not long-lasting, and tissue engineering might provide a promising alternative. Within the RegMed XB/InSciTe Mimicart project, regenerative osteochondral implants for focal defects are being developed. The ability of the implants to regenerate cartilage and bone and integrate with the adjacent tissues needs to be explored and maximized. Aim of the present study is to characterize human osteochondral tissue at various stages of OA, and to develop an approach to culture OA tissue for 28 days while maintaining the biochemical and histological composition of the explants.
A double chamber culture platform was used to culture Ø10mm human osteochondral explants. In this culture platform the cartilage and bone tissue of an osteochondral plug can be cultured in two separated media compartments, allowing for the supplementation of tissue specific medium components. Previous research has shown that Ø8mm porcine osteochondral explants can be cultured up to 56 days in this system, while maintaining cartilage tissue content, structure and mechanical properties.1
Osteochondral explants with either a smooth or fibrillated cartilage surface, representing different OA stages, were harvested from human tibia plateaus obtained from total knee replacement surgeries at the Maxima Medical Centre, Eindhoven. The native and cultured explants were evaluated for their viability using an MTT assay and an LDH assay, biochemical tissue content and cell gene expression. Moreover, samples were histologically stained with safranin-O/fast green and picrosirius red for analysis of the proteoglycan and collagen distribution and to determine the grade of OA using the Mankin scoring system.
OA cartilage of various stages, including tissue with smooth and fibrillated surfaces, was evaluated and statistical differences were found in biochemical and histological analyses at day 0 (average Mankin score of 3.4 vs. 5.1, average proteoglycan content of 8.4±1.7%dw vs. 13.5±3.2%dw, average collagen content of 39.9±3.8%dw vs. 29.3±4.6%dw, for tissue with a smooth or fibrillated surface respectively). Preliminary results of the cultures reveal that chondrocyte viability was maintained over 28 days, but bone tissue was less viable. Osteochondral explant culture is ongoing to explore whether the biochemical and histological composition of the cartilage is preserved over time.
Inter-Laboratory Comparison of Gait Waveforms in Individuals with Knee Osteoarthritis
Jim Schrijvers, Derek Rutherford, Rosie Richards, Josien van den Noort, Martin van der Esch, Jaap Harlaar
Abstract: Purpose: Gait analysis is regularly used to test efficacy of interventions and to better understand knee function in individuals with knee osteoarthritis (OA). Assessment of gait is performed in many laboratories across the world utilizing custom setups and protocols. This creates the potential for large inter-lab variance when comparing results. Dalhousie University (DAL, Canada) and VU medical center (VUmc, the Netherlands) are striving to establish a collaborative initiative to combine resources in future gait studies in individuals with knee OA. Therefore, the aim was to evaluate the inter-laboratory comparison of previous collected knee gait waveforms of individuals with knee OA.
Methods: Knee gait waveforms (three-dimensional (3D) angles, 3D moments and muscle activations) were analyzed of the affected leg of individuals with moderate KOA previously collected with a R-mill instrumented treadmill, a motion capture system and electromyography at DAL (n=55) and VUmc (n=40). Inter-laboratory comparison was performed using analysis of variance models of the population characteristics, common discrete values of the waveforms (peak, initial contact (IC) and impulse values) and principal pattern (PP) scores (resulted from principal component analysis (PCA)).
Results: The DAL population was shorter (0.04m; p=0.02), heavier (5.9kg; p=0.04), had higher knee flexion strength (20Nm, p<0.01) and lower pain scores (WOMAC1 Pain: 6; p<0.01). PCA identified an amplitude offset in the sagittal and frontal plane angles (PP1-scores: p<0.01; 11.8˚ and 4.1˚, respectively), as well as in the sagittal plane moments (PP1-scores: p<0.01, 0.16 Nm/kg). The frontal plane moments were identical, as shown by similar PP-scores and discrete values. The gait waveforms of the quadriceps, hamstrings and gastrocnemius had a phase shift between laboratories (PP2 or PP3-scores: p<0.01; 4%, 4% & 6% of gait cycle, respectively).
Conclusion: This inter-laboratory comparison of treadmill gait in individuals with knee OA revealed systematic offsets (amplitude and phase) that obstruct the merging of gait datasets from laboraties. Marker models, collection and processing decisions (i.e. filters and IC identification) may explain these results. There is a need for a standardized, harmonized gait analysis protocol to enable combined databases and multi-center trials for knee OA gait analyses.
Effect of Beta-Blockade on Renal Autoregulation in Experimental Endotoxic Shock
L.M. van Loon
Heart rate (HR) control with esmolol could benefit central hemodynamics during septic shock, but might reduce organ perfusion by reducing perfusion pressure. We assessed the effect of HR control with esmolol on renal blood flow during early sepsis.
Sepsis was induced in 10 healthy anesthetized and mechanically ventilated sheep by continuous IV-administration of lipopolysaccharide. After successful resuscitation of the septic shock with fluids and vasoactive drugs, esmolol was infused to reduce HR but stopped 30-min after reaching a targeted HR reduction of 30%. Arterial and venous pressures, and renal blood flow (RBF) were recorded continuously.
Renal perfusion pressure (RPP) - i.e. arterial minus venous pressure - was not significantly altered during resuscitated septic shock while beta-blockade with esmolol significantly reduced both RPP and RBF. Positive correlations were found between RPP and RBF. This correlation was weaker when RPP was above 70% of baseline, with RBF displaying a relative ‘plateau’ above this threshold. Esmolol increased the lower limit of renal autoregulation. This resulted in the disappearance of the physiological ‘plateau’ and returned after stopping esmolol. Showing the reversibility of these effects.
In our experimental animal model, renal autoregulation remained preserved despite acute endotoxic shock. Nevertheless, during the subsequent strong beta-blockade with esmolol the lower limit of autoregulation was increased and RBF was significantly reduced as a consequence of a low RPP.
Effect of Calcification & Fibrous Tissue Features on Rupture Risk in Atherosclerotic Plaques
Bas Vis, Hilary Barrett, Frank Gijsen, Ali Akyildiz
Abstract: Myocardial infarction and stroke are majorly caused by atherosclerotic plaque rupture, which correlates with high mechanical stresses in plaques1. A potential stress concentration location in plaques is the calcification-fibrous tissue interface. Histopathological examinations demonstrated plaque ruptures located at this interface2. This study aims to identify critical geometric features of calcification and fibrous tissue structure for stress concentrations, through a comprehensive morphometric analysis and implementing fibrous tissue anisotropy in a plaque calcification model for the first time.
Histological cross-sections (n=65) of carotid plaques (n=16) were morphometrically analyzed for calcification-related geometric features and for fiber orientation patterns surrounding calcification. Based on these measurements, 145 finite element models (FEMs) were generated. Using anisotropic, hyperelastic material models; FEMs were solved for an intraluminal pressure of 140 mmHg, and plaque stresses were computed.
