Image Classification of Stroke Blood Clot Origin Using Deep Convolutional Neural Networks and Visual Transformers: A Kaggle Competition

Abstract

Stroke is one of two main causes of death worldwide. Many individuals suffer from ischemic stroke every year. Only in US, more over 700,000 individuals meet ischemic stroke due to blood clot blocking an artery to the brain every year. «Stroke symptoms typically start suddenly, over seconds to minutes, and in most cases do not progress further. The symptoms depend on the area of the brain affected. The more extensive the area of the brain affected, the more functions that are likely to be lost. » «Ischemic stroke occurs because of a loss of blood supply to part of the brain, initiating the ischemic cascade. » In the case of subsequent strokes, it is possible to mitigate consequences if physicians can determine stroke etiology. The paper describes particular approach how to apply Artificial Intelligence for purposes of separating two major acute ischemic stroke (AIS) etiology subtypes: cardiac and large artery atherosclerosis. Four deep neural network architectures and simple ensemble method are used in the approach.