Overview

Schema Version

https://atlas.rsna.org/schemas/2025-11/dataset.json

Name

Postoperative glioblastoma MRI

Link

https://doi.org/10.1148/ryai.220231

Indexing

Keywords: brain tumor segmentation, glioblastoma, postoperative, FLAIR, T1-weighted, contrast-enhanced, deep learning, segmentation quality

Content: NR, MR, OI, RS

RadLex: RID35976, RID4591, RID35806, RID11221, RID4044, RID4026, RID49531, RID10794

SNOMED: 1163375002

Organization(s)

Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital

Stephen E. and Catherine Pappas Center for Neuro-Oncology, Massachusetts General Hospital

Harvard-MIT Division of Health Sciences and Technology, Massachusetts Institute of Technology

MGH and BWH Center for Clinical Data Science

The University of Texas MD Anderson Cancer Center

Brigham and Women’s Hospital

University of Texas Southwestern Medical Center

University of Colorado Anschutz Medical Campus

Contact

Jayashree Kalpathy-Cramer (corresponding author)

Funding

Supported by NIH U01CA242879; NIH R01CA129371; K23CA169021; P41EB015896.

Ethical review

Institutional review board approval with waiver for written consent.

Comments

Secondary analysis of imaging data from two clinical trials (ClinicalTrials.gov: NCT00756106, NCT00662506) to study expert-centered evaluation of deep learning brain tumor segmentation.

Date

Published: 2023-11-22

References

[1] Hoebel KV, Bridge CP, Ahmed S, et al.. "Expert-centered Evaluation of Deep Learning Algorithms for Brain Tumor Segmentation". Radiology: Artificial Intelligence. 2024. doi:10.1148/ryai.220231. PMID: 38197800. PMCID: PMC10831514.

[2] . "ClinicalTrials.gov Identifier: NCT00756106". ClinicalTrials.gov. . Available from: https://clinicaltrials.gov/ct2/show/NCT00756106

[3] . "ClinicalTrials.gov Identifier: NCT00662506". ClinicalTrials.gov. . Available from: https://clinicaltrials.gov/ct2/show/NCT00662506

Dataset

Motivation

Assess expert-centered quality perception of deep learning brain tumor segmentations and its relationship to commonly used quantitative metrics.

Sampling

Newly diagnosed glioblastoma patients post-biopsy or partial resection with residual contrast-enhancing tumor ≥1 cm at enrollment.

Partitioning scheme

Participant-level split: all visits from each participant assigned to only one of training, validation, or test subsets.

Missing information

Image file formats, voxel resolutions, and imaging site details not reported.

Relationships between instances

Longitudinal dataset with repeated MRI visits per participant (713 visits from 54 individuals).

External data

Imaging originates from two clinical trials (NCT00756106, NCT00662506).

Confidentiality

Clinical trial imaging data; IRB-approved secondary analysis with waiver of consent.

Sensitive data

Health-related imaging data of human subjects.