Investigating Acute Ischemic Stroke Subtypes: Predicting Cardiac and Large Artery Atherosclerosis using Image Classification
Authors: Myco Torres, Maureen Montes, Ryan Flannery
Faculty Supervisor: Ilmi Yoon
Department: Computer Science
Strokes are a leading cause of cognitive deficits and chronic disability, accounting for 5.5% of all moralities worldwide. Ischemic strokes are characterized by the blockage and reduction of blood vessels supplying the brain, resulting in restricted blood flow and consequent cell death. This study aims to investigate acute ischemic stroke subtypes, focusing specifically on classifying blood clot origins and predicting cardiac and large artery atherosclerosis. We will focus on analyzing microscopy images to develop predictive machine learning models in order to identify underlying causes of stroke. Multiple convolutional neural networks (CNNs) will be trained to classify high-resolution whole-slide pathology images of patient blood clots to predict its origin. We will evaluate the efficacy of a custom CNN model versus models created using various pretrained models, such as VGG16 and ResNet. Through the use of machine learning and deep neural networks, this study aims to enhance the capacity of healthcare providers to prescribe correct treatments and improve patient outcomes in stroke survivors.