AI Science Tracker
EN

A brain-connectivity model maps task-defined functional regions in tumor patients

Johns Hopkins University · 21 Aug 2021 · medicine

Discovery

A brain-connectivity model maps task-defined functional regions in tumor patients

Source check: Source checked. Not independent scientific confirmation.

Finding: A graph neural model used resting-state brain scans to estimate language and motor regions in 62 brain-tumor patients. Repeated cross-validation against task-fMRI-derived labels reported language-region AUC 0.76. The reference was task-fMRI, and the study did not establish agreement with invasive mapping or improved surgical outcomes.

Technology: Multi-task graph neural network for resting-state fMRI connectivity

Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.

AI contribution
The learned graph model combined resting-state connectivity with tumor-region information to estimate multiple functional-network labels simultaneously.
Test and result
The retrospective 62-patient tumor cohort supplied expert-confirmed task-fMRI pseudo-reference labels. Resting-state connectivity and manually identified tumor regions were used at inference. Repeated ten-fold evaluation reported parcel-level language AUC .76, finger .83, foot .78 and tongue .80, with different task denominators. Hyperparameters were selected using a separate healthy HCP dataset. An exploratory holdout trained on 57 unilateral-language patients and tested five bilateral cases.
Limitation / still unknown
Task-fMRI labels and atlas parcels are indirect references rather than invasive stimulation mapping. The bilateral holdout is only five patients. Task participation, expert thresholding and tumor segmentations constrain applicability to patients unable to perform tasks, and a single-center retrospective cohort does not establish clinical benefit. Simulated tumors in healthy HCP subjects are a separate synthetic experiment, not additional tumor patients or independent clinical confirmation.
Evidence sources and annotation context

Evidence context (descriptions follow the cited sources; test type is a catalog classification):

Evidence sources: https://pmc.ncbi.nlm.nih.gov/articles/PMC9245684/

Source checked · editorial details

The source page was checked against this entry. Independent confirmation is not assessed by this label.

Source type: paper. Paper: linked separately.

Review date: 21 Nov 2026. Review scope: Primary clinical-cohort, computational model/split Methods and clinical Results; synthetic healthy-brain experiments kept distinct..

No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.

Editorial assessment: Impact 1/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.

Open in the full catalog for filters, charts and related discoveries. Browse medicine discoveries.

Related catalog entries

Links use documented relationships where available, plus shared catalog context. A shared topic or model family does not establish a scientific dependency.