BME Seminar: Aditi Deshpande
Monday, April 18th, 2022, 12:00 p.m.
Aditi Deshpande
BME Ph.D. Candidate
Laksari Lab, PI: Kaveh Laksari
The University of Arizona
"Automatic Stroke Detection & Outcome Prediction Using Cerebrovascular Imaging Biomarkers for Efficient Stroke Triage"
Keating 103
Live Zoom | Passcode: BearDown
(Instructor permission required for enrolled students to attend via Zoom)
Hosts: Dr. Beth Hutchinson and Dr. Russ Witte
Persons with a disability may request a reasonable accommodation by contacting the Disability Resource Center at 621-3268 (V/TTY).
ABSTRACT: Cerebrovascular diseases are a leading cause of death and disability in the US and worldwide and stroke is a major contributor with ~800,000 cases per year in the US - 1 every 4 seconds, leading to a death every 4 minutes. Thus, timely diagnosis and efficient treatment planning are crucial to improving outcomes in stroke patients. Altered brain vasculature is a key phenomenon of stroke and other neurologic disorders and hence, qualitative and quantitative assessment of cerebrovascular morphology is fundamental to understanding brain health and disease.
In this work, we present an automatic method to segment cerebral angiography scans to create a 3D map of brain vasculature and to extract novel geometric features of the vessel network. This method was used to create a probabilistic atlas of human cerebral vasculature using Magnetic Resonance Angiography (MRA) scans of 175 healthy adults and the unique geometric properties of the vascular tree were then used to study morphological changes due to normal aging and also quantify inter-subject neuro-anatomical variations in the major cerebral vessels of healthy adults. This knowledge of healthy vascular patterns and geometry was then exploited for quantification and statistical analysis of vascular alterations in Acute Ischemic Stroke (AIS) and Alzheimer’s Disease (AD) patients, the biggest cerebrovascular and neurodegenerative disorders.
Using these cerebrovascular atlas and vessel features, we developed an algorithm to automatically detect the presence and location of large vessel occlusions (LVO) to meet the clinical need for validated, standardized, fast and automatic diagnosis and triage of AIS patients. Lastly, we propose a method to improve the protocol for patient selection in Endovascular Thrombectomy (EVT) treatment of AIS patients using patient-specific imaging-based features over the current protocol which is based on a fixed 6-hour time to treat window. This was demonstrated by the significantly improved accuracy of prediction of the 90-day functional outcome in stroke patients by incorporating our novel geometric features of the cerebrovasculature, over current methods in literature. Our approach addresses the variations in vascular anatomy which often translate to varying hemodynamics and response to treatment in AIS patients and can significantly improve patient outcomes. The unique geometric features and quantitative comparisons presented here demonstrate the potential for using vascular morphology as a non-invasive imaging biomarker for neurologic disorders.