# Histogram analysis based on DTI and NODDI for differentiating atypical high-grade glioma from primary central nervous system lymphoma

## TL;DR

This study uses advanced MRI techniques to distinguish between two brain tumors, high-grade glioma and lymphoma, with high accuracy.

## Contribution

The study introduces histogram analysis of DTI and NODDI parameters to improve differentiation between atypical high-grade glioma and PCNSL.

## Key findings

- Histogram features of DTI and NODDI parameters can classify high-grade glioma and PCNSL with statistical significance.
- The 75th percentile of the orientation dispersion index (ODI75th) achieved 94.55% accuracy in distinguishing the tumors.
- ODI75th within the enhancing tumor region showed high sensitivity (96%) and specificity (93.33%).

## Abstract

Distinguishing between high-grade glioma (HGG) and primary central nervous system lymphoma (PCNSL) is of paramount clinical importance, as these entities necessitate substantially different therapeutic approaches. The differential diagnosis becomes particularly challenging when HGG presents without characteristic magnetic resonance imaging (MRI) features, making it difficult to differentiate from PCNSL. The diffusion tensor imaging (DTI) and neurite orientation dispersion and density imaging (NODDI) offer quantitative assessments of water molecule diffusion within tissues, thereby providing potential means to characterize microstructural differences between HGG and PCNSL. This study aims to evaluate the diagnostic efficacy of histogram analysis based on DTI and NODDI parameters in differentiating atypical HGG from PCNSL.

We retrospectively reviewed patients who underwent multi-b-value diffusion-weighted imaging (DWI) at our institution. The multi-b-value DWI was performed using a single-shot echo-planar imaging (EPI) sequence with six b-values (0, 500, 1,000, 1,500, 2,000, and 2,500 s/mm2) distributed across 30 directions. The DTI and NODDI model were employed to derive the parametric maps of apparent diffusion coefficient (ADC), fractional anisotropy (FA), intracellular volume fraction (ICVF), isotropic volume fraction (ISOVF), and orientation dispersion index (ODI). Two regions of interest (ROIs) were manually delineated within the enhancing tumor area and the peritumoral edema. Histogram features were extracted from these ROIs. Comparisons between HGG and PCNSL were performed. Receiver operating characteristic (ROC) curves were drawn, and the area under the curve (AUC), sensitivity, specificity, and accuracy were calculated. p < 0.05 was considered statistically significant.

A total of 55 patients (30 with atypical HGG and 25 with PCNSL), were included in this study. Several histogram features of parameters could be used to classify the HGG and PCNSL (p < 0.05). The 75th percentile of the orientation dispersion index (ODI75th) within the enhancing tumor region demonstrated the highest diagnostic performance (AUC = 0.985). At an optimal threshold of 0.604, ODI75th yielded a sensitivity of 96%, a specificity of 93.33%, and an accuracy of 94.55% for distinguishing HGG from PCNSL.

DTI-and NODDI-based histogram analysis demonstrates the potential to differentiate between atypical HGG and PCNSL. ODI75th within the enhancing tumor region showed the most favorable diagnostic performance.