Talk Title: Augmented Reality Based and EEG-guided Neglect Detection, and Assessment
Abstract: From binary detection to scalable, connectivity-informed, and personalized assessment
Spatial neglect (SN) is a common visuospatial attention disorder following stroke, affecting roughly 30% of survivors and strongly predicting long-term disability. The clinical gold standard—the pen-and-paper Behavioral Inattention Test—yields only a binary pass/fail outcome, is confounded by compensatory movements, and cannot map the extent of the neglected visual field. To address these gaps, we developed AREEN, an augmented-reality, EEG-guided system that presents controlled visual stimuli through a HoloLens headset while recording synchronized 16-channel EEG.
Building on AREEN, we present four advances evaluated in a cohort of 28 stroke patients (16 with
neglect, 12 without). First, using leave-one-participant-out cross-validation, a boosted-tree model
generalized neglect detection to unseen patients, correctly grouping 90.9% of neglect and 80.0% of non-neglect patients and revealing a lateralized interhemispheric imbalance. Second, a frequency-specific functional-connectivity analysis using graph metrics distinguished groups with 87.0% accuracy at rest (AUC 0.90) and 80.9% during a visuospatial task, identifying reduced right frontal and parieto-occipital centrality as network biomarkers and candidate rehabilitation targets. Third, a spatio-temporal neural network (ESTNet) combined with Bayesian fusion of response-time priors estimated each patient’s neglected field of view, raising sensitivity from 53.6% to 76.7% and achieving 79.6% accuracy. Patients rated the system as highly usable.
Together, these results show that AREEN provides scalable, connectivity-informed, and personalized
neglect assessment, laying the groundwork for real-time, neurofeedback-driven rehabilitation.
Bio: Murat Akcakaya received his B.Sc. in Electrical and Electronics Engineering from Middle East Technical University in Ankara, Turkey, in 2005, and his M.Sc. and the Ph.D degrees in Electrical Engineering from Washington University in St. Louis, in May and December 2010, respectively. He is currently an Associate Professor in the Electrical and Computer Engineering Department of the University of Pittsburgh. His research interests include statistical signal processing and machine learning with applications to noninvasive electroencephalography (EEG) based brain-computer interface (BCI) systems, array signal processing, and physiological signal analysis for health informatics. Dr. Akcakaya was a recipient of the NSF 2019 CAREER Award. He has over 160 publications in leading engineering journals and conferences. His research is supported by NSF, NIH, DOD and DOE.