Go-previous
13:00   Neurophysiology
A multimodal test-battery for clinical assessment of visual attention in neurological disorders
Valentina Barone, Johannes van Dijk, Mariette Debeij-van Halld, Michel van Putten
Abstract: Attention is an important aspect of human brain function. Patients with neurological disorders, such as epilepsy and traumatic brain injury, can experience complaints related to attention and connected cognitive abilities. Therefore, assessment of attention with quantifiable parameters may assist in patient care, both for diagnostics and prognostication. To quantify visual attention in the clinic, we developed a compact test-battery combining a choice reaction time task, eye-tracking and EEG. Within a single and rapid (15 min) experimental design, the system assesses reaction time, parameters of eye movements (i.e. saccade metrics and fixations) and event related potentials (ERPs). We present pilot data from controls, patients with mild traumatic brain injury and epilepsy. Reaction times and eye metrics such as fixation duration, saccade duration and latency show significant differences between neurological patients and controls. Late ERP components are detected in all subjects, but no significant group differences could be found in the peak latencies and mean amplitudes. Our test-battery can measure key features of visual attention in the clinic. Significant differences between controls and patients are found for reaction times and eye metrics. These findings illustrate that our system is potentially beneficial for diagnostics and prognostication of neurological patients.
Altered nociceptive function in patients with morbid obesity compared to healthy controls
Tom Berfelo, Imre Poldino Krabbenbos, Boudewijn van den Berg, Eva Kleinveld, Jan Reinoud Buitenweg
Abstract: For improved observation of nociceptive dysfunction, we are developing a novel measurement technique that combines the nociceptive detection threshold (NDT) with brain evoked potentials (EP) using intra-epidermal electrical stimulation. To validate the applicability of the NDT-EP method in patients with higher body mass index (BMI) values, we explored the feasibility of the NDT-EP method in pain-free patients with morbid obesity (MO). Subsequently, we compared the NDT-EP outcomes in MO with those of healthy controls (HC) at the various stimulus types used. Seventeen pain-free MO patients (BMI: 45.9 ± 4.6) and sixteen HC (BMI: 22.0 ± 2.0) were measured at the St. Antonius hospital. Three stimulus types (i.e., single- and double-pulse stimuli with 10 and 40 ms inter-pulse interval) consisting of a total of 450 trials were applied to each subject during two measurements. Subsequently, NDT-EP outcomes related to stimulus properties were calculated using (generalized) linear mixed regression models. The NDT-EP method was completed successfully in all MO patients. NDT results demonstrated that the detection probability was significantly (p=0.014) associated with the effect of MO diagnosis. The detection probability was significantly (p=0.020) decreased by the interaction of diagnosis with the amplitude of the first pulse. Furthermore, EP results showed that the positive peak of the EP amplitude at 485 ms post-stimulus was significantly (p=0.031) decreased by the interaction of diagnosis with the amplitude of a second pulse after 10 ms. The NDT-EP method was feasible to use in pain-free MO patients. Different NDT-EP outcomes were seen in MO compared to HC, which may indicate altered nociceptive function. Therefore, we need future research into clinical features, e.g., BMI, related to nociceptive dysfunction.