Of the 145 calcified plaque regions analyzed, four distinct fiber patterns were identified: attached to calcification (40%), encircling (25%), pushed-aside (15%) and randomly distributed (10%). Most calcifications had the long axis oriented circumferentially, with the median[Q1:Q3] length/width ratio of 1.97 [1.23:3.02]. Calcification width/plaque thickness ranged from 0.02 to 0.96 with a median of 0.12. For attached fiber pattern, calcifications were evenly distributed over plaque thickness whereas for other three patterns calcifications were closer to outer edge.
For attached fiber FEMs, max principal stress at calcification-fibrous tissue interface demonstrated a wide variation, with a median [Q1:Q3] of 277 [156:543] kPa. For pushed-aside pattern FEMs, the median [Q1:Q3] stress was 85 [41:165] kPa. Encircling and random pattern models showed stress values lower than 10 kPa. Multivariable regression analysis (r2=0.88) showed that interface stresses significantly positively correlated with calcification length/width ratio and this effect was amplified with increased calcification width/plaque thickness ratio.
This study is to first to 1.)comprehensively evaluate calcification-related morphometric features in atherosclerotic plaques and fiber orientation patterns around calcifications, 2.)implement fibrous tissue anisotropy for stress analysis at calcification-fibrous tissue interface, and 3.)analyze the effect of morphometric features and fiber orientation on interface stresses. Study results demonstrate that combination of larger calcification length/width ratio, larger calcification width/plaque thickness, and attached fiber pattern significantly increases stresses at calcification-fibrous tissue interface, a plaque rupture location.
 Cheng, G. C., Loree, H. M., Kamm, R. D., Fishbein, M. C., and Lee, R. T. (1993), Distribution of circumferential stress in ruptured and stable atherosclerotic lesions: A structural analysis with histopathological correlation, Circulation, 87(4), 1179-1187.
 Daemen M. J., Ferguson M. S., Gijsen F. J., Hippe D. S., Kooi M. E., Demarco K., van der Wal A. C., Yuan C., and Hatsukami T. S., Carotid plaque fissure: an underestimated source of intraplaque hemorrhage, Atherosclerosis 254, 102 (2016).
The Role of Hemodynamics in the Rupture of an Intracranial Aneurysm
Romana Perinajová, Merel Toussaint, Pim van Ooij, Saša Kenjereš
Abstract: Intracranial aneurysms cause almost 500,000 deaths in the world every year. Currently, the processes behind their genesis and rupture are not well known. New opportunities for both the treatment as well as the prevention may be found if we have better understanding these processes. Here we show how the combination of Magnetic Resonance Imaging (MRI) and Computational Fluid Dynamics (CFD) provides a non-invasive alternative for studying the intracranial aneurysm. The importance of boundary conditions in an aneurysm simulation was assessed by comparing the simulation results to 7T MRI velocity data, obtained from the Academic Medical Centre (AMC) in Amsterdam. Adequate similarities were found in velocity values, together with qualitative agreement in the wall shear stress.
In previous studies, it was hard to find the precise location of rupture in an aneurysm. The aneurysm geometry of the CFD Rupture challenge from 2013 (J of Biomech. Eng. 2015) was simulated to predict the location of a known rupture site. This rupture location was predicted by combining few hemodynamic criteria. Namely, the time-averaged wall shear stress (WSSTA), oscillatory shear index (OSI) and vortex-saddle point structure during systole with accompanying local pressure minima. Thanks to a sensitivity study on these criteria a critical threshold for rupture risk was proposed and it was possible to predict the exact rupture site for the intracranial aneurysm from the CFD challenge. Therefore, the risk assessment of the rupture site in an aneurysm can be used as a more accurate tool to evaluate the need for a dangerous preventive surgery. By applying the same evaluation of the hemodynamical properties in the patient aneurysm, we were able to localize a possible rupture site in this case as well. These results are the first glance into the better understanding the origins of the aneurysm rupture.
Multimodal Non-Invasive Measurement of Cardiovascular Responses to Postural Changes
Arjen Mol, Andrea Maier, Carel Meskers, Richard van Wezel
Abstract: Background: Orthostatic hypotension (OH) is associated with poor physical performance, but this association is weak. This may be due to the involved compensatory physiological mechanisms, which are not fully captured by the onetime OH diagnosis as reflected by its poor reproducibility. Continuous, non-invasive, ambulatory assessment of blood pressure (BP), posture and activity of compensatory physiological mechanisms is needed to obtain insight in the relationship between posture, blood pressure and physical performance. Photoplethysmography (PPG) and near-infrared spectroscopy (NIRS) are measurement techniques reflecting systemic and cerebral circulation and suitable for continuous, non-invasive measurements. However, responses of these signals to postural changes in healthy adults must first be addressed to perform measurements in patients with OH. This study assessed signal sensitivity and reliability in healthy adults.
Methods: PPG applied to the radial and digital artery of the left hand, bifrontal NIRS, continuous BP and ECG were measured in 34 healthy subjects (mean age 31.9 years, standard deviation 14.4). 16 were subject performed sit to stand (preferred speed), slow supine to stand (within 3 seconds) and rapid supine to stand (10 seconds) test conditions, while 18 performed sit to stand (preferred speed), supine to stand (preferred speed), head up tilt (15 seconds), Valsalva (15 seconds) and squat (1 minute) test conditions. Sensitivity was defined as the change of the signals during the test conditions relative to baseline variability. Test-retest reliability of the signals for each test condition was assessed and expressed using intraclass correlation coefficients.
Results: The PPG and NIRS signals were sensitive to the test conditions. NIRS signals showed a large drop within 30 seconds, and a small drop between 30 and 180 seconds after standing up, both with a test-retest reliability ICCs > 0.8.
Conclusion: PPG and NIRS measures were sensitive to postural change in healthy adults, indicating their potential value as a combined barocontrol monitor in older adults with orthostatic hypotension.
Influence of Casting Induced Asymmetry and Velocity of Gait on the Margin of Stability
Mique Saes, Mohamed Irfan Mohamed Refai, Ilona Visser, Peter Veltink, Jaap Buurke, Hans Bussmann, Gert Kwakkel, Bert-Jan van Beijnum, Erwin van Wegen, Carel Meskers
Many stroke patients suffer from distorted gait patterns. Previous research show that the Margin of Stability (MoS), defined as the shortest distance between the projection of the extrapolated center of mass (XCoM) and the Base of Support (BoS)(Hof et al., 2008), is a potential relevant metric that varies with different balance impairment levels (Van Meulen et al. 2006; Bruijn et al., 2013). However, the influence of different gait patterns on MoS is not yet systematically explored. We investigated the influence of casting induced asymmetry and changes in walking speed on MoS in healthy individuals, which can be used to understand compensatory strategies and ultimately gait recovery post stroke.
Eight healthy participants walked at different speeds (0.4m/s, 0.8m/s, 1.2m/s and preferred speed), under two conditions: with and without a complete restriction of the dominant ankle using plaster casting. Kinematic gait measurements were performed on a GRAIL treadmill system (Motek, Inc.) including force platforms and ten infrared cameras. MoS was calculated in anterior-posterior (MOSAP) and mediolateral (MOSML) direction (van Meulen et al., 2016) at initial contact of the right and left foot separately.
