Hierarchical detection of neonatal brain pathologies using a two-stage deep learning framework from magnetic resonance imaging
- 1African Institute for Mathematical Sciences, Accra, Ghana
- 2Department of Computer Science, AGH University of Kraków, Kraków, Poland
Abstract
Automated neonatal brain pathology detection from magnetic resonance imaging (MRI) faces a dual challenge: preterm birth produces a characteristic but confounding pattern of delayed myelination that can obscure concurrent structural abnormalities, while conditions such as hypoxic-ischaemic encephalopathy (HIE) and hydrocephalus demand timely identification in their own right. Existing frameworks typically address these clinical questions in isolation or collapse them into a single flat classifier, discarding the natural clinical decision hierarchy. This study presents a hierarchical two-stage deep learning pipeline that emulates clinical reasoning from paired T1-weighted and T2-weighted neonatal brain MRI. An initial Stage-1 deploys a brain-age gap classifier, built on a 3-D ResNet-18 regressor followed by a gap-thresholded binary decision intended to screen for myelination delay. Then the same subjects advance to a Stage-2, where a second 3-D ResNet-18, trained for structural pathology, performs three-class classification: normal, HIE, and hydrocephalus. Experiments demonstrate that a regression from MRI scan to age with an R2 of 0.941 at Stage-1, the proposed hierarchical pipeline achieves a Stage-2 area under the curve (AUC) of 0.919 ± 0.024.
Related
- Extracellular Vesicles and Choline Metabolism in Neonatal Bronchopulmonary Dysplasia: A Systems-Pharmacology Framework for Regenerative Therapy
- Cardiovascular Magnetic Resonance for Early Detection of Cancer Therapy-Related Cardiac Dysfunction: From Multiparametric Tissue Signatures to Clinical Pathways
- HAFI-DTA: an interpretable multimodal framework for drug-target affinity prediction based on hierarchical attention guidance and feature iterative interaction
- Second Ray Pathologies Associated with Hallux Valgus: A Radiographic Comparative Study
- Comparative Analysis of Manual and Deep LIIF Extended Global Scoring Method for Ki-67 Across a Spectrum of Oral and Maxillofacial Pathologies: An Exploratory Study
BibTeX
@article{ampiah2026hierarchical,
title = {{Hierarchical detection of neonatal brain pathologies using a two-stage deep learning framework from magnetic resonance imaging}},
author = {Christian Ahwoi Ampiah and Tina Malalanirainy and Alessandro Crimi},
journal = {Frontiers in Neuroimaging},
year = {2026},
month = oct,
doi = {10.3389/fnimg.2026.1981436},
url = {https://doi.org/10.3389/fnimg.2026.1981436},
note = {CC BY},
}
RIS
TY - JOUR
TI - Hierarchical detection of neonatal brain pathologies using a two-stage deep learning framework from magnetic resonance imaging
AU - Christian Ahwoi Ampiah
AU - Tina Malalanirainy
AU - Alessandro Crimi
PY - 2026
DA - 2026/10/30
JO - Frontiers in Neuroimaging
DO - 10.3389/fnimg.2026.1981436
UR - https://doi.org/10.3389/fnimg.2026.1981436
AB - Automated neonatal brain pathology detection from magnetic resonance imaging (MRI) faces a dual challenge: preterm birth produces a characteristic but confounding pattern of delayed myelination that can obscure concurrent structural abnormalities, while conditions such as hypoxic-ischaemic encephalopathy (HIE) and hydrocephalus demand timely identification in their own right. Existing frameworks typically address these clinical questions in isolation or collapse them into a single flat classifier, discarding the natural clinical decision hierarchy. This study presents a hierarchical two-stage deep learning pipeline that emulates clinical reasoning from paired T1-weighted and T2-weighted neonatal brain MRI. An initial Stage-1 deploys a brain-age gap classifier, built on a 3-D ResNet-18 regressor followed by a gap-thresholded binary decision intended to screen for myelination delay. Then the same subjects advance to a Stage-2, where a second 3-D ResNet-18, trained for structural pathology, performs three-class classification: normal, HIE, and hydrocephalus. Experiments demonstrate that a regression from MRI scan to age with an R2 of 0.941 at Stage-1, the proposed hierarchical pipeline achieves a Stage-2 area under the curve (AUC) of 0.919 ± 0.024.
SN - 2813-1193
ER -