Textual analysis of carotid vessel wall for stroke risk stratification

Abstract

The main objective of this project is to predict the probability of stroke in the future by applying machine learning approaches in analyzing the features of carotid vessel. Stroke is a serious neurological disease which caused millions of people dead all over the world. It is mainly lead by a poor blood supply from the carotid vessel, thus, analysis of the graphs of this artery will probably provide some useful indication of future stroke occurrence. Machine learning approach is adopted to inspect the characteristics of the patients’ carotid vessel wall. The ultrasound scan images of arteries are first transformed into textual formats through a variety of processes. Then the statistics are feed into a neural network as input. Finally, the label indicating stroke-caught or not is used for supervision of the prediction. Following the common approach, the whole data is going to be divided into different sets for training, testing and evaluating.