MoSAP and MoSML increased with gait speed (p<0.001). Step length of the restricted foot was smaller when walking at >0.8m/s compared to the non-restricted condition (p<0.05). MoSAP was further outside the BoS in casted compared to unrestricted condition at 0.8m/s (p<0.05). MoSAP is outside the BoS at 0.8m/s or faster. Casting resulted in increased MoSML towards the restricted side (p<0.001). This increase was correlated with the increase in step width (r=0.74,p=0.04).
MoS changes with gait speed and asymmetry. The shift of MoS to the casted side is probably caused by an increased step width. Healthy individuals are able to keep their XCoM outside their BoS in AP direction with unilateral casting of the ankle joint, compensation strategies in stroke patients are probably different.
MoS is responsive to asymmetrical walking patterns and changes in gait speed. MoS might be a candidate metric to detect changes and progress in stroke rehabilitation during gait.
Review and Simulations of Upper Extremity Movement Smoothness Measures in Stroke for Reaching
Bouke Scheltinga, Mohamed Irfan Mohamed Refai, Carel Meskers, Erwin van Wegen, Gert Kwakkel, Bert-Jan van Beijnum
Abstract: A measure for quality of movement during reaching tasks is ‘movement smoothness’. It is related to the continuity of a movement, independent of amplitude and duration of the movement . In stroke patients, it is shown that as recovery proceeds the movements become smoother. More specifically, this occurs mainly within the first 8 weeks post stroke, which is approximately the same time window as the improvement of motor control and capacity according to clinical assessments . It is assumed that these improvements early after stroke are mainly the results of spontaneous neurological recovery . This spontaneous recovery is however yet poorly understood. Studying movement quality during stroke recovery is vital to better understand the recovery process after stroke . During the past two decades, over 20 different smoothness measures were used in research with stroke patients, yet there is no standardised measure for smoothness. This makes it hard to compare the different measures reported in literature. The purpose of this study is to make a step towards identifying a standardised measure for movement smoothness for stroke subjects.
First a literature review was done to get an overview of all available measures for smoothness in reaching with stroke subjects. Subsequently, velocity profiles that mimic forward reaching movements were simulated. The behaviour of the metrics was studied while we varied different parameters in a reaching simulation, such as movement duration, movement distance, movement segmentation or added sinusoids and noise. Based on the findings in combination with the definition of smoothness, their suitability to measure smoothness is discussed.
Results show that different measures can give contradicting outcomes, have a non-monotonic behaviour and are sensitive for noise. Furthermore, only three metrics out of 27 investigated, fully satisfied the definition of smoothness. From these simulations, it can be said that the correlation metric and spectral arc length are most suitable to determine the movement smoothness in reaching movements. This study could help identify standardised movement smoothness metrics for stroke subjects.
 Balasubramanian, S., Melendez-Calderon, A., Roby-Brami, A., & Burdet, E. (2015). On the analysis of movement smoothness. Journal of neuroengineering and rehabilitation, 12(1), 112.
 van Kordelaar, J., van Wegen, E., & Kwakkel, G. (2014). Impact of time on quality of motor control of the paretic upper limb after stroke. Archives of physical medicine and rehabilitation, 95(2), 338-344.
 Buma, F., Kwakkel, G., & Ramsey, N. (2013). Understanding upper limb recovery after stroke. Restorative neurology and neuroscience, 31(6), 707-722.
Realizing Soft High Torque Actuators for Complete Assistance Wearable Robots
Allan Veale, Kyrian Staman, Herman van der Kooij
Abstract: Wearable robots enhance the ability of their wearers to physically interact with the world, and can benefit rehabilitation efficiency, assistive devices' effectiveness, and ergonomic support of workers. Wearable robots' ergonomics and safety can be promoted by using actuators made of soft materials, but soft actuators in the literature are unable to produce the high torques required for lower limb activities of daily living (ADLs), for example, extension of the knee for sit-to-stand (STS).
This paper presents and validates a method for realizing a soft high torque actuator, the pleated pneumatic interference actuator (PIA), for knee extension. Compared to a PIA, the pleated PIA is able to produce a greater torque and a torque when straight. A pleated PIA integrated into pants was shown to produce a knee extension torque that increased with both pressure and angle, as predicted theoretically. This actuator produced a maximum torque of 163 Nm at 4 bar and was able to provide complete knee extension support through 61 % of the range of the STS motion. Future work will improve the pleated PIA’s ability to generate torques at lower bending angles so it can provide complete support through the whole STS motion.
Eye Blinks Related to Freezing of Gait in Parkinson’s Patients
Abstract: Parkinson’s Disease patients sometimes experience a symptom called freezing of gait (FOG). During a FOG episode, patients are temporarily unable to produce effective leg movements. Appearance of FOG might either be triggered or prevented by perceptual input from a person’s environment. Whereas assistive perceptual cues during walking might alleviate cognitive load and decrease FOG prevalence, distractive non-relevant surroundings might increase cognitive load and therefore increase FOG. Differences in cognitive load may thus be a distinctive factor for triggering or preventing FOG. A well-examined correlate for cognitive load is the frequency of eye blinks (Bacher & Smotherman, 2004). Eye blink frequency is known to go up with an increase in cognitive load and potentially blinks could function as a measure of this load. In turn, cognitive load might be a predictive factor for prevalence of FOG
This research is aimed at uncovering the relationship between eye blinks and occurrence of FOG. Specifically, the goal is to investigate deviating patterns preceding FOG, which might help in detecting or predicting a FOG episode before its onset. Parkinson patients (n=15) experiencing FOG were measured during different motor tasks; normal turning, rapid turning and stepping in place. During these tasks an electro-oculogram (EOG) was recorded. The measurements were split into three types of fragments: moments preceding a FOG episode, the FOG episode itself and fragments without FOG. For each fragment, several features of eye movements are extracted such as blinks per minute. These features are compared over conditions to investigate differences in eye blinks related to FOG. Results of this research have high clinical relevance. If any predictive value is present preceding FOG, this could be one step forward in recognizing FOG before it sets in. Moreover, eye blinks yield relatively robust signals in an EOG measurement and can also be easily measured with camera-based systems. Together, predictive values in robust signals might eventually contribute to developing a wearable device that announces FOG and assists patients in daily life.
The Atheroprotective Role of Helical Flow Patterns in Coronary Arteries. an Explanatory Study.
Giuseppe De Nisco, Annette M. Kok, Claudio Chiastra, Diego Gallo, Ayla Hoogendoorn, Francesco Migliavacca, Umberto Morbiducci, Jolanda J. Wentzel
In recent years a distinguishable arterial intravascular blood flow feature, helical flow (HF), has been identified as an atheroprotective hemodynamic phenotype, because of its capability to suppress flow disturbances[1,2,3]. In this study, we (1) describe in depth for the first time helical blood flow in swine coronary arteries, and (2) explore the existence of associations between helical flow descriptors and descriptors of disturbed shear stress.
