A multi-institutional meningioma MRI dataset for automated multi-sequence image segmentation

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Oncology

Background:

• Meningiomas are common primary brain tumors causing significant morbidity and mortality.
• Accurate diagnosis and monitoring rely heavily on multi-sequence brain MRI.
• Current limitations exist in automated, objective, and quantitative tools for meningioma assessment.

Purpose of the Study:

• To introduce the BraTS Pre-operative Meningioma Dataset, the largest multi-institutional expert-annotated dataset for meningioma research.
• To facilitate the development of automated computational methods for meningioma segmentation.
• To expedite the clinical integration of these automated tools for improved patient management.

Main Methods:

• Compiled 1,141 multi-sequence MR images from six institutions.
• Included four structural MRI sequences: T2, T2/FLAIR, pre-contrast T1, and post-contrast T1-weighted.
• Provided expert-annotated segmentations for enhancing tumor, non-enhancing tumor, and surrounding non-enhancing T2/FLAIR hyperintensity, along with demographic data and CNS WHO grade.

Main Results:

• Established the largest multi-institutional, expert-annotated, multi-sequence meningioma MRI dataset to date.
• The dataset comprises 1,141 cases with detailed segmentation of tumor sub-compartments.
• Includes demographic and clinical data to support comprehensive research.

Conclusions:

• The BraTS Pre-operative Meningioma Dataset is a valuable resource for advancing automated meningioma analysis.
• Its release aims to spur innovation in computational methods for segmentation.
• Ultimately, the goal is to enhance the clinical care of patients with meningiomas.