Support Vector Machine Approach to Cardiac SPECT Diagnosis

摘要

This article presents the use of Support Vector Machines (SVM) to diagnose the ischemic heart disease using heart images obtained from Single Proton Emission Computed Tomography (SPECT). The data set came from 267 different patients and was divided into several sub-sets containing training and validation data. The study consisted in comparing results of classifying cardiac SPECT images using SVMs with those obtained using another method of machine learning CLIP3 which is a combination of the decision tree algorithm and the rule induction algorithm. Validations carried out using a SPECT image database have shown that SVMs are good in generalising knowledge gained about multi-dimensional data with relatively little training data.