The lumen geometry of 30 coronary arteries (10 left anterior descending, 10 left circumflex, and 10 right coronary arteries) of adult low density lipoprotein receptors mutation pigs was acquired at the early stage of a high fat diet protocol and reconstructed from fusion of CT angiography and IVUS images. Personalized computational hemodynamics approach was applied to solve the discretized Navier-Stokes equations, with swine-specific flow rates at inflow and outflow sections (at the side branches) obtained from Doppler velocity measurements.
The luminal surface averaged values of three “established” descriptors of low/oscillatory wall shear stress (WSS), i.e. the time-averaged WSS (TAWSS), the oscillatory shear index (OSI), and the relative residence time (RRT), and descriptors quantifying WSS multidirectionality, were computed in each vessel.
Helicity-based hemodynamic descriptors proposed elsewhere were computed to describe helical flow structure in coronary arteries. Possible associations between HF and WSS-based descriptors were explored by Spearman correlation coefficient ρ, significance was assumed for p<0.05.
It was observed that distinguishable counter-rotating HF structures characterize the intravascular hemodynamics of all the investigated coronary arteries, although with different intensity. Significant associations were observed between helicity intensity h2 and WSS-descriptors. In particular, h2 was found to be strongly positively correlated with TAWSS (ρ=0.92), i.e. high helicity intensity values imply higher, more atheroprotective, TAWSS values. A result of interest is that the investigated coronary arteries resulted to be exposed to a moderately-to-scarcely multidirectional WSS.
This study demonstrates that HF is naturally present in healthy coronary arteries and that, as in other arteries[1,3], it has an atheroprotective nature, because of its positive influence on pro-atherogenic hemodynamic WSS-descriptors. Future analysis will clarify if HF in coronary arteries is also correlated to lower prevalence of atherosclerosis.
The Validation of a New Phase-Dependent Gait Stability Measure
Abstract: Falling is a major threat to health of the growing population of older adults. Identification of individuals at high risk of falling is increasingly important. Currently employed stability measures to assess gait stability (1) appear limited in predicting falls in older adults and (2) reduce stability to an average over the gait cycle. Inspired by recent findings of time-dependent changes of local stability within a stride cycle , we present a new phase-dependent stability measure. This measure is closely related to the often-employed local divergence exponent, also referred to as the maximum Lyapunov exponent. Our measure entails linearizing the system in a rotating hypersurface orthogonal to the gait cycle trajectory, and estimating the local divergence rates. The corresponding metric, i.e. the maximum of this local divergence rate over the gait cycle, can serve as predictor of gait robustness. We validated this using a mathematical 2D model of a compass biped walker , which has two configuration parameters, namely slope and mass ratio. For a given set of different configurations of the walker, a stable period-one gait was identified. For comparison we also determined the conventional maximum Lyapunov exponent . To assess robustness, we computed the maximum floor variation the biped walker model may handle by increasing the size of a single step up/down until falls occurred. The new stability measure turned out to be highly correlated with gait robustness, with much better correlation than the maximum Lyapunov exponent. It may provide additional value in predicting falls in older adults from real human walking data.
1. Ihlen, E. A. F., Goihl, T., Wik, P. B., Sletvold, O., Helbostad, J., & Vereijken, B. (2012). Phase-dependent changes in local dynamic stability of human gait. Journal of Biomechanics, 45(13), 2208–2214.
2. Bruijn, S. M., Bregman, D. J. J., Meijer, O. G., Beek, P. J., & van Dieën, J. H. (2012). Maximum Lyapunov exponents as predictors of global gait stability: A modelling approach. Medical Engineering and Physics, 34(4), 428–436.
3. Norris, J. A., Marsh, A. P., Granata, K. P., & Ross, S. D. (2008). Revisiting the stability of 2D passive biped walking: Local behavior. Physica D: Nonlinear Phenomena, 237(23), 3038–3045.
The Impact of Background Phase Offset Errors in Cardiovascular Magnetic Resonance Phase Contrast Imaging in Daily Clinical Practice: a Multi-Scanner Study
Abstract: Background – Phase contrast (PC) CMR flow measurements are widely used for blood flow assessment, but they suffer from phase offset errors (POE). It is crucial to correct for them and the most reliable method is to additionally scan a stationary phantom. An alternative is to use stationary tissue (ST) in the body. The aim of this study was to evaluate the impact of POE in daily practice by comparing the difference of the flow measurements with and without correction using the stationary phantom and the stationary tissue. Furthermore, the interscanner variation was assessed.
Methods – We included 166 patients in which both aorta and main pulmonary artery flow measurements were acquired including a static phantom. Subjects were scanned on three different 1.5T scanners: GE Signa Artist (MRI 1, n=63), GE Discovery MR450 (MRI 2, n=51), and GE Signa Explorer (MRI 3, n=52). Breath-hold PC images were acquired perpendicular to the vessel using fast cine PC pulse sequence with the same parameters on all scanners. Medis Q-flow was used to obtain uncorrected and ST corrected flow measurements, which were compared with phantom corrected measurements and corrected for BSA. A difference of >10% in net flow was defined as clinically relevant. POE influences on regurgitation fraction were assessed by Bland-Altman plots.
Results - Of the 166 cases included, with a median age was 27 (5-74) years. In all scanners the median difference in flow between no correction and phantom correction was ≤3%, however, with a wide range of over- and underestimation. ST correction resulted in larger differences compared to no correction (p<0.01). Clinically significant differences were seen in 19% of all measurements with no correction and in 30% with ST correction (p<0.01). Furthermore, there were significant differences between scanners (no correction, p<0.01; ST correction, p<0.01). Differences in regurgitation fraction with ST correction and no correction are shown (figure).
Conclusion - Background POE have a significant impact on flow quantification in many patients. Unexpectedly, background phase correction using ST correction worsens accuracy compared to no correction. POE vary greatly between scanners.
Kinematic Comparison of the Oxford Foot Model and Rizzoli Foot Model in Healthy Adults
Wouter Schallig, Josien van den Noort, Jennifer McCahill, Mario Maas, Jaap Harlaar, Marjolein van der Krogt
Abstract: Several multi-segment foot models have been developed to measure foot motion during gait1. The Oxford Foot Model2 (OFM) and Rizzoli Foot Model3,4 (RFM) are used most frequently1. The kinematic output of these models has not yet been extensively compared, especially not in atypical gait patterns. Therefore, the aim of this study was to determine the differences in kinematic output between OFM and RFM during normal walking, voluntary equinus and crouch gait.
Ten healthy adults (26.8±2.6 years) underwent 3D gait analysis with the Newington marker model5 placed on both lower extremities and a combined OFM and RFM markerset on the right foot2,3,4. After a static trial, subjects walked six times in normal, equinus, and crouch gait. Markers were captured by a 12-camera Vicon system. Three successful trials were time-normalized to 100% of the gait cycle and averaged. Joint angle waveforms and corresponding range of motion (ROM) values were compared between models using repeated measures ANOVAs and post-hoc analyses with Bonferroni correction. For the joint angles these tests were performed with statistical parametric mapping6.
