Preoperative diagnosis of gastric cancer using 2-D magnetic resonance imaging with 3-D reconstruction techniques.

OBJECTIVE: To investigate the clinical value of 2-D magnetic resonance imaging (MRI) with 3-D reconstruction techniques for the preoperative diagnosis and TNM-staging of gastric cancer.

METHODS: Using a Philips Gyroscan NT 1.0T superconductive unit, MRI using the water-filling method was performed in 15 patients with suspected gastric cancers. The 2-D MRI sequences included TSE-T1WI, TSE-T2WI and fat suppression (SPIR). The source images of magnetic resonance hydrography (heavily TSE-T2WI sequence) were reconstructed using the Philips EasyVision viewing workstation. Four 3-D postprocessing algorithms, including maximum intensity projection, surface shaded viewing, volume rendering and virtual endoscopy, were performed and compared with the results of a barium study and endoscopy. All 15 patients with 16 gastric cancers had their diagnosis confirmed by postoperative pathological findings.

RESULTS: 2-D MRI and 3-D reconstruction images were successfully obtained for all 15 patients. The maximum intensity projection, surface shaded viewing, and volume rendering images corresponded to the upper gastrointestinal series findings, and the virtual endoscopy images corresponded to the gastroscopic views. In 16 gastric lesions, MRI correctly diagnosed 14 (87.5%) advanced gastric cancers, and the tumor location, size and classification were also accurately identified. The accuracy of MRI for determining the preoperative TNM stage was 64.3% (9/14), and there was significant correlation between these results and those from the histopathological studies (P < 0.01). Based on T, N and M factors, the staging accuracy of MRI was 71.4% (10/14), 57.1% (8/14) and 85.7% (12/14), respectively.

CONCLUSIONS: 2-D MRI with 3-D reconstruction is an effective method for the preoperative diagnosis and TNM staging of gastric cancer. However, the detection of early cancers or benign lesions and N-staging should be further studied.