5th Dutch Bio-Medical Engineering Conference 2015
22-23 January 2015, Egmond aan Zee, The Netherlands
13:30   Motor Control, Neuro Control & Patient Models II
15 mins
Dirk Weenk, Fokke van Meulen, Bert-Jan van Beijnum, Ed Droog, Daniel Roetenberg, Hermie Hermens, Peter Veltink
Abstract: Introduction and past research: Inertial sensors are great for orientation estimation, but they cannot measure relative positions of human body segments directly. In previous work we used ultrasound to estimate distances between body segments [1]. In [2] we presented an easy to use system for gait analysis in clinical practice but also in-home situations. Ultrasound range estimates were fused with data from foot-mounted inertial sensors, using an extended Kalman filter, for 3D (relative) position and orientation estimation of the feet. Validation: From estimated 3D positions we calculated step lengths and stride widths and compared this to an optical reference system for validation. Mean (±standard deviation) of absolute differences was 1.7 cm (±1.8 cm) for step lengths and 1.2 cm (±1.2 cm) for stride widths when comparing 54 walking trials of three healthy subjects. Clinical application: Next, the system presented in [2] was used in the INTERACTION project, for measuring eight stroke subjects during a 10 m walk test [3]. Step lengths, stride widths and stance and swing times were compared with the Berg balance scale score. The first results showed a correlation between step lengths and Berg balance scale scores. To draw real conclusions, more patients and also different activities will be investigated next. Future work: In future work we will extend the system with inertial sensors on the upperand lower legs and the pelvis, to be able to calculate a closed loop and improve the estimation of joint angles compared with systems containing only inertial sensors.
15 mins
Lucas Dobbe, Paulina Bank, Jurriaan de Groot, Erwin de Vlugt
Abstract: A highly prevalent disease like stroke may affect key organ systems involved in motor performance, e.g., the central nervous system, the musculoskeletal system, and sensory systems involved in vision and proprioception. This compromises the ability to properly adapt to fluctuating environmental conditions and task demands [1] by means of feed forward and feedback control mechanisms. Loss of sensory (re)weighting between visual and proprioceptive/haptic feedback may play an essential role in this compromised motor adaptability, but is yet unknown. Emerging technologies such as virtual reality and haptic robots provide new opportunities to manipulate environmental conditions in order to identify the contributors to movement disorders and guide selection of the optimal strategy for treatment. For meaningful use of virtual reality in a clinical setting, however, it is essential to understand the extent and manner in which visual information can be exploited to assess the adaptability of motor control. To this end, twenty healthy subjects performed a visuomotor tracking task within a virtual environment that was presented on a 60” LED screen [2]. With a virtual car, subjects tracked a curved road (f < 1 Hz) by means of flexion and extension movements (≈ 10°) of the wrist in a haptic robot (Wristalyzer, MOOG FCS). Additionally small torque perturbations were applied to the wrist (1.25-20 Hz multisine) for system identification [3] in order to quantify changes in neuromechanical properties of the wrist (i.e., joint stiffness, joint damping, and reflexive and visual feedback) due to specific manipulations of the visual scenery related to task demands (i.e., road width, velocity, and preview of the upcoming trajectory) and presentation of visual information (i.e., the zoom factor and the presence or absence of a virtual tunnel). As expected, healthy subjects adapted their control strategy to changes in the visual scenery. Manipulations of road width, velocity and preview led to changes in joint stiffness and damping, which probably reflects prominent co-contraction and voluntary steering behaviour. The contribution of visual and proprioceptive feedback depended on movement velocity. Manipulations of the visual scenery that were related to visual presentation only (i.e., without effect on task demands) had less pronounced effects on motor behaviour. This study showed that visual information induced changes in neuromechanical properties underlying motor behaviour. The ability to manipulate and evaluate both the visual and the proprioceptive sensory feedback systems provides opportunities for high resolution quantification of (the lack of) motor adaptability in patients, for the purpose of diagnosis as well as tailored training. REFERENCES [1] C.G.M. Meskers, A.C. Schouten, J.H. de Groot, E. de Vlugt, J.J. van Hilten, F.C.T. van der Helm, and J.H. Arendzen, “Muscle weakness and lack of reflex gain adaptation predominate during post-stroke posture control of the wrist”, J. NeuroEng. Rehab., 6:29 doi:10.1186/1743-0003-6-29 [2] T. Geijtenbeek, F. Steenbrink, E. Otten, and O. Even-Zohar, “D-flow: immersive virtual reality and real-time feedback for rehabilitation”, In: Proc. of the 10th ACM international conference on virtual reality continuum and its applications in industry, pp. 201-208, (2011). [3] A.C. Schouten, W. Mugge, and F.C.T. van der Helm, “NMClab, a model to assess the contributions of muscle visco-elasticity and afferent feedback to joint dynamics”, J. Biomech., Vol. 41, pp. 1659-1667, (2008). ACKNOWLEDGMENT This work is part of the research programme IMDI Neurocontrol (Neuras project) financed by the The Netherlands Organisation for Health Research and Development