Here we describe only sagittal plane results of the hindfoot-shank (HF-SH) and the forefoot-hindfoot (FF-HF) angles. For both angles, an interaction effect between gait pattern and model was present. For HF-SH angle, OFM showed more dorsiflexion compared to RFM during the late stance for normal and crouch gait and consequently a larger ROM. For FF-HF angle, RFM showed more plantar flexion compared to OFM towards the end of stance for all gait patterns as well as a higher ROM. For voluntary equinus gait this difference was most prominent, since it was present from 33-63% of the gait cycle.
This study demonstrated relevant differences in kinematic output between OFM and RFM. Interestingly, the differences between the models depend on the gait pattern. It is important to keep this in mind when comparing literature or when applying a specific model to study the gait pattern in a clinical population. The differences between the models are probably a result of different axes definitions and different marker locations with a different sensitivity to skin movement artefacts.
Improvement of Fog Detection in Parkinson’s Disease Patients Via Multimodal Data Analysis
Floris Beuving, Ying Wang, Rick Helmich, Jorik Nonnekes, Mike X Cohen, Richard van Wezel
Abstract: Freezing of gait (FOG) is a symptom of Parkinson’s disease patients. It can be described as a disruption of gait despite the intention to walk. Sensory stimulation with cues offers a remedy to tackle this symptom by helping patients to overcome FOG. However, it has been observed that on-demand cueing is more efficient to decrease the FOG duration compared to continuous cueing. Therefore, this study aims to aid in the development of FOG detection and prediction algorithms.
The experiment consisted of three movement in place tasks (stepping, normal half turning and rapid half turning) with a duration of two minutes for each session. All sessions were recorded on video and rated for the occurrence of FOG by two independent annotators. Fifteen idiopathic Parkinson’s disease patients were measured at off-medication (Hoehn and Yahr scale 2-4). Freezing was observed in all patients. In total 709 FOG episodes were detected with an average duration of 7 seconds. The equipment consisted of a TMSI Porti system (accelerometers, EMG, footswitches and ECG), and a 64 channel 10-10 EEG ActiCap system (Brainproducts). This study focuses on the correlation of body movements with FOG and non-FOG episodes. The uncovered correlations are used for multimodal data analysis to improve FOG detection. The multimodal data analysis allows modalities to complement each other, and is therefore interesting for investigating new perspectives for better FOG detection. The study is in progress and preliminary results will be presented.
Single-Channel Eeg Classification by Multi-Channel Tensor Subspace Learning and Regression
Simon Van Eyndhoven, Martijn Boussé, Borbála Hunyadi, Lieven De Lathauwer, Sabine Van Huffel
Abstract: The classification of brain states using neural recordings such as electroencephalography (EEG) finds applications in both medical and non-medical contexts, such as detecting epileptic seizures or discriminating mental states in brain-computer interfaces, respectively. Although this endeavour is well-established, existing solutions are typically restricted to lab or hospital conditions because they operate on recordings from a set of EEG electrodes that covers the whole head. By contrast, a true breakthrough for these applications would be
the deployment ‘in the real world’, by means of wearable devices that encompass just one (or a few) channels. Such a reduction of the available information inevitably makes the classification task more challenging.
To address this issue, we developed a pipeline for classifying brain states based on data from only a single EEG channel, after a calibration phase in which information of multiple channels is exploited. The pipeline relies on 1) a multilinear subspace learning step, in which spectral filters are tuned that extract frequency bands that are discriminative for the classification task, 2) solving a tensor regression problem with a low-rank structure and 3) the application of an off-the-shelf classifier on the estimated regression coefficients.
We demonstrate the feasibility of this approach for EEG data recorded during a mental arithmetic task. However, it is generic; it may be applicable for the detection of e.g. epileptic seizures as well, as these also induce spectral changes, and for non-EEG data. The proposed framework can readily be improved, by using a few channels instead of only one, and by rigorously tuning the parameters in the pipeline with cross-validation.
 Van Eyndhoven S., Bousse M., Hunyadi B., De Lathauwer L., Van Huffel S., “Single-channel EEG classification by multi-channel tensor subspace learning and regression”, ESAT-STADIUS, KU Leuven (Leuven, Belgium), 2018. Proc. of the 28th IEEE International Workshop on Machine Learning for Signal Processing (MLSP 2018), September, Aalborg, Denmark)
 M. Boussé, G. Goovaerts, et al., “Irregular heartbeat classification using Kronecker Product Equations,” in Proc. of the 39th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC 2017, Jeju Island, South Korea), 2017, pp. 438–441.
High Frame Rate 3D Intracardiac Echography
Mehdi Soozande, Hendrik J. Vos, Johan G. Bosch, Nico de Jong
Abstract: The initial application of IntraCardiac Echography (ICE) is in electrophysiology, to guide and monitor catheter ablations for atrial fibrillation. A 3-D ICE could provide a real-time ultrasound imaging to guide the transseptal puncture and ablation device navigation in a 3-D rendered environment of the left atrium and pulmonary veins. In addition, catheter-based interventions such as closure of atrial septal defects, mitral valve therapies, occlusion of the left atrial appendage, and transcatheter aortic valve replacement also could be performed using 3-D ICE guidance.
Electrophysiologists are interested in determining the site of origin or specific underlying mechanism of atrial arrhythmias, such as macro-reentrant atrial flutter (AFL) and focal atrial tachycardia for RF-ablation treatment. Cardiac electrical activity maps are conventionally used to determine the ablation sites and also evaluate the treatment. This electrical activity causes a transient contraction in myocytes which propagates through the atrial wall. Recent studies demonstrated a high correlation between the cardiac electrical activity and its consequent Electromechanical Wave (EW) in both healthy and arithmetic cases.
Electromechanical waves are transient phenomena and propagate with a velocity of approximately 1m/s. Therefore, a high frame rate imaging (>1000Hz) is required to capture them accurately.
A high frame rate 3-D ICE could offer a 3-D map of electromechanical wave front propagating through an atrial wall. However, realizing a catheter based transducer for such a high frame rate imaging introduces several challenges including a high channel count, low lateral resolution in limited direction, low SNR, and limited power dissipation. In this project, we are trying to address these challenges. A matrix array with a maximum size of 3mm in the limited direction will be mounted on top of an ASIC and encapsulated in a 10Fr intracardiac catheter. A low number of diverging waves should be used to achieve a high volume rate over a large field of view. we are working on approaches to reduce the channel count in transmit and receive, while maintaining a good lateral imaging resolution and SNR. This involves both electronics architecture, transmission / reception schemes and signal processing.
Endobronchial Ultrasound Strain Elastography Imaging of Mediastinal Lymph Nodes; can it Predict Malignancy?
