Glioma grade assessment by using histogram analysis of diffusion tensor imaging-derived maps

Area of Science:

• Neuro-oncology
• Radiology
• Medical Imaging

Background:

• Morphological validation of imaging in neuro-oncology relies on histology, molecular, and immunohistochemical techniques.
• Diffusion Tensor Imaging (DTI) is a key intracranial diagnostic method for neoplasia studies.
• Histopathology remains the gold standard for glioma grading.

Purpose of the Study:

• To evaluate the feasibility of using discriminant analysis on preoperative DTI-derived image histograms for glioma grading.
• To validate DTI-based glioma grading against histomorphology.

Main Methods:

• Analysis of preoperative DTI data (including fractional anisotropy, diffusivity maps) from 40 glioma patients.
• Generation of histograms from gross tumor volumes with 25 bins per scalar map.
• Multivariate discriminant analysis to select histogram bins for low-grade (LG) vs. high-grade (HG) glioma classification, with leave-one-out cross-validation for accuracy.

Main Results:

• Statistical descriptors of voxel distribution in DTI histograms did not differentiate between LG and HG tumors.
• The developed histogram model demonstrated 88.5% specificity and 85.7% sensitivity in separating LG and HG gliomas.
• Specificity of the model improved when cases with oligodendroglial components were excluded.

Conclusions:

• Histograms constructed from preoperative DTI images of the entire tumor volume can represent glioma grade.
• This histogram-based approach enables discrimination between LG and HG gliomas, validated by histopathology.
• DTI histogram analysis offers a promising non-invasive method for preoperative glioma grading.