Inthispaper,anelectricalresistancetomography(ERT)imagingmethodisusedasaclassifier,andthentheDempster-Shafer’sevidencetheorywithfuzzyclusteringisintegratedtoimprovetheERTimagequality.Thefuzzyclusteringisappliedtodeterminingthekeymassfunction,anddealingwiththeuncertain,incompleteandinconsistentmeasuredimagingdatainERT.Theproposedmethodwasappliedtoimageswiththesameinvestigatedobjectundereighttypicalcurrentdrivepatterns.ExperimentswereperformedonagroupofsimulationsusingCOMSOLMultiphysicstoolandmeasurementswithapieceofporcinelungandapairofporcinekidneysastestmaterials.Comparedwithanysingledrivepattern,theproposedmethodcanprovideimageswithaspatialresolutionofabout10%higher,whilethetimeresolutionwasalmostthesame.