Related Experiment Video
Updated: Mar 31, 2026
06:48
Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images
Published on: January 7, 2019
9.6K
Reliability of Semi-Automated Segmentations in Glioblastoma
1Department of Neuroradiology, Klinikum rechts der Isar, Technical University of Munich, Ismaninger Str. 22, 81675, Munich, Germany. thomas-huber@tum.de.
Clinical Neuroradiology
|October 23, 2015
Summary
Semi-automated tumor volumetry for glioblastoma is reliable across different users. Segmentation of FLAIR changes is more accurate than contrast enhancement, aiding early detection of disease progression.
Area of Science:
- Neuro-oncology
- Medical imaging analysis
- Quantitative MRI
Background:
- Accurate assessment of glioblastoma (GBM) tumor volume is crucial for evaluating treatment response.
- Quantitative volumetric measurements of contrast-enhancing (CE) or fluid-attenuated inversion recovery (FLAIR) hyperintense tumor compartments are needed.
- Semi-automated segmentation tools aim to provide objective tumor volume assessment.
Observation:
- A semi-automated, region-growing segmentation tool was used for 320 segmentations of FLAIR and CE tumor compartments in GBM patients.
- Raters included neuroradiologists, medical students, and volunteers, assessed for intra- and inter-rater reliability using intra-class correlation (ICC) and Dice scores.
- Precision errors were quantified using root-mean-square error (RMSE).
Findings:
- Excellent intra- and inter-rater reliability (ICC > 0.985) was observed across all user groups.
- FLAIR segmentations demonstrated significantly smaller precision errors and higher Dice scores compared to CE segmentations.
- Single raters achieved the lowest RMSE for FLAIR (3.3%), with both single raters and neuroradiologists showing minimal precision error in longitudinal FLAIR evaluations.
Implications:
- Semi-automated glioblastoma volumetry is reliable even for non-expert users.
- FLAIR-based segmentation offers higher reliability than CE segmentation for tumor volume assessment.
- Quantitative analysis of longitudinal FLAIR changes enables reliable detection of disease progression (<15%) earlier, potentially improving patient management.
