Scholarlink

Objective <p>This study aims to develop a voxel-wise quantitative framework for the T2-FLAIR mismatch signal in routine MRI scans to differentiate IDH-mutant astrocytomas from oligodendrogliomas.</p> Methods <p>This retrospective study analyzed 43 patients with confirmed diffuse gliomas (25 astrocytomas and 18 oligodendrogliomas) using preoperative T2-weighted and FLAIR MRI. A voxel-wise log-ratio mismatch map was created, and whole-tumor regions of interest (ROIs) were defined through manual segmentations and inward tumor boundary restrictions. Five mismatch biomarkers were extracted: PositiveMismatchRatio, MeanMismatchStrength, MismatchStd, PositiveMismatchMean, and LargestMismatchClusterRatio. Group comparisons utilized the Mann–Whitney U test, and diagnostic performance was assessed through receiver operating characteristic (ROC) analysis.</p> Results <p>The analysis revealed that several voxel-wise mismatch biomarkers presented significantly higher values in astrocytomas compared to oligodendrogliomas, particularly in core-restricted ROIs. The best overall performance was achieved by LargestMismatchClusterRatio in the 3-mm core ROI (AUC, 0.786; 95% CI, 0.640–0.914), followed closely by PositiveMismatchRatio in the same ROI (AUC, 0.779; 95% CI, 0.631–0.910). Threshold-based analysis showed that PositiveMismatchRatio in the 3-mm core ROI provided the strongest rule-in performance, with 60.0% sensitivity, 94.4% specificity, a positive likelihood ratio of 10.80, and a diagnostic odds ratio of 25.5. Overall, core restriction improved discrimination performance compared to whole-tumor assessments.</p> Conclusions <p>Core-based voxel-wise quantification of T2-FLAIR signal discordance showed preliminary diagnostic value for differentiating IDH-mutant astrocytomas from oligodendrogliomas, particularly as a high-specificity rule-in approach. External validation in larger multicenter cohorts is required before clinical application.</p> Graphical Abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Core-based voxel-wise T2-FLAIR mismatch analysis for differentiation of IDH-mutant astrocytomas from oligodendrogliomas

  • 期刊论文

  • 发表日期：

    2026年7月1日

查看全文
  • Seyit Erol,
  • Ahmet Baytok,
  • Ayşe Arı,
  • Halil Özer,
  • Hakan Cebeci

摘要

Objective

This study aims to develop a voxel-wise quantitative framework for the T2-FLAIR mismatch signal in routine MRI scans to differentiate IDH-mutant astrocytomas from oligodendrogliomas.

Methods

This retrospective study analyzed 43 patients with confirmed diffuse gliomas (25 astrocytomas and 18 oligodendrogliomas) using preoperative T2-weighted and FLAIR MRI. A voxel-wise log-ratio mismatch map was created, and whole-tumor regions of interest (ROIs) were defined through manual segmentations and inward tumor boundary restrictions. Five mismatch biomarkers were extracted: PositiveMismatchRatio, MeanMismatchStrength, MismatchStd, PositiveMismatchMean, and LargestMismatchClusterRatio. Group comparisons utilized the Mann–Whitney U test, and diagnostic performance was assessed through receiver operating characteristic (ROC) analysis.

Results

The analysis revealed that several voxel-wise mismatch biomarkers presented significantly higher values in astrocytomas compared to oligodendrogliomas, particularly in core-restricted ROIs. The best overall performance was achieved by LargestMismatchClusterRatio in the 3-mm core ROI (AUC, 0.786; 95% CI, 0.640–0.914), followed closely by PositiveMismatchRatio in the same ROI (AUC, 0.779; 95% CI, 0.631–0.910). Threshold-based analysis showed that PositiveMismatchRatio in the 3-mm core ROI provided the strongest rule-in performance, with 60.0% sensitivity, 94.4% specificity, a positive likelihood ratio of 10.80, and a diagnostic odds ratio of 25.5. Overall, core restriction improved discrimination performance compared to whole-tumor assessments.

Conclusions

Core-based voxel-wise quantification of T2-FLAIR signal discordance showed preliminary diagnostic value for differentiating IDH-mutant astrocytomas from oligodendrogliomas, particularly as a high-specificity rule-in approach. External validation in larger multicenter cohorts is required before clinical application.

Graphical Abstract