Transfer learning in Deep Neural Networks for epileptic seizure detection with wearable EEG
Miguel Bhagubai, Maarten De Vos
Abstract: Epilepsy patients’ monitoring is currently done via visual inspection of video-electroencephalography (video-EEG) by neurologists. However, it is important to monitor patients remotely on their daily lives. Currently, patients recur to self-monitoring by annotating seizure occurrences in diaries. These diaries are often very inaccurate, making them unreliable. Automatic and portable EEG-based systems, developed to monitor epilepsy patients and automatically detect epileptic seizures, are very important and useful tools to support diagnosis and keep track of the progression of the disease. The typical full scalp-EEG set-ups that are used in the hospital are too cumbersome to wear during daily life. The development of automated seizure detection algorithms, based on behind-the-ear EEG (bhe-EEG) data measured with minimally intrusive and wearable devices, allows monitoring patients 24/7 outside the hospital in a more reliable way. Most state-of-the-art algorithms use full-scalp EEG data. This work investigated Deep Neural Networks (DNNs) using bhe-EEG data from epileptic patients. Wearable EEG contains less information since it only includes a reduced number of channels and the location of the electrodes is further away from the seizure source. DNNs require large amounts of data to achieve acceptable performance. In this work, we showed the difference in performance of a deep convolutional neural network (DCNN) trained and evaluated on the different EEG modalities. This work also investigated the use of transfer learning in order to improve the classification on the bhe-EEG by taking advantage of the model trained on full-scalp EEG and tunning with the bhe-EEG data. The results showed that the performance of a bhe-EEG based network is lower than a model trained and tested on full-scalp EEG (52% sensitivity with 2.5 false alarms per hour and 76% sensitivity with 4.58 false alarms/hour respectively). The weight transfer approach proved to be effective in developing a bhe-EEG based model, increasing the sensitivity to 82% with a 1.71 false alarm rate per hour. This work presented alternative methods for automatic wearable EEG classification, however, the limited amount of data still dampens the performance of DNNs when compared to traditional feature-based algorithms. Further work is needed for implementing such frameworks in practice.
Fit for purpose of on-the-road and simulated driving effect of sleep deprivation
Ingrid Koopmans, Robert-Jan Doll, Marieke de Kam, Geert Jan Groeneveld, Adam Cohen, Rob Zuiker
Abstract: Drivers should be aware of possible impairing effects of alcohol, medicinal substance, or fatigue on their driving performance. Such effects can be assessed in clinical intervention trials including a driving task. However, a choice between simulated and on-the-road driving must be made. Here, we assess the sensitivity of simulated and on-the-road driving variables to sleep-deprivation induced sedation in healthy experienced drivers. This two-way cross over study included 24 healthy males (age 25.7+/-1.6) in possession of a valid driver’s license with a minimum driving experience of 3000 km per year. Simulated and on-the-road driving behaviour were assessed in the morning, after a well-rested night and after a sleep-deprived night (with at least 7 days to recover from the sleep deprivation). All subjects started with simulated driving followed by the on-the-road driving test. The driving simulator (Green Dino BV, the Netherlands) was a fixed base simulator with three pedals, a steering wheel with indicators, and three LCD screens. The on-the-road car (modified VW Caddy Combi 2017) used a forward-facing camera (Mobileye Vision Technologies, Ltd, Israel) to measure the relative road position. Driving behaviour was examined by calculating the Standard Deviation of Lateral Position (SDLP). Additionally, subjective scores for perceived driving performance and effort were collected using Visual Analogue Scales (VAS). Sleep deprivation significantly increased the SDLP during both simulated (10 cm [33%], 95%CI: 6.7-13.3) and on-the-road driving (2.8 cm [13%], 95%CI: 1.9-3.7). Perceived driving performance decreased after a night of sleep deprivation with 3.8 (45%, 95%CI: 5.0-2.6) for the simulator and 3.5 (38%, 95%CI: 4.8-2.2) for the on-the-road test. Driving effort increased after a night of sleep deprivation for both the simulator and the on-the-road test with 6.9 (202%, 95%CI: 5.7-8.1) and 5.9 (178%, 95%CI: 4.4-7.4), respectively. The significant correlation between the on-the-road test and the simulator for the well-rested morning (0.63, p < .001) was no longer present after a night of sleep deprivation (0.31, p=.18). The minimal detectable effect size in a cross-over study design (n=16) based on this study is 5.9 cm for the simulator and 1.5 cm for the on-the-road test. The difference in sensitivity to the effects of sleep deprivation on simulator compared to on-the-road driving are an indication that the tasks are not interchangeable and that they may assess different aspects of driving behaviour.
