MR Microimaging of Benign and Malignant Nodes in the Neck

Abstract

We evaluated the diagnostic criteria of high-resolution MRI in differentiating benign and malignant cervical nodes that were palpable and superficial in the neck. We performed MR microimaging on 24 histologically proven metastatic nodes, 14 histologically proven lymphomas, and 35 histologically or clinically proven benign nodes in the necks of 26 patients. The lymph nodes were imaged with T1-weighted spin-echo, fat-suppressed T2-weighted turbo spin-echo, and spin-echo diffusion-weighted echo-planar sequences using a 47-mm microscopy coil. MR microimaging provided high-resolution images of the nodes. Hilar fat was lost in 92%, 79%, and 46% of the metastatic nodes, lymphomas, and benign nodes, respectively. Smooth nodal margins were lost in 58%, 23%, and 9% of metastatic nodes, lymphomas, and benign nodes, respectively. Heterogeneous nodal parenchyma on T1- or fat-suppressed T2-weighted images, or both, was observed in 88%, 29%, and 23% of metastatic nodes, lymphomas, and benign nodes, respectively. The apparent diffusion coefficients were significantly different among these three node groups (p < 0.001), with metastatic nodes being the highest, followed by benign nodes. Logistic regression analyses showed that heterogeneous nodal parenchyma and apparent diffusion coefficient levels were significant in discriminating metastatic nodes, and apparent diffusion coefficient levels in discriminating lymphomas. Combined use of these MR microscopic criteria on nodal architecture and apparent diffusion coefficients yielded 90% accuracy (86% sensitivity, 94% specificity) and 93% accuracy (85% sensitivity, 95% specificity) for discriminating metastatic nodes and lymphomas, respectively. The nodal architecture and apparent diffusion coefficient levels on MR microimaging may provide useful information in diagnosing benign and malignant nodes in the neck.