Utilizing multiple diffusion metrics in evaluation of corticospinal tract injury in patients with glioblastoma

Ting Chen, Peipei Wang, Eryuan Gao, Jie Bai, Guohua Zhao, Kai Zhao, Gaoyang Zhao, Yu Zhang, Yong Zhang, Mengzhu Wang, Guang Yang, Boyan Xu, Xiaoyue Ma, Jingliang ChengFrontiers in Neuroscience 2025 · DOI

Abstract

Evaluating corticospinal tract (CST) injury is critical for surgical planning in glioblastoma (GBM) patients. This study compared diffusion parameters from four advanced models: Diffusion Tensor Imaging (DTI), Diffusion Kurtosis Imaging (DKI), Neurite Orientation Dispersion and Density Imaging (NODDI), and Mean Apparent Propagator-MRI (MAP-MRI). While both DKI-derived Mean Kurtosis (MK) and NODDI-derived Intracellular Volume Fraction (ICVF) achieved high diagnostic accuracy (AUC = 0.813), MK provided a more balanced sensitivity-specificity profile. Furthermore, MK significantly correlated with clinical muscle strength, suggesting it is a superior biomarker for assessing microstructural CST damage in clinical settings.

Introduction

Glioblastoma (GBM) is a highly invasive brain tumor that frequently involves the corticospinal tract (CST), the primary pathway for motor signals. Neurosurgeons face a high-stakes balancing act: maximizing tumor resection while preserving the CST to avoid permanent motor deficits. Traditional DTI, the current clinical standard, is limited by its assumption of simple Gaussian water diffusion, which often fails in the complex microenvironment of tumors and crossing fibers. This study investigates whether advanced non-Gaussian diffusion models can provide a clearer picture of microstructural injury to guide surgical decisions.

Methodology

The researchers analyzed a clinical dataset of 76 GBM patients (22 with motor weakness, 54 with normal function). Imaging was performed on a 3T MRI scanner using a multi-shell diffusion protocol with 10 b-values. Four distinct mathematical models were compared:

  • DTI: The standard clinical model.
  • DKI: An extension of DTI that accounts for tissue complexity (non-Gaussianity).
  • NODDI: A model that separates signals into intracellular, extracellular, and CSF compartments.
  • MAP-MRI: An advanced model calculating the 3D probability of water displacement.

CST reconstruction was performed using Generalized Q-sampling Imaging (GQI). To account for individual biological differences, "relative" metrics were calculated by comparing the tumor-affected side to the healthy contralateral side.

Results and Key Insights

The study demonstrated that metrics reflecting tissue complexity are more sensitive indicators of CST health than simple directionality metrics like Fractional Anisotropy (FA).

  • Diagnostic Performance: Relative MK and ICVF tied for the highest diagnostic accuracy with an AUC of 0.813. However, MK was deemed more clinically useful due to its balanced sensitivity (81.82%) and specificity (79.63%).
  • Functional Correlation: A significant correlation was found between relative MK values and physical muscle strength (r = 0.577, p < 0.001). This bridges the gap between mathematical imaging metrics and a patient's physical reality.
Figure 4: Visualization of Mean Kurtosis (MK) mapping on the corticospinal tract in a patient with glioblastoma, showing the sensitivity of kurtosis metrics to microstructural complexity.
  • Axonal Integrity: The study found that metrics related to "free water" (ISOVF) did not change significantly between groups. This suggests that GBM-induced weakness is primarily caused by the destruction of nerve fibers themselves rather than just peritumoral edema.
Figure 6: Comparative diffusion metric map illustrating microstructural alterations within the CST, highlighting the diagnostic potential of non-Gaussian models.

Clinical Implications and Limitations

While DTI remains the standard, this research makes a compelling case for DKI-derived Mean Kurtosis (MK) as a superior metric. It is sophisticated enough to capture complex pathology yet practical enough to potentially guide the next generation of neurosurgery.

However, several hurdles remain before widespread adoption. The advanced models require specialized software and significant computing power, which can be a bottleneck in fast-paced clinical settings. Additionally, larger multi-center studies and longitudinal data are needed to validate these findings and determine if preoperative metrics can predict long-term motor recovery.

Conclusion

Advanced diffusion models, particularly DKI, provide a more accurate evaluation of CST injury in glioblastoma patients than traditional DTI. The balanced performance and clinical relevance of Mean Kurtosis (MK) support its potential role in aiding surgical planning and preserving motor function.

Summary updated

Abstract

Diffusion parameters from DTI, DKI, NODDI, and MAP-MRI are useful for evaluating CST injury. While DKI-derived MK and NODDI-derived ICVF achieved identical high AUC values, MK exhibited a more balanced sensitivity-specificity profile for assessing microstructural alterations in CST injury, this advantage of DKI may better address clinical demands, potentially aiding in surgical planning and preserving motor function in patients with GBM.
BibTeX
@inproceedings{chen2025utilizing,
  author = {Ting Chen and Peipei Wang and Eryuan Gao and Jie Bai and Guohua Zhao and Kai Zhao and Gaoyang Zhao and Yu Zhang and Yong Zhang and Mengzhu Wang and Guang Yang and Boyan Xu and Xiaoyue Ma and Jingliang Cheng},
  title = {{Utilizing multiple diffusion metrics in evaluation of corticospinal tract injury in patients with glioblastoma}},
  booktitle = {Frontiers in Neuroscience},
  year = {2025},
  doi = {10.3389/fnins.2025.1605786},
}
Figure