Fiber density mapping of gliomas: histopathologic evaluation of a diffusion-tensor imaging data processing method

Radiology. 2010 Dec;257(3):846-53. doi: 10.1148/radiol.10100343. Epub 2010 Sep 30.

Abstract

Purpose: To evaluate fiber density mapping (FDM) in the quantification of the extent of destruction of white matter (WM) structures in the center, transition zone, and border zone of intracranial gliomas.

Materials and methods: This retrospective study was approved by the institutional review board. Diffusion-tensor imaging (DTI) and magnetic resonance (MR) imaging-guided biopsies were performed in 20 patients with glioma. FDM is a three-step approach that includes diagonalization of the diffusion tensor, fiber reconstruction for the whole brain, and calculation of fiber density values. Coregistration of FDM data with MR imaging data used for stereotactic biopsy guidance enabled us to correlate these results with histopathologic findings. Data were analyzed by using regression analyses and Hoetelling-Williams and Wilcoxon signed rank tests.

Results: Histopathologic correlation revealed strong negative correlations with both the logarithm of tumor cell number (CN) (R = -0.825) and the percentage of tumor infiltration (TI) (R = -0.909). Complete destruction of WM structures was found when the percentage of TI was 60% or greater and when the tumor CN was 150 or greater. We estimated a fiber density value of 18 as a limit in the identification of fiber structures that are infiltrated with tumor yet are still potentially functional.

Conclusion: FDM provides histologic insight into the structure of WM; therefore, it may help prevent posttreatment neurologic deficits when planning therapy of brain tumors.

MeSH terms

  • Adult
  • Biopsy
  • Brain Neoplasms / pathology*
  • Diffusion Magnetic Resonance Imaging / methods*
  • Female
  • Glioma / pathology*
  • Humans
  • Image Interpretation, Computer-Assisted
  • Linear Models
  • Magnetic Resonance Imaging, Interventional
  • Male
  • Neoplasm Staging
  • Regression Analysis
  • Retrospective Studies
  • Statistics, Nonparametric