15 mins
Tricia Gibo, David Abbink
Abstract: Haptic technology has the potential to be used for the training of a variety of skills, such as teleoperation, vehicle control, and movement training. Previous studies have investigated the use of haptic guidance to improve the training of motor skills, i.e., providing assistive forces to help the trainee learn the correct movement needed to accomplish a task. Unfortunately, most studies have shown a null or even detrimental effect of haptic guidance on learning the desired movement trajectory [1,2]. For more complex motor tasks, it may be possible that the task can be accomplished via different movements. Here, motor skill learning consists of: 1) identifying a movement strategy to use, and 2) refining execution of this movement strategy. Prior work on haptic guidance for training has mainly focused on improving the latter. While the former has not been directly studied, results from a few prior studies suggest the potential of using haptic guidance to explore the motor workspace, thus assisting the search for a better strategy. Gillespie et al. showed that haptic guidance did not improve subjects’ performance as they learned to control a pendulum-cart system; however, subjects who trained with guidance were more likely to attempt the demonstrated optimal control strategy [1]. Additionally, other studies have shown that subjects adopt different coordination strategies when other forms of feedback are provided via the haptic sensory modality [3]. In this study, we conducted an experiment to investigate whether haptic guidance can be used to help people discover new movement strategies to control a complex dynamical system. Subjects used a 1-degree-of-freedom haptic manipulator to control a cursor attached to a virtual mass-spring system. The task consisted of moving both the cursor and simulated mass to a target as fast as possible. Subjects completed 150 trials in one of the following two conditions. In the No Guidance (NG) condition, subjects did not feel any forces from the manipulator. In the Haptic Guidance (HG) condition, guiding forces were applied to the subject’s hand to help them accomplish the task, guiding them along different velocity profiles and damping any resulting oscillations of the mass-spring system. Guidance forces were only presented on 50 trials interspersed throughout the training. Haptic feedback of the mass-spring system dynamics was not present in either condition. Final performance was compared between the groups on trials when no haptic guidance was present. We found that the two groups eventually settled on different movement strategies to perform the task. HG subjects adopted movement trajectories with higher velocities and allowed greater separation between the cursor and virtual mass (similar to the strategy used when haptic feedback of system dynamics is present). Alternatively, subjects in the NG condition moved slower, keeping the mass near the cursor. These results suggest that haptic guidance can aid in the explicit identification of a different movement control strategy during learning.
15 mins
Huan Yang, Hil Meijer, Robert-Jan Doll, Jan Buitenweg, Stephan van Gils
Abstract: Capsaicin is the pungent substance in chili peppers. As a tool in pain research and therapy, capsaicin can induce distinct effects in the human nociceptive system, i.e. peripheral degeneration and central sensitization. Using a previously developed computational model of the nociceptive pathway [1], we aim to identify these effects as perturbations of model parameters. We consider data from a topical application of a high dose capsaicin patch, where nociceptive detection thresholds and epidermal nerve densities were measured during 3 months [2]. We modeled the variation of parameters during the experiment to account for the degeneration and regrowth of nerve endings as well as temporarily enhanced synaptic efficiency and membrane excitability of dorsal horn neurons. We compared the simulated detection thresholds to the experimentally observed phenomena. Our model simulation captures the changes in thresholds as experimentally observed. Our study validates the model for capsaicin-perturbed nociceptive function and also reveals the capsaicin-induced effects on nociceptive function. First, our study suggests that the increase of detection thresholds after the application of capsaicin is mainly due to the degeneration of nerve endings. Second, our model simulation suggests that peripheral and central neuroplasticity can compensate each other, resulting in different patterns of detection thresholds for single or multiple pulse stimuli. Third, our simulation implies the existence of long-lasting transcription-dependent central neuroplasticity. Future work could estimate parameters to quantify these opposite effects in individual subjects. REFERENCE [1] Yang, H., Meijer, H.G.E., Doll, R.J., Buitenweg, J.R. and van Gils, S.A. Computational modeling of nociceptive stimulus detection. Submitted. [2] Doll, R.J., Buitenweg, J.R., van Amerongen, G., Hay, J.L., Groeneveld, G.J. and Veltink, P.H. Capsaicin (8%) patch increases multiple electrical nociceptive thresholds in healthy human subjects. In the 8th Congress of European Federation of IASP Chapters (EFIC), 9-12 Oct 2013, Florence, Italy. pp. 756-756.
