Magnetic Resonance Imaging Gastrography: Evaluation of the Dark Lumen Technique Compared With Conventional Gastroscopy in Patients With Malignant Gastric Disease.

摘要

OBJECTIVES:: We sought to demonstrate the feasibility of depicting gastric tumors using magnetic resonance imaging (MRI) while applying the dark lumen technique. The findings were correlated with conventional gastroscopy. In addition, we evaluated the screening for lymph nodes and liver metastases during the same session to identify potential tumor spread. MATERIALS AND METHODS:: The study included 15 patients with known malignant gastric disease. Conventional gastrointestinal endoscopy was performed in all patients as gold standard. All patients were examined with computed tomography for tumor staging. MRI was performed using 1 L of tap water as oral contrast agent for all protocols. The MRI program included an axial T1-weighted (T1w; 2D-FLASH) sequence, an axial STIR and T2w (TSE) sequence, and 2 postcontrast T1w (3D coronal /2D axial FLASH) sequences using 0.2 mmol/kg gadolinium diethylenetriaminepenta-acetic acid as intravenously injected contrast agent. Qualitative analysis and comparison with conventional gastroscopy were performed. RESULTS:: The images obtained with the postcontrast 3D coronal T1w-FLASH sequence were the most suitable in identifying gastric tumors. Complete correlation with conventional gastroscopy was achieved in 80% of the cases and partial correlation in 13% of the cases. The same imaging sequence was also appropriate for the evaluation of lymph nodes. For the identification of liver metastases, the images obtained with the axial postcontrast T1w 2D-FLASH sequence provided a higher diagnostic confidence as compared with other imaging protocols. CONCLUSIONS:: Applying the dark lumen technique through MRI is suitable for imaging gastric tumors and has the potential to become a "one-stop shopping" method because of the possibility for lymph node evaluation and screening for metastases during the same session based on the same images.