Abstract
Twenty one patients undergoing surgical treatment for rectal carcinoma were examined by means of the first prototype model of MR endoscope (designed and provided by Yokokawa Medical Co. and Olympus Opt. Co., Tokyo, Japan) for the staging before surgery. The same patients also underwent the staging with endoscopic ultrasonography (EUS) before surgery. As a surface coil at the tip of antenna probe was accurately positioned on the cancerous lesion using non-magnetic fiberscope, all lesions were detected by MR endoscopy. The whole rectal wall was apparent as 3 or 5 layers on images obtained with 3D fast SPGR method of MR endoscopy. The diagnostic accuracies in detection of rectal wall infiltration by MR endoscopy were 100% (2/2) in the cases in which cancerous invasion was detected in the submucosa by MR endoscopy, 50% (3/6) in those of the muscularis propria, 83% (10/2) in those of perirectal fat, 100% (1/1) in those of the adjacent organs and 76% (16/21) in all cases. The diagnostic accuracies in detection of rectal wall infiltration by EUS were 100% (3/3) in the cases in which the cancerous invasion was detected in the submucosa by EUS, 57% (4/7) in those of the muscularis propria, 90% (9/ 10) in those of perirectal fat, 100% (1/1) in those of the adjacent organs and 81% (17/21) in all cases. The sensitivity of MR endoscopy in detection of lymph node infiltration was 85.7% ; specificity, 61.5% ; and accuracy, 70.0%. EUS showed the same sensitivity, specificity and accuracy. These findings suggest that MR endoscopy may be as accurate as EUS in the preoperative staging of rectal carcinoma. In comparison with EUS, MR endoscopy has multiplanar capabilities and wide scanning area. But, MR endoscopy needs MRI apparatus and MR imaging cannot be acquired in real time. The improvement of MR endoscope is desired.
