Colorectal　cancer　is　a　type　of　malignant　from　the　intestinal　tract.　The　accurate　diagnosis　of　colorectal　polyps　can　effectively　guarantee　the　life　safety　of　potential　patients.　There　are　supervised　radionics　methods　and　deep　learning　methods　when　determining　whether　polyps　exist.　This　paper　proposes　to　obtain　global　features　set　from　computed　tomographic　colonography　(CTC)　images　by　radionics　methods　and　the　local　features　set　using　deep　convolutional　neural　network　simultaneously.　Specifically,　we　use　the　chaotic　evolution　algorithm　to　optimize　the　parameters　in　the　support　vector　machine　classifier　and　random　forest　classifier.　Finally,　our　hybrid　method　achieved　better　classification　result　by　random　forest　classifier　on　combinational　features　in　which　accuracy　is　91.318%　from　the　experiment.　©　2020,　Springer　Nature　Singapore　Pte　Ltd.