Diagnostic performance of MRI relative to CT for metastatic nodes of head and neck squamous cell carcinomas

J Magn Reson Imaging. 2007 Dec;26(6):1626-33. doi: 10.1002/jmri.21187.

Abstract

Purpose: To compare the diagnostic abilities of magnetic resonance imaging (MRI) and computed tomography (CT) based on the architectural changes in the nodal parenchyma.

Materials and methods: We retrospectively studied histologically proven 70 metastatic and 52 reactive nodes in the necks of 38 patients with head and neck squamous cell carcinomas who had undergone both CT and MRI. We assessed the detectability of the architectural changes in the nodal parenchyma that were suggestive of cancer focus (cancer nest, necrosis, and keratinization). The diagnostic abilities of CT and MRI were assessed by three observers separately for the small (<10 mm in minimum axis diameter) and large (>or=10 mm) nodes.

Results: MRI was significantly more effective than CT in diagnosing small metastatic nodes, yielding 83% sensitivity, 88% specificity, and 86% accuracy. However, the diagnostic abilities of MRI and CT were similar for large metastatic nodes; MRI yielded 100% sensitivity, 98% specificity, and 99% accuracy. receiver operating characteristic analysis also indicated that the Az values were significantly higher for MRI than for CT (0.927 vs. 0.822, P = 0.00054) for the detection of small nodes.

Conclusion: MRI is superior to CT in the diagnosis of metastatic nodes from head and neck squamous cell carcinomas.

Publication types

  • Comparative Study

MeSH terms

  • Aged
  • Analysis of Variance
  • Carcinoma, Squamous Cell / diagnosis*
  • Carcinoma, Squamous Cell / diagnostic imaging
  • Carcinoma, Squamous Cell / pathology
  • Contrast Media
  • Female
  • Gadolinium DTPA
  • Head and Neck Neoplasms / diagnosis*
  • Head and Neck Neoplasms / diagnostic imaging
  • Head and Neck Neoplasms / pathology
  • Humans
  • Lymphatic Metastasis
  • Magnetic Resonance Imaging / methods*
  • Male
  • Middle Aged
  • ROC Curve
  • Tomography, Spiral Computed / methods*

Substances

  • Contrast Media
  • Gadolinium DTPA