OBJECTIVETo evaluate the clinical value of diffusion weighted imaging (DWI) combined with conventional sequences of magnetic resonance imaging (T1 and T2-weighted imaging) for the diagnosis of rectal cancer.

METHODSDWI and conventional sequences were performed in 29 patients with endoscopically diagnosed rectal cancer and 15 patients without rectal cancer. Two doctors who were blind to the history of the patients interpreted the imaging findings. The sensitivity and specificity of conventional sequences with and without DWI were analyzed using receiver operating characteristic curve (ROC).

RESULTSThe areas under ROC were 0.915 and 0.930 for conventional sequences alone, and 0.990 and 0.994 for conventional sequences with DWI, respectively, indicating that although both of them were optimal methods for the diagnosis of rectal cancer, the accuracy of conventional sequences with DWI was significantly superior to that of conventional sequence alone (P < 0.05). The Kappa value was 0.850 for conventional sequences alone and 0.858 for DWI with conventional sequences.

CONCLUSIONDWI was necessary for the diagnosis of rectal cancer when performing conventional sequences.