BRISC: Annotated Dataset for Brain Tumor Segmentation and Classification.

Area of Science:

• Medical Image Analysis
• Artificial Intelligence in Oncology
• Radiology

Background:

• Accurate brain tumor segmentation and classification from MRI are crucial but hindered by limited, unbalanced datasets.
• Existing datasets often lack expert annotations and high-resolution segmentation masks, impeding robust AI model development.

Purpose of the Study:

• To introduce the BRISC dataset, a novel, large-scale resource for brain tumor segmentation and classification.
• To address the need for high-quality, diverse, and expertly annotated brain MRI data.
• To facilitate the development and validation of advanced deep learning models for brain tumor analysis.

Main Methods:

• Collated 6,000 contrast-enhanced T1-weighted MRI scans from public sources lacking segmentation labels.
• Performed expert annotation by certified radiologists and physicians, including high-resolution segmentation masks.
• Included three major tumor types (glioma, meningioma, pituitary) and non-tumorous cases across axial, sagittal, and coronal planes.

Main Results:

• The BRISC dataset provides 6,000 expertly annotated brain MRI scans with high-resolution segmentation masks.
• Benchmark results using standard deep learning models demonstrate the dataset's utility for segmentation and classification tasks.
• The dataset facilitates robust model development and cross-view generalization due to multi-planar categorization.

Conclusions:

• The BRISC dataset represents a significant advancement in resources for brain tumor image analysis.
• Public availability of BRISC will accelerate research in AI-driven brain tumor diagnosis and treatment planning.
• This dataset supports the development of more accurate and reliable AI tools for medical imaging.