AI Science Tracker
EN

PedDx: An End-to-End Deep Learning Framework for Pediatric Posterior Fossa Tumor Diagnosis with Agentic AI–Enabled Independent External Evaluation

Affiliation not stated in the reviewed material · 29 Oct 2026 · medicine

Discovery

PedDx: An End-to-End Deep Learning Framework for Pediatric Posterior Fossa Tumor Diagnosis with Agentic AI–Enabled Independent External Evaluation

Source check: Source checked. Not independent scientific confirmation.

Finding: Results report five-class accuracy of 81.4% in the external CHOA cohort (88.7% among non-abstained cases) and 92.5% across 40 independently processed WCH cases (97.3% among non-abstained cases); WCH contained only three histologies, and the preprint also reports confidently incorrect OOD cases. Methods describe a frozen pretrained 3DINO backbone, multi-crop MRI sampling, calibrated class probabilities and confidence-based abstention; the deployed workflow also includes a quality-control agent. Testing: Methods/results distinguish discovery (n=492), replication (n=114), external CHOA (n=129) and independent WCH workflow cases (n=40), with pathology/clinical diagnoses and stated exclusions. The independent cohorts support the narrow diagnostic-performance result; small WCH sample, missing two histologies, unreviewed preprint and no evidence of improved patient outcomes limit interpretation.. Independent external evaluation of a pediatric posterior-fossa tumor MRI classifier and full workflow reports performance on separate CHOA and WCH cohorts; bounded to diagnostic classification, not clinical benefit. Source affiliation was not stated in the reviewed case checks and was unavailable in the local bibliographic cache.

Technology: Methods describe a frozen pretrained 3DINO backbone, multi-crop MRI sampling, calibrated class probabilities and confidence-based abstention; the deployed workflow also includes a quality-control agent.

Publication: Peer-reviewed paper · Test context: prospective or independent check. Catalog categories, not an evidence grade.

AI contribution
Methods describe a frozen pretrained 3DINO backbone, multi-crop MRI sampling, calibrated class probabilities and confidence-based abstention; the deployed workflow also includes a quality-control agent.
Test and result
Methods/results distinguish discovery (n=492), replication (n=114), external CHOA (n=129) and independent WCH workflow cases (n=40), with pathology/clinical diagnoses and stated exclusions. The independent cohorts support the narrow diagnostic-performance result; small WCH sample, missing two histologies, unreviewed preprint and no evidence of improved patient outcomes limit interpretation.
Limitation / still unknown
Independent external evaluation of a pediatric posterior-fossa tumor MRI classifier and full workflow reports performance on separate CHOA and WCH cohorts; bounded to diagnostic classification, not clinical benefit.
Evidence sources and annotation context

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

Evidence sources: https://www.medrxiv.org/content/10.64898/2026.10.27.26366616v1.full.pdf · https://api.biorxiv.org/details/medrxiv/2026-10-29/2026-10-29/0

Source checked · editorial details

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

Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.

Review date: 22 Nov 2026. Review scope: Primary result-bearing material for the stated method, result and evaluation, with bibliographic version information..

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 3/5 · Validation 4/5 · Novelty 3/5. 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.