Overview

Schema Version

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

Name

HEAL trial neonatal MRI cohort used in AI-OPiNE study

Link

https://clinicaltrials.gov/study/NCT02811263

Indexing

Keywords: HEAL trial, neonatal MRI, hypoxic-ischemic encephalopathy, outcome prediction, diffusion tensor imaging, apparent diffusion coefficient, Bayley Scales, therapeutic hypothermia, erythropoietin

Content: NR, PD, MR

RadLex: RID12698, RID10312, RID5056, RID38778

SNOMED: 419620001, 128188000, 95628005

Author(s)

Christopher O. Lew

Evan Calabrese

Joshua V. Chen

Felicia Tang

Gunvant Chaudhari

Amanda Lee

John Faro

Sandra Juul

Amit Mathur

Robert C. McKinstry

Jessica L. Wisnowski

Andreas Rauschecker

Yvonne W. Wu

Yi Li

Organization(s)

Duke University Medical Center

University of California San Francisco

University of Washington

Saint Louis University

Washington University School of Medicine, Mallinckrodt Institute of Radiology

Children’s Hospital Los Angeles, University of Southern California

Contact

Corresponding author: Evan Calabrese, Department of Radiology, Duke University Medical Center, Durham, NC, USA.

Funding

NIH/NINDS and NICHD support acknowledged for HEAL-related work (e.g., U01NS092764, U01NS092553); authors declared no additional funding for this secondary analysis.

Ethical review

Duke Health IRB–approved retrospective analysis; written informed parental consent obtained in HEAL; HIPAA compliant; images de-identified at each site.

Comments

Secondary analysis of imaging and outcomes from the HEAL randomized clinical trial across 17 U.S. institutions; harmonized neonatal brain MRI protocol including T1-weighted, T2-weighted, and diffusion tensor imaging acquired within the first week after birth (typically 4–6 days).

Date

Published: 2024-07-10

References

[1] Lew CO, Calabrese E, Chen JV, et al.. "Artificial Intelligence Outcome Prediction in Neonates with Encephalopathy (AI-OPiNE)". Radiology: Artificial Intelligence. 2024-09-01. doi:10.1148/ryai.240076. PMID: 38984984. PMCID: PMC11427921.

[2] Wu YW, Comstock BA, Gonzalez FF, et al.. "Trial of Erythropoietin for Hypoxic–Ischemic Encephalopathy in Newborns". N Engl J Med. 2022-01-01. PMID: 35830641. PMCID: PMC10542745. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10542745/

Dataset

Motivation

Develop and evaluate deep learning models using neonatal brain MRI plus basic clinical variables to predict 2-year neurodevelopmental outcomes (death or any NDI).

Sampling

Prospective HEAL enrollment of term neonates (≥36 weeks) with moderate to severe encephalopathy at 17 U.S. sites between Jan 25, 2017 and Oct 9, 2019; this secondary analysis included 414 participants after excluding incomplete MRI or follow-up.

Partitioning scheme

Participants from 17 institutions split into discovery (15 institutions) and two test sets: in-distribution (from same 15 institutions) and out-of-distribution (all cases from 2 institutions). Discovery further split into training and validation (80/20).

Missing information

Radiology AI paper does not provide public dataset access details or full demographic breakdown by race/ethnicity.

Relationships between instances

Each participant has a multisequence MRI exam (T1-weighted, T2-weighted, DTI-derived trace and ADC) acquired within the first week of life; tabular clinical variables linked per participant.

Noise

Multisite data with real-world variation in vendors/platforms; harmonized MRI protocol used to minimize heterogeneity.

External data

Data analyzed were provided by a third party (HEAL Consortium). Requests for data should be directed to the provider acknowledged in the HEAL publications.

Confidentiality

De-identified imaging data per site; HIPAA compliant.

Re-identification

Low risk; images were de-identified at each site before transfer.

Sensitive data

Yes—pediatric health data from a clinical trial; de-identified for analysis.