Importance: Perinatal hypoxic-ischemic encephalopathy (HIE) is one of the most common causes of neonatal death and neurodevelopmental impairment worldwide. Accurate prognostication of developmental outcomes following perinatal HIE is an important component of family-centered and evidence-based care. Objective: To utilize magnetic resonance imaging (MRI)-based radiomic measures together with machine learning to produce automated and objective predictions of developmental outcomes after perinatal HIE. Design: This was a retrospective cohort study of infants born between January 2018 and January 2022 with HIE. Setting: The data for this study were acquired at the neonatal neurocritical care unit of a quaternary care center based on the center's institutional criteria for diagnosis and for the use of therapeutic hypothermia. Participants: Neonates with a gestational age of >= 35 weeks and a diagnosis of neonatal encephalopathy. Exposure(s): Therapeutic hypothermia, with a whole-body cooling system, was begun within 6 hours after birth and was continued for 72 hours. Main Outcome(s) and Measure(s): Brain MRI data were acquired on postnatal day 4-5, after rewarming after completion of therapeutic hypothermia. At 18-months of age, developmental outcome measures were assessed with the Bayley Scales of Infant and Toddler Development. We extracted radiomic measures from the deep-gray matter structures and from 2224 cubic tiles across the entire brain, in multiple modalities, and provided these measures to an elastic-net penalized linear regression model to predict the 18-month developmental outcomes. Results: MRI-based radiomic measures from 160 neonates were used in a 10-fold cross-validation framework to predict the 18-month Bayley outcome scores. Across cognitive, language, and motor domains, the mean correlation between the predicted outcomes and the observed outcomes was 0.947, and the mean coefficient of determination was 0.879. Conclusions and Relevance: A machine learning model using MRI-based radiomic measures from infants with HIE can reliably predict their 18-month developmental outcomes with excellent accuracy across the full range of motor, cognitive, and language domains. In addition, our approach allowed us to map the predictor weightings into neuroanatomical space, producing atlases of the brain regions responsible for the developmental impairments; these may prove useful in the search for novel interventions.

The authors have declared no competing interest.

This research was enabled in part by support provided by the TD Bank Group Charity Classic Golf Tournament.

I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.

Yes

The details of the IRB/oversight body that provided approval or exemption for the research described are given below:

This retrospective cohort study was conducted at the neonatal neurocritical care unit of the Hospital for Sick Children in Toronto, Canada. The study protocols (REB:1000064940, 1000079302) were reviewed and approved by the Institutional Research Ethics Board of The Hospital for Sick Children in Toronto, Canada, and informed consent was waived.

I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals.

Yes

I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).

Yes

I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable.

Yes

Our multi-contrast population-specific neonatal brain MRI template and the labels for the cubic tiling, the deep gray-matter structures, and the PLIC, can be found here: https://gin.g-node.org/johndlewis/HIE3/Template/ ; the scripts used to process the data can be found here: https://gin.g-node.org/johndlewis/HIE3/Tools ; and the linear regression model can be found here: https://gin.g-node.org/johndlewis/ HIE3/Models/ .