What question did this study set out to answer?

This research aims to provide a comprehensive dataset to facilitate the understanding of gliomas and their molecular profiles using MRI.

April 24, 2026Open Access

The University of Texas Southwestern Glioma Dataset - MRI, Molecular Markers and Segmentations

Key Points

This research aims to provide a comprehensive dataset to facilitate the understanding of gliomas and their molecular profiles using MRI.
Developed a curated dataset of MRI images, molecular markers, and tumor segmentations from 625 glioma patients.
Included multi-contrast MRI and genetic information like IDH status and MGMT methylation.
Dataset supports deep learning model validation and exploration of MRI genetics relationships.
The dataset offers MRI contrasts and genetic profiles for over 600 glioma patients treated between 2006 and 2023.
It provides essential resources for non-invasive predictions of molecular markers.
Serves as a benchmark for developing deep learning applications in tumor analysis.

Abstract

Abstract Gliomas are the most common type of primary brain tumors. Their management options and outcomes depend significantly on the underlying molecular-marker profile. Traditionally, molecular markers are determined through pathological testing on a tissue specimen acquired through biopsy. Several Magnetic Resonance Imaging (MRI) based Deep Learning (DL) methods offer a promising, non-invasive approach to predict these markers. However, they often require high-quality, well-annotated datasets. To support this need, we present a well-curated brain tumor dataset developed at The University of Texas Southwestern (UTSW) Medical Center. This dataset includes multi-contrast-MRI, demographics, molecular-markers, and multi-label tumor segmentations for 625 patients treated at UTSW between 2006 and 2023. Each patient record contains four MRI contrasts: pre-contrast-T1w, post-contrast-T1w, T2w, and T2-weighted fluid-attenuated inversion recovery (T2w-FLAIR) images. The dataset also provides comprehensive genetic information, including IDH mutation-status, 1p19q co-deletion, MGMT promoter methylation, tumor-type, and tumor-grade. This dataset offers a valuable resource for exploring the relationship between MRI characteristics and tumor genetics. It also serves as a robust benchmark for developing and validating DL models for various downstream tasks.

KI fragen

Bookmark

View Full Paper