15 mins
Rick van der Vliet, Maarten Frens, Opher Donchin, Ruud Selles
Abstract: Over the last decade, transcranial Direct Current Stimulation (tDCS) over the motor cortex (M1) has been discovered as a neuromodulation technique that can increase neuronal excitability (Nitsche et al., 2003) and LTP (Fritsch et al., 2010), skill learning (Lefebvre et al., 2012; Reis et al., 2009; Reis et al., 2013; Waters-Metenier, Husain, Wiestler, & Diedrichsen, 2014) and rehabilitation after stroke (Khedr et al., 2013). Effort is currently directed towards optimizing current flow through the cortex (Dmochowski, Datta, Bikson, Su, & Parra, 2011; Huang et al., 2013) and comparing the clinical efficacy of different non-invasive brain stimulation protocols. The results of these studies should establish the feasibility and potential of M1 tDCS as a clinical and scientific tool. tDCS over the cerebellum might be similarly useful in accelerating adaptation of learned motor skills such as reaching and walking and indeed, several pilot studies have shown beneficial effects of stimulation on rotational (Galea, Vazquez, Pasricha, de Xivry, & Celnik, 2011) and locomotor adapation (Jayaram et al., 2012). However, development of cerebellar tDCS is seriously hampered by the intricacies of cerebellar anatomy and the lack of mechanistic understanding of both tDCS and cerebellar learning. Different areas of the cerebellum are involved in distinct forms of adaptation and it is not clear how these areas should be targeted. That is, the range of electric field strengths affecting neuronal firing and eventually behaviour is unknown. Ideally, several lines of research would be designed to (1) establish the range of electrical field strengths influencing neuronal excitability and behaviour in mice and slices, (2) model current flow in humans and monkeys and relate behavioural modulation to electric field strength, and (3) design targeted tDCS protocols that can increase electric field strength in an area of interest to definitively test specific hypotheses on electric field strength, cerebellar anatomy and behaviour. Here, we would like to present preliminary data on behavioural modulation and current flow modelling in humans. We have found that anodal transcranial stimulation over the cerebellum increases the rate of reaching adaptation (in line with Galea et al. (2011)), saccadic adaptation and eyeblink conditioning (in line with Zuchowski et al. (2014)), but not vestibulo-ocular reflex (VOR) adaptation (not in line with results from animal experiments (Das et al., 2014). In line with these findings, preliminary modelling results suggest electric field strength in the flocculus (involved in VOR adaptation) is several orders lower than in the areas involved in eyeblink conditioning, reaching and saccadic adaptation. Possibly, this explains why VOR adaptation is not modulated by tDCS using the standard electrode configuration. Combining behavioural with modeling data could therefore prove to be an interesting approach towards increased understanding of cerebellar tDCS.
15 mins
Mark Vlutters, Tjitske Boonstra, Alfred Schouten, Herman van der Kooij
Abstract: Background and Aim To ensure upright balance the ankle joint stiffness must be sufficient to resist the gravitational pull. This stiffness arises from both intrinsic and reflexive components. Determining their individual contri-bution might give insight in neuromuscular and balance related disorders. Ankle joint stiffness is often investigated by fitting a parametric model to a torque response, obtained from an applied joint rotation. Direct, non-parametric estimation is often not applicable because the applied rotations cannot rule out reflex activity. Here the rotational amplitude dependency of the intrinsic ankle stiffness was estimated in standing, using fast ramp-and-hold stretches to circumvent reflexive contributions. Methods Eight healthy subjects participated in the study. Subjects stood on the Bilateral Ankle Perturbator (BAP, figure top), with which 0.08-0.005 rad plantar- and dorsiflexion rotations were applied to the individual ankle joints. Rotations consisted of 40 ms ramp-and-hold minimum jerk profiles [1]. The intrinsic stiffness was obtained by dividing the difference in torque exerted on the platform before and after rotation onset by the rotational amplitude. These values were normalized to the critical stiffness [2]. EMG data of the triceps surae and tibialis anterior muscles were recorded to investigate reflex activity. Results The EMG signals of the streched muscles showed short latency reflex activity starting approximately 5 ms after the rotation ended (figure middle). The EMG of the gastrocnemius medialis is shown. The intrinsic ankle stiffness decreased non-linearly with increasing rotation amplitude. There was no significant difference in stiffness between plantar- and dorsiflexions. A fit to all subjects' pooled data in comparison with other values in literature [2-4] is shown (figure bottom). Conclusions The intrinsic ankle stiffness is insufficient to ensure balance, hence changes in muscle activation are required to realize upright stance. Reflex activity is not expected to have influenced the stiffness estimates due to the short latency of the perturbations and the electro-mechanical delay of muscle tissue. The decrease in stiffness is attributed to muscle cross-bridge breakage leading to sliding of filaments, decreasing the overall stiffness. References [1] Burdet e.a.–2000–J.Biomech. [2] Casadio e.a.–2005–Gait Posture [3] Loram e.a.–2002–J.Physiol. [4] Loram e.a.–2007–J.Physiol.