Overview

Schema Version

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

Name

Pediatric low-grade glioma (pLGG) MRI-genomics cohorts for BRAF subtype classification

Link

https://pmc.ncbi.nlm.nih.gov/articles/PMC11140508

Indexing

Keywords: pediatric low-grade glioma, BRAF, BRAF fusion, BRAF V600E, T2-weighted MRI, brain MRI, molecular subtyping, deep learning, transfer learning, self-supervised learning

Content: MR, NR, OI, PD

RadLex: RID10312, RID10796

Organization(s)

Dana-Farber/Boston Children’s Hospital (DF/BCH)

Children’s Brain Tumor Network (CBTN)

Contact

Corresponding author email provided in article: ude.dravrah.icfd@nnaK_nimajneB

Funding

National Institutes of Health (U24CA194354, U01CA190234, U01CA209414, R35CA22052, U54CA274516, K08DE030216), NCI SPORE (2P50CA165962), European Union – European Research Council (866504), RSNA (RSCH2017), Pediatric Low-Grade Astrocytoma Program at the Pediatric Brain Tumor Foundation, William M. Wood Foundation, Botha-Chan Low Grade Glioma Consortium.

Ethical review

IRB-approved at DF/BCH and CBTN with waiver of informed consent due to use of public datasets and retrospective design.

Comments

Two cohorts were used: a single-institution DF/BCH development cohort (n=214) and a publicly available CBTN external cohort (n=112), each with pretreatment T2-weighted brain MRI and linked BRAF mutational status (wild type, fusion, V600E).

Date

Published: 2024-03-06

References

[1] Tak D, Ye Z, Zapaischykova A, et al.. "Noninvasive Molecular Subtyping of Pediatric Low-Grade Glioma with Self-Supervised Transfer Learning". Radiology: Artificial Intelligence. 2024-05-01. doi:10.1148/ryai.230333. PMID: 38446044. PMCID: PMC11140508.

Dataset

Motivation

Enable noninvasive MRI-based classification of BRAF mutational status (wild type, fusion, V600E) in pediatric low-grade glioma when tissue diagnosis is infeasible.

Sampling

Included all eligible patients meeting criteria (age 1–25 years; WHO grade I–II glioma; pretreatment T2-weighted MRI; known BRAF status) within the stated time frames.

Partitioning scheme

Development dataset (DF/BCH) with a 25% randomly selected internal test set; external testing on CBTN cohort.

Missing information

Image file formats, scanner vendors, and detailed acquisition parameters are not fully specified in the main text (details in appendices).

Relationships between instances

Each patient scan was sectioned into multiple axial tumor images for model training and then aggregated to patient-level predictions by averaging probabilities.

Noise

Heterogeneous MRI parameters across institutions may affect performance.

External data

External testing used the publicly available CBTN pLGG cohort meeting inclusion criteria.

Confidentiality

Patient imaging and genomic data; IRB-approved with waiver of consent.

Sensitive data

Pediatric health and imaging data with linked genomic information.