AUTOMATIC LIVERDISEASESDIAGNOSISFOR CTIMAGESUSING KERNEL-BASEDCLASSIFIERS
Chien-Cheng Lee · 2006
Inthis paper, akernel-based classifier forautomatic liver diseases diagnosis ofCT images is introduced. Three kinds ofliver diseases areidentified including cyst, hepatoma andcavernous hemangioma. Thediagnosis schemeincludes twosteps: features extraction andclassification. Thefeatures, derived fromgraylevels, co-occurrence matrix, andshape descriptors, areobtained fromtheregion ofinterests (ROIs) amongthenormal andabnormal CTimages. Then, a3-layer hierarchical scheme isadopted intheclassifier. Finally thereceiver operating characteristic (ROC) curve isemployed toevaluate theperformance ofthediagnosis system.