Comparison between Glioblastoma and Primary Central Nervous System Lymphoma Using MR Image-based Texture Analysis

Area of Science:

• Neuroradiology
• Medical Imaging Analysis
• Oncology

Background:

• Glioblastoma (GBM) and primary central nervous system lymphoma (PCNSL) are distinct primary brain tumors.
• Differentiating between GBM and PCNSL is crucial for appropriate treatment and patient management.
• Magnetic Resonance (MR) imaging is a key modality for evaluating brain lesions.

Purpose of the Study:

• To identify and differentiate between glioblastoma (GBM) and primary central nervous system lymphoma (PCNSL) using MR image-based texture features.
• To evaluate the efficacy of texture analysis in distinguishing these two challenging brain tumors.

Main Methods:

• Retrospective analysis of 44 GBM and 16 PCNSL patients.
• Calculation of 67 texture features from post-contrast T1-weighted MR images.
• Application of Principal Component Analysis (PCA) and statistical comparisons (FDR < 0.05) to identify distinguishing features.

Main Results:

• Twenty-one texture features met reproducibility and statistical significance criteria.
• Principal Component Analysis identified first-order entropy, median, GLRLM run length non-uniformity, and run percentage as key discriminators.
• GBM showed significantly higher entropy and run length non-uniformity, and lower run percentage and median compared to PCNSL.

Conclusions:

• MR image-based texture analysis, particularly using first-order entropy, median, GLRLM run length non-uniformity, and run percentage, can effectively differentiate between GBM and PCNSL.
• These texture features offer valuable insights for non-invasive tumor characterization in clinical practice.