Comparison of paired-click paradigms to study auditory sensory gating in healthy human subjects
Annika de Goede, Liam van der Aa, Robert Doll
Abstract: Sensory gating refers to the neuronal process of filtering out irrelevant or redundant stimuli to protect the brain from an information overload. It is commonly tested with the paired-click paradigm by measuring event-related potential (ERP) components using electroencephalo-graphy. In healthy subjects, the P50, N100, and P200 components are largely inhibited after the second stimulus, while gating deficiencies are described for various neuropsychiatric conditions and drugs. Despite extensive use of the paired-click paradigm, there is a lack of consensus for several methodological settings. This study aims to evaluate test parameters in terms of 1) maximizing gating and 2) minimizing test duration. Ten healthy subjects participated in five configurations of the paired-click paradigm to study the effect of inter-pair interval (IPI) duration (3 s versus 8 s), IPI variability (fixed versus ± 0.5 s), and stimulus type (click versus tone). Each configuration consisted of 150 pairs of either clicks (5 ms square wave at 85 dB) or tones (100 ms 1 kHz tone at 85 dB), with an interstimulus interval of 500 ms. Conditions were randomized and separated by a 5-minute break. The difference between P50, N100, and P200 amplitude to the first and second stimulus (S1-S2) were analysed by using linear mixed models. Each ERP component showed stronger gating for the 8 s IPI compared to the 3 s IPI (p < .001). No significant effect of IPI variability was found, nor an interaction between IPI duration and variability. Clicks induced stronger gating compared to tones for the P50 component (p < .001), whereas N100 and P200 gating was stronger when tones were applied (p < .01 and p < .05, respectively). Sensory gating was most reliably measured using an 8 s IPI, regardless of IPI being fixed or variable. The fact that gating decreased for the 3 s IPI limits the possibility to shorten the paired-click test duration, which would make it more suitable for use in early phase drug development. Since the components are differently affected by clicks and tones, decisions regarding the stimulus type should be based on the component of interest.
The impact of electrode interaction on the auditory processing of CI users
Elisabeth Noordanus, John van Opstal
Abstract: Cochlear implant (CI) users display a large variability in speech recognition in silence and – even more – in noise. Possible sources of this variability are multifold. These can be device-related, for example, 1) differences in the precise locations of CI electrodes in the cochlea, and 2) interference between electrodes because of overlapping auditory nerve populations, together with 3) suboptimal individual settings of the sound processor. Alternatively, or in addition, the cause of variability may be subject-related, for example 4) degeneration of the auditory nerve, 5) deficiencies in the more central auditory processing, or 6) cognitive factors. Currently, no methods are available to disentangle the influence of these six sources of variability on speech recognition. We investigate two new objective methods to determine the spectro-temporal resolution on electrode-level and take, in a controlled way, the impact of CI electrode interactions on the processing of temporal and spectral information into account, while excluding influences of the sound processor (point 3) and largely excluding cognitive influences (point 6). Both objective methods use direct stimulation of the CI electrodes, bypassing the sound processor. In our first experiments we stimulated two electrodes, each with a different modulation frequency. One method (electrophysiology) analyses the EEG response to this known input, without any active involvement of the subject, the other (psychophysics) measures the subject’s reaction time to a perceived change in the modulation frequency of one of the electrodes or to a change in stimulated electrode. We tested both methods with 20 CI users. In addition, we used existing measurement methods to characterize the device (cross-impedance) and the response of the auditory nerve (spread of excitation). We also determined the spectro-temporal performance by including the clinical sound processor, and assessed speech-recognition scores. Preliminary analysis indicates that both objective methods give relevant results. The interpretation of the reaction-time results is more straightforward, and its implementation for clinical use is easier, than that of the EEG based method. Further analysis will relate the outcomes of our methods to those of the device, the auditory nerve, and to speech-understanding performance.


end %-->