Roel Verhoeven, Chris de Korte, Erik van der Heijden
Abstract: Background – For lung cancer staging, mediastinal lymph nodal spread needs to be assessed in a systematic and complete way. This is routinely done by EndoBronchial UltraSound – TransBronchial Needle Aspiration (EBUS-TBNA), a technique in which a flexible bronchoscope with an ultrasound transducer at the distal tip is inserted into the airways for ultrasound guided needle aspiration of the lymph node. Choosing which mediastinal lymph nodes are suspected and need to be aspirated for staging is currently based on FDG-PET avidity, and, CT- and ultrasound based nodal size. EBUS strain elastography (EBUS-SE) is a relatively new technique that may further help predict malignancy in these mediastinal lymph nodes. However, a standardized assessment strategy for EBUS-SE measurements is lacking.
Objective - To determine an optimal assessment strategy of EBUS-SE measurements and to investigate its ability to predict mediastinal lymph node malignancy.
Methods - Two qualitative visual analogue scale strain scores and two semi-quantitative strain elastography scores (mean strain histogram and strain ratio) were acquired in 120 lymph nodes suspected of lung cancer. After data collection, scorings were randomized into an 80% training dataset to determine scoring cut-off values. Performance was consecutively tested on the remaining 20% and the overall dataset.
Results - Lymph nodes low in strain showed more likely to be malignant. The semi-quantitative strain histogram scoring strategy was able to best and most reproducibly predict malignancy with 93% overall sensitivity, 75% specificity, 69% positive predictive value, 95% negative predictive value, and, 82% accuracy. Combining the EBUS-SE histogram scoring with PET-CT information increased post-test likelihood of disease in relevant clinical scenarios, having a positive test likelihood ratio of 4.16 (2.98-8.13 95% CI) and a negative test likelihood ratio of 0.14 (0.04-2.81 95% CI) in suspicious lymph nodes based on PET- or CT imaging.
Conclusions - EBUS Strain elastography can potentially help predict lymph node malignancy in patients with (suspected) lung cancer. In this pilot study, the best semi-quantitative assessment method showed to be the mean strain histogram technique. Standardization of strain elastography measurement acquisition is required to facilitate widespread clinical use of this technique.
Ultrasound-Guided Breast Biopsy of Ultrasound Occult Lesions Using Multimodality Image Co-Registration
Anton Nikolaev, Hendrik H. G. Hansen, Eleonora Tagliabue, Bogdan Mihai Maris, Vincent Groenhuis, Chris L. de Korte
Abstract: Fusion-based ultrasound(US)-guided biopsy in a breast is challenging due to the high flexibility of the tissue combined with the fact that the breast is usually differently deformed for CT, MR, and US acquisition which makes registration difficult.
This phantom study demonstrates the feasibility of a fusion-based (MRI to 3D US) ultrasound-guided method for breast biopsy. We built a dedicated setup that facilitated 3D ultrasound scanning of the prone-positioned breast while minimally deforming it. In our setup, the phantom was hanging freely inside a cone-shaped water tank. A 152-mm-width ABVS transducer (Siemens Healthcare, Issaquah, WA, USA) was mounted in the cone wall. The cone was able to revolve around the phantom (360° with 1.14° increment) allowing the acquisition of 2D ultrasound data for multiple image planes. Using functional based reconstruction (Rohling, Gee, & Berman, 1999), a volumetric ultrasound image with a resolution of 0.5x0.5x0.5mm resembling the whole breast phantom could be reconstructed within 2 minutes. An aperture array (20 x 14G diameter) was created on the opposite side of the transducer to enable biopsy needle insertion. For lesion localization, the reconstructed US volume was registered to an MRI volume of the phantom acquired using a Siemens Skyra MRI scanner. Based on the lesion location in 3D space, custom-made software rotated the US transducer such that the lesion was in real-time visible in a 2D imaging plane suitable for biopsy and indicated the number of the aperture, and the lateral and axial coordinates on the US image to aid the radiologist in navigating the biopsy needle to the target region for biopsy.
The principle was validated by performing a biopsy of a rigid PVA phantom with 4 lesions (1 cm diameter, 14G needle). Lesions were similar to the phantom in echogenicity but different in color. All 4 biopsied specimen contained the color of the lesion. This demonstrates the applicability of the method towards US occult lesions.
The tissue deformation due to needle insertion was not taken into account yet. The next step is an incorporation of real-time lesion displacement estimation for needle trajectory correction to further improve the accuracy of the biopsies.
Effect of Humeral Elevation Plane on Muscle Activation in People with Facioscapulohumeral Dystrophy
Hans Essers, Anneliek Peters, Kenneth Meijer, Alessio Murgia
Abstract: Facioscapulohumeral Dystrophy (FSHD) is a progressive muscle-wasting disease, which leads to declines in upper extremity functionality. Although the scapulohumeral joint’s stability and functionality are affected, evidence on possible muscle control alterations in FSHD individuals is still lacking. From previous findings it was observed that FSHD persons had a decreased muscle activity of the scapula rotators, Serratus Anterior and Trapezius Ascendens, during humeral elevation in the sagittal and frontal plane. Moreover almost half of the FSHD participants had greater elevations in the sagittal plane than the frontal plane, while no such difference was shown in healthy control participants. The aim of this study is to investigate if the differences in muscle activity, in FSHD participants, are accompanied by a higher elevation in the sagittal than in the frontal plane.
Upper arm kinematics and electromyograms of eight upper extremity muscles were recorded during humeral elevation in the sagittal and frontal plane in ten FSHD participants (Brooke scale 3-4). Maximum normalized muscle activities were extracted from electromyograms and peak humeral elevations were derived from kinematic data. The main outcomes were the differences in maximal muscle activities and humeral elevations, between the sagittal and frontal plane respectively. An independent-samples Mann-Whitney U test was used to test for significant group differences at alpha: 0.05.
Four FSHD participants elevated the humerus higher in the sagittal plane (range: 21-54°, p<0.05) compared to the remaining six who did not show any difference (range: -11-3°). Activations were on average 4.3±2.0 times larger in the sagittal plane for the Serratus Anterior muscle (p<0.05). For the remaining six FSHD participants the same muscle was activated comparably in both planes (1.4±1.0 times in the sagittal compared to the frontal plane). The remaining seven muscles did not show any significant group differences.
Higher humeral elevations were characterized by a higher activity of the scapula rotator Serratus Anterior. This greater activation of the scapular muscle accompanied by an increase in kinematic performance suggests a better preserved muscle quality in an otherwise clinically identical population. Future research should investigate whether this muscle activity difference between humeral elevation planes is underlined by differences in muscle control.
