Applying soft computing techniques for emboli detection

Abstract

Transcranial Doppler (TCD) ultrasound can be used for detecting asymptomatic circulating cerebral emboli. Asymptomatic embolic signals (ES) can be considered as an important indicator of increased stroke risk. When an ES is reflected by an embolus, these signals have usually larger amplitude than the signals from normal blood flow. Also, they show a transient characteristic. In this study, a number of data mining methods were tested and compared for the classification of embolic signals. A data set containing 100 ES, 100 speckle and 100 artifact were used. The data mining algorithms were compared with each other in order to obtain the best sensitivity in embolic signal detection. Receiver operating curve (ROC) analysis results show that Naive Bayes Tree and Naive Bayes algorithms perform better than the others. © 2019 Elsevier B.V., All rights reserved.