Preoperative assessment of recurrence-prone subregions within the peritumoral region of glioblastoma: a comparison between diffusion kurtosis and diffusion tensor imaging.

Authors

Yu Zhang, Peipei Wang, Kai Zhao, Eryuan Gao, Guohua Zhao, Gaoyang Zhao, Ting Chen, Xiaoyue Ma, Jie Bai, Yong Zhang, Shaoqiang Han

Published in

Quantitative imaging in medicine and surgery. Volume 16. Issue 9. Pages 675. Sep 01, 2026. Epub Aug 05, 2026.

Abstract

Glioblastoma (Gb) is characterized by extensive infiltration beyond the contrast-enhancing tumor margin, and most recurrences occur within the peritumoral region. Preoperative identification of recurrence-prone subregions within the peritumoral area may help optimize local treatment strategies. Diffusion kurtosis imaging (DKI) and diffusion tensor imaging (DTI) can provide quantitative information on tissue microstructure, but their value in distinguishing recurrence-prone and non-recurrent subregions within the peritumoral region of Gb remains unclear. This study aimed to evaluate and compare the value of preoperative DKI- and DTI-derived histogram parameters in distinguishing recurrence-prone from non-recurrent subregions within the peritumoral region of Gb.
This retrospective study included 55 patients with Gb who subsequently developed recurrence and underwent preoperative magnetic resonance imaging (MRI), including conventional sequences, DKI, and DTI. DKI data were processed to generate DKI- and DTI-derived parametric maps. All preoperative structural images and post-recurrence contrast-enhanced images were automatically coregistered to the preoperative diffusion parametric map space using ITK-SNAP. The edema region on preoperative fluid-attenuated inversion recovery (FLAIR) images and the contrast-enhancing lesion at first recurrence were initially segmented using an nnU-Net model and then jointly reviewed by two radiologists. The overlap between the preoperative FLAIR-hyperintense region and the contrast-enhancing lesion at first recurrence was defined as the recurrence-prone region, whereas the remaining FLAIR-hyperintense area was defined as the non-recurrent region. Histogram parameters were extracted from both regions. Differences in diffusion parameters were analyzed using paired-samples t-tests. P values for parameters showing significant differences were further adjusted using the Benjamini-Hochberg procedure. Receiver operating characteristic (ROC) analysis was performed, and the area under the curve (AUC) was calculated.
In the DKI model, maximum radial kurtosis (RK_max) demonstrated the highest discriminatory power between recurrence-prone and non-recurrent regions, with an AUC of 0.868, sensitivity of 0.891, and specificity of 0.746. In the DTI model, maximum fractional anisotropy (FA_max) showed the best diagnostic performance, with an AUC of 0.834, sensitivity of 0.818, and specificity of 0.746. In this cohort, RK_max showed a higher AUC than FA_max.
Preoperative DKI histogram analysis may help identify recurrence-prone subregions within the peritumoral region of Gb. In this cohort, the best-performing DKI-derived parameter showed the highest AUC compared with the best-performing DTI-derived parameter. These findings are exploratory and require validation in larger studies.

PMID:
42701598
Bibliographic data and abstract were imported from PubMed on 06 Sep 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.