Ex Vivo Detection of Lymph Node Metastasis Using Photoacoustic and Ultrasound Imaging in Head and Neck Cancers
Farzin Ghane Golmohamadi, Khalid Daoudi, Robert Takes, Ilse van Engen-vanGrunsven, Chris de Korte
Abstract: The metastatic status of regional lymph nodes (LNs) is a significant prognostic factor in head and neck cancer patients. Therefore, in these patients, neck dissections are often part of the treatment and the presence of cancer cells is determined through pathological examination. However, this procedure is associated with significant morbidity. More accurate assessment of the metastatic status of the regional lymph nodes would allow for more selective treatment. In this study, we investigated the feasibility of a fast method to identify metastatic lymph nodes ex vivo based on photoacoustic (PA) and high-frequency ultrasound (US) imaging for patients suffering from head and neck cancer. First, we collected freshly dissected LNs from the pathology department, and subsequently, we performed 2D PA and US imaging in random cross-section through the LNs using a PA imaging system with laser wavelength ranging from 670-980nm. Subsequently, we carried out conventional histopathological analysis on the LNs. We select a region of interest (ROI) for each PA and US image and then proceed to quantitative analysis (QA) [1, 2]. The mean and variance were extracted from 2D PA and US images, and next the slope, intercept and midband obtained from the 2D PA spectral analysis was determined. Results showed two distinctive sets of values for the reactive and metastatic LNs (Mann–Whitney U-test, p<0.05) for all QA parameters. Furthermore, we implemented logistic regression to classify LNs as metastatic or reactive for each QA parameter. Our preliminary results reveal that PA and US imaging have potential to facilitate differentiation between reactive and metastatic LNs based on absorption and tissue microstructure. This work shows that ultrasound and photoacoustic spectral imaging can be a promising tool in assisting surgeons to preoperative diagnosis.
 Xu, G., et al. (2014). "The functional pitch of an organ: quantification of tissue texture with photoacoustic spectrum analysis." Radiology 271(1): 248-254.
 Wang, T., et al. (2013). "Characterization of ovarian tissue based on quantitative analysis of photoacoustic microscopy images." Biomedical optics express 4(12): 2763-2768.
3D Geometry Assessment Carotid Artery Bifurcation Using Ultrasound
Joerik de Ruijter, Frans van de Vosse, Marc van Sambeek, Richard Lopata
Abstract: Plaques in carotid arteries are a common cause of stroke. Rupture risk is related to high peak stresses in the shoulders or cap of the plaque. Patient-specific, high-resolution, 3-D biomechanical models are required for a reliable calculation of the magnitude of the peak stresses, which requires an accurate 3-D geometry. In this study, a high-resolution 2-D linear array is used in combination with a magnetic probe tracking device, for 3-D geometry reconstruction of the carotid bifurcation. An automated segmentation and meshing method was developed to create a patient-specific mechanical model for simulation of flow and stress, with minimal user input, and tested in a volunteer study.
The method was tested in vivo on 20 healthy volunteers. By making a slow sweep (6 cm) over the patient’s neck, the full geometry of the bifurcated geometry of the carotid artery is captured. In the first frame, seeds were placed in the centers of the internal and external carotid. The Star-Kalman method was used to approximate the center and the size of the vessel(s) for every frame. Images were filtered with a Gaussian high-pass filter before conversion into the 2-D monogenic signals. Multiscale asymmetry features were extracted from these data, enhancing low lateral wall-lumen contrast. These images, in combination with the initial ellipse contours, were used for an active deformable contour model to segment the vessel lumen. Distension of the wall due to the change in blood pressure is removed using a filter approach. Results were compared to manual segmentation performed by experienced observers. Finally, the contours were converted into a 3-D hexahedral mesh for finite element analysis in ANSYS. Simulations were performed to estimate stress in the wall, and flow in the lumen.
The segmentation algorithm showed good agreement with an average Similarity Index of 0.90 and Hausdorff distance of 0.8±0.4 mm. For the CCA a diameter of 6.8±1.1 mm was found and for ICA/ECA 5.4±0.7 mm, which is comparable with clinical data. The presented algorithm can be applied to generate fluid structure interaction models automatically based on in vivo measurements. In future, these models will be extended to diseased carotid arteries.
3D Ultrasensitive Doppler Imaging Based on Multi-Plane Ultrafast Ultrasound and Eigen-Based Filtering
Chuan Chen, Gijs A.G.M. Hendriks, Stein Fekkes, Hendrik H.G. Hansen, Chris L. de Korte
Abstract: Introduction: Doppler imaging is used routinely for visualizing vascularization and quantifying blood flow. Full-view Doppler is subject to low framerates and has low sensitivity for detecting low velocities. Recent development of ultrafast ultrasound extensively increased the framerate, and 2D studies have shown that in combination with eigen-based filtering it also improves Doppler performance at low blood velocity, a technique called ultrasensitive Doppler. We extended this technique to 3D by mechanical translation of the probe to enable volumetric scanning of low blood velocities, and investigated the impact of the translation speed of the probe on Doppler performance.
Method: We designed a prototype consisting of an L7-4 ultrasound linear array transducer translated orthogonally to the imaging plane (elevational direction) by a stepper motor. While being translated, the transducer transmitted steered plane waves in packages of 9 angles, which data were compounded in receive resulting in 500 high resolution images per second. Acquired signals were decomposed using a typical eigen-based filtering, singular value decomposition (SVD), to separate the slow blood signals from tissue clutter and noise signal. A theoretical model was developed which considers the effects of beam slice thickness, translation speed, and SVD settings. Our theoretical model was experimentally evaluated using data acquired on phantoms, in which water containing ultrasonic scatterers was pumped resulting in flow velocities of 5 and 10 cm/s through two approximately perpendicular 4 mm-diameter tubes. Probe translation speeds were changed from 0.5 cm/s to 1 cm/s, in addition to the stationary 0 cm/s case for reference. The accuracy of the obtained Power Doppler images was quantified in terms of Contrast-to-Noise Ratio (CNR) between delineated flow regions and background regions.
Results: At a flow velocity of 5 cm/s, the measured CNRs were 2.32, 2.01 and 1.87 for translation speeds of 0, 0.5 and 1 cm/s, respectively. At 10 cm/s flow velocity, corresponding values were 2.84, 2.54 and 2.15.
Discussion: We developed 3D ultrasensitive Doppler method based on multi-plane ultrafast ultrasound and eigen-based filtering. More accurate Power Doppler was produced for slower translation speeds and for higher blood velocities. This was theoretically expected because smaller Doppler phase shifts need to be captured by longer acquisition periods.
Repeatability of Natural Shear Wave Elastography Measurements
L.B.H. Keijzer, M. Strachinaru, D.J. Bowen, M.D. Verweij, A.F.W. van der Steen, J.G. Bosch, N. de Jong, H.J. Vos
Abstract: Diastolic dysfunction is an important cause of heart failure with a preserved ejection fraction, where a decreased active relaxation and/or increased passive stiffness prevent the heart to completely relax . Currently, there is no accurate method for non-invasive stiffness measurements of the myocardium. However, early diagnosis is important for preventing further development of heart diseases and could likely help in accommodating more personalized treatment. Shear waves (SWs) can potentially be used to perform noninvasive stiffness measurements , called shear wave elastography (SWE). The propagation speed of SWs is linked to the stiffness of the medium in which the SWs propagate. This study focusses on SWs naturally induced by the aortic and mitral valve closure (AVC and MVC). For clinical diagnosis, high precision of the SWE measurements is important and therefore we tested the reproducibility in 10 healthy volunteers. Inter-system variability was researched by comparing a clinical scanner in conventional TDI mode (Philips; 490-570 Hz frame rate) with a second clinical system with a high frame rate mode using a diverging-wave pulse-inversion transmission scheme (Zonare; 1000 Hz). Furthermore, test-retest, inter-scan, intra-scan and inter-observer variability was investigated. For each volunteer, multiple measurements were performed with both systems within 30 minutes. Furthermore, these measurements were repeated 21-93 days later. Moreover, the effect of stress on the SWE measurements was tested by performing handgrip tests.
