03/25/2026
By Danielle Fretwell

The Francis College of Engineering, Department of Biomedical Engineering & Biotechnology, invites you to attend a Doctoral Dissertation defense by Vitaly Dobromyslin on: "Integrating fMRI-Derived Biomarkers and Machine Learning for Enhanced Cerebral Infarct Detection."

Candidate Name: Vitaly Dobromyslin
Degree: Doctoral
Date: Wednesday, April 8, 2026
Time: 1 - 2 p.m.
Location: Virtual defense via Zoom. Please contact advisor for Zoom link.

Committee:

  • Advisor: Wenjin Zhou, Assistant Professor, Miner School of Computer & Information Sciences, UMass Lowell
  • Benuyan Liu, Professor, Director, Miner School of Computer & Information Sciences, UMass Center for Digital Health (CDH), Computer Networking Lab, CHORDS, UMass Lowell
  • Yu Cao, Professor, Director, Miner School of Computer & Information Sciences, UMass Center for Digital Health (CDH), UMass Lowell

Abstract:
Aging population, lifestyle changes, and increasing healthcare costs make stroke diagnosis and prevention one of the critical public health issues globally. Incidence of stroke roughly doubles every 10 years after the age of 55, with U.S. Centers for Disease Control and Prevention (CDC) naming stroke as the fourth leading cause of death and a leading cause of serious long-term disability in US. The repeat infarcts are associated with higher risk of successive infarcts and increased morbidity. Unfortunately no single brain imaging modality can currently provide accurate and safe stroke detection at both acute and chronic stages. Furthermore there is a need to develop imaging biomarkers that correlate with stroke recovery and provide prognostic information for the patient. This work aims to expand stroke detection capabilities of the structural T1-weighted (T1-w) MRI scans by identifying promising resting state functional MRI (rs-fMRI) features with auto machine learning techniques. By training vision transformer model on the T1-w stroke data from one dataset, and evaluating it on an independent dataset, we highlight the dataset drift concerns. We propose separate rs-fMRI processing pipelines for detecting gray and white matter stroke, as well as offer insights for future stroke detection improvements.