A Machine-Learning Algorithm Toward Color Analysis for Chronic Liver Disease Classification, Employing Ultrasound Shear Wave Elastography

摘要: Abstract The purpose of the present study was to employ a computer-aided diagnosis system that classifies chronic liver disease (CLD) using ultrasound shear wave elastography (SWE) imaging, with stiffness value-clustering and machine-learning algorithm. A clinical data set 126 patients (56 healthy controls, 70 CLD) analyzed. First, an RGB-to-stiffness inverse mapping technique employed. five-cluster segmentation then performed associating corresponding different-color regions certain value ranges acquired from SWE manufacturer-provided color bar. Subsequently, 35 features (7 for each cluster), indicative physical characteristics existing within image, were extracted. stepwise regression analysis toward feature reduction used derive reduced subset fed into support vector machine classification algorithm classify CLD cases. highest accuracy in subject discrimination model 87.3% sensitivity specificity values 93.5% 81.2%, respectively. Receiver operating characteristic curve gave area under 0.87 (confidence interval: 0.77–0.92). quantifies information terms images discriminates cases is introduced. New objective parameters criteria employing provided by can be considered important step color-based interpretation, could assist radiologists' diagnostic performance on daily basis after being installed PC employed retrospectively, immediately examination.