Schedule of Events | Symposia

A Neuroinformatics-Computational Approach to the Assessment of Visuospatial Neglect

Poster Session A - Saturday, March 29, 2025, 3:00 – 5:00 pm EDT, Back Bay Ballroom/Republic Ballroom

Vinoth Jagaroo^1,2 (jagaroo@bu.edu), Woo Zhong Han³, Kiran Jagaroo⁴, Sophia Andrienko³, Annie Chen³, Darian Cheung³, Daniel Doh³, Zekai Wu³, Joshua Yip³, Justin Zheng³; ¹Boston Univ. School of Medicine (Behavioral Neuroscience), ²Emerson College, ³Boston University Center for Computing & Data Sciences and SPARK! Program, ⁴University of Massachusetts-Boston, College of Science & Mathematics

Hemispatial neglect, a complex neurocognitive disorder with differential neural mechanisms, manifests in 30% of stoke patients. Neuropsychological assessment of neglect is hindered by outmoded ‘paper-and-pencil’ tests, ill-suited to complex cognitive and neural dynamics of neglect. There is pressing need for computational methods in the assessment of neglect. We describe the successful development of an informatics platform for the analysis of neglect. Theoretical impetus for the platform stems from the spatiotopic model of neglect, centered on Brodmann’s area 7 (posterior parietal) – representational neglect is function of spatiotopic compression that can be mathematically described. The platform uses a scalable grid -- array of cells of a computer screen that record the coordinates of content displayed. Grid-based mapping applies Manhattan Distance -- coordinate translations with the center (0,0 coordinate) as the reference point. Two visuospatial tests, Letter Cancellation and the Boston Visuospatial Battery (BVB), were digitized. The clinician selects the letter cancellation task or an image from the BVB. Selected test is displayed on the grid matrix. Patient uses a mouse or touchpad to trace the image. Based on neglect gradient extracted by the software, the image can be incrementally moved. Patient then retraces the image, and the process is repeated until no neglect is recorded. A mathematical rendering of the gradient of neglect is generated. With further data acquisition, we aim to (a) apply machine learning methods to combine several predictive models to make predictions about patients’ visuospatial prognosis, and (b) tie their computational patterns of neglect to fMRI imaging data.

Topic Area: ATTENTION: Spatial