With the Zonare system, propagation speeds were found to be in the range of 3.2-4.3 m/s (AVC) and 2.1-4.7 m/s (MVC). During the handgrip test, heartrates increased, but the propagation speeds did not significantly differ from the rest measurements. Test-retest variabilities were found to be in the same range as the inter-scan variabilities. The scanning-view quality, the limited length of the interventricular septum (IVS) and the variations in angle between the IVS and the probe are potential causes. With the Philips system, statistically different propagation speeds were obtained (AVC: 1.8-4.8 m/s; MVC: 3.7-6.4 m/s). These differences are likely caused by differences in methods, such as different frame rates and the methods used to determine propagation speeds.
 M. R. Zile, et al., N. Engl. J. Med., 2004;  O. Villemain et al., JACC Cardiovasc. Imaging, 2018.
The Design of a Novel Activity Tracker for the Hand: Improving Daily-Life Hand Functioning after Stroke by Objective Feedback
Ruben Regterschot, Gerard Ribbers, Agaath Sluijter, Ruud Selles, Hans Bussmann
Abstract: People that have suffered from a stroke frequently have an impaired function of the affected arm in daily life. Studies show that high intensity of practice and high levels of arm-hand usage in daily life conditions contributes to better functioning of the more affected arm. Therefore, the aim of our project is to develop a hand activity tracker that applies objective feedback to stimulate the use of the more affected arm in stroke patients during daily life conditions. The current design of the hand activity tracker consists of two wrist-worn accelerometers and an accelerometer on the leg. The hand activity tracker will combine arm activity data with body postures and movements to estimate arm-hand use. Furthermore, it will apply reminders and real-time goal achievement to stimulate arm-hand use. In the reminders strategy, when the more affected arm is not used enough then the device will send reminders on an hourly basis to remind the patient to use the more affected arm. The reminder consists of a visual instruction combined with a vibro-tactile trigger to draw the patient’s attention to the display of the device. After a succesful reminder a visual reward is given in combination with a vibro-tactile trigger. In the real-time goal achievement strategy, the device will display the current status compared to the daily arm-hand use goal. When the goal is reached the device will give a visual reward combined with a vibro-tactile trigger to attract the patient’s attention to the display. We evaluated the design of the hand activity tracker on acceptability and feasibility by performing interviews in six stroke patients (5 males, 1 female; age: 58.2±18.2 yrs; 1 month - 5 years post stroke). The results of the interviews indicate the acceptability and feasibility of the design of the hand activity tracker. As next steps we will develop a prototype of the hand activity tracker and evaluate it on usability during daily life conditions in stroke patients.
Explorative Study on Using Classification of Electrical Appliances in Monitoring Systems: a Comparison of Classifiers
Marc Mertens, Glen Debard, Bart Vanrumste, Jesse Davis
Abstract: Since the population is ageing, more elderly people will depend on the support of others. Automatic monitoring systems that can provide information regarding the status of elderly living alone, such as a gradual or sudden change of behavior can aid the caregiver in assessing the health status and self-reliance. Several sensor types can be used in such a monitoring system. In our study, we monitor the usage patterns of electrical appliances. By linking usage of appliances to Activities of Daily Living (ADL), behavioral patterns can be constructed. The first step in this approach is the recognition of used appliances. This abstract describes a comparison of classifiers which can be used to discern between different appliances. It builds on an earlier exploration using the J48 tree classifier as described in .
The dataset consists of electrical current profiles of several appliance types (e.g., TV sets, coffee machines, vacuum cleaners, electrical heaters, etc.), sampled at 5kS/s. From this data, several features were extracted such as Irms, Imax, correlation of the electrical current with a perfect 50Hz sinewave, etc. A total of 109 instances of one period were collected. The data set is split up randomly into a training set of 85 and a test set of 24 instances. We compared four classification algorithms on accuracy; the above mentioned J48, Naive Bayes, Nearest Neighbors (kNN, k=1..3) and Logistic Regression.
Comparing accuracy results, it is shown that kNN (n=1) gives the best result (95.8%) and in order of decreasing accuracy: J48 (90%), kNN (n=2 or 3) (87.5%), Naïve Bayes (79%) and finally Logistic Regression (75%). kNN (n=1) giving the best result, indicates that a near perfect prediction can be done if we first measure the relevant appliances in the house and use those as a training set for detection.
With this experiment, we showed it is possible to recognize electrical appliances from their electrical current signature in an accurate way. For future work, we will analyze appliances running at the same time. We acquired four datasets in real living environments and will use this classification method to construct ADL patterns based on appliance usage.
 MONITORING ACTIVITIES OF DAILY LIVING THROUGH DETECTION OF USED ELECTRICAL APPLIANCES Marc Mertens, Glen Debard, Bert Bonroy, Els Devriendt, Koen Milisen, Jos Tournoy, Jesse Davis, Tom Croonenborghs, Bart Vanrumste, BME2013
A Comparison of Three Models for Gait Analysis in Children with Cerebral Palsy
Eline Flux, Jaap Harlaar, Marjolein van der Krogt
Abstract: Multiple models exist to assess gait kinematics, with different underlying marker sets and biomechanical models. This study assessed differences between two conventional and one more recently developed model for gait analysis in children with cerebral palsy. Furthermore, effects of different measurement sites were analyzed and reliability in knee axis alignment was determined. Twenty-five children with cerebral palsy participated. 3D instrumented gait analysis was performed in three laboratories across Europe, using a comprehensive retroreflective marker set including three models: the Newington Model , known as Plug-in-Gait (PiG), the calibrated anatomical system technique (CAST) and the Human Body Model (HBM). Kinematic outcomes from the models were compared using statistical parametric mapping (SPM) and RMSE values were used to quantify differences. Sagittal plane outcomes were comparable (discrepancies between models were less than 5°) for most comparisons, whereas for frontal and transversal planes, differences for almost all segment and joint angles exceeded the threshold for possible influence on clinical decision-making (i.e. >5°). Furthermore, an interaction effect between model and measurement site was found, which suggests a possible influence of specific marker placement experience of researchers, although this should be further studies in future researches, for example by analyzing the same patients in multiple laboratories. CAST and PiG knee adduction angles showed low reliability, with data collected using an online optimization method showing higher reliability, while HBM constrains frontal and transversal knee movements. It can be concluded that studies using different models cannot be compared, especially with respect to the frontal and transverse plane kinematics. A translation algorithm needs to be developed for comparing output from different marker models, taking into account the offsets, relative errors and error band.
 Davis, Hum Mov Sci 10 (1991) 575-587
 Cappozzo, Clin Biomech 10 (1995) 171-178
 Van den Bogert, Med Biol Eng Comput 51 (2013) 1069-1077