Use of combination of PCA and ANFIS in infarction volume growth rate prediction in ischemic stroke

Stroke is one of the leading causes of death in the world today. Treatment of stroke using a procedure called Decompressive Hemicraniectomy requires the patient to undergo multiple CT scans in order to determine the size of the stroke affected area, also known as the infarction volume. Recent studies have focused on the automation of infarction growth rate prediction by the utilization of machine learning techniques. These, when applied correctly significantly reduce the amount of time required to determine the infarction volume in stroke patients. In this paper, we propose a system that is able to predict the infarction volume growth rate based on only one CT scan and several clinical measurements. The proposed technique uses a combination of Principal Component Analysis (PCA) and Adaptive Neuro-Fuzzy Inference System (ANFIS) and has shown to perform better in predicting the infarction volume. Dimensionality reduction in clinical data is first performed by reducing the number of features in the given stroke dataset. Then the target infarction volume growth rate is predicted using Adaptive Neuro-Fuzzy Inference System. The dataset used had 122 instances with 15 features. The obtained prediction from our proposed system consisting of a combination of PCA and ANFIS had a root mean square error of 0.196, cosine distance of 0.464 and outperformed that obtained by prediction with Adaptive Neuro-Fuzzy Inference System alone which had an error of 0.439 and a cosine distance of 0.616. This paper was made possible by National Priorities Research Program (NPRP) grant No. 7342109 from the Qatar National Research Fund (a member of Qatar Foundation). The statements made herein are solely the responsibility of the authors. Scopus