Hydrocephalic Brain Volume Estimation from Low-Field MRI: Topologically-Enriched Cross-Modal Enhancement and Segmentation

Abstract

Objective Accurate volumetric analysis of brain and cerebrospinal fluid (CSF) is essential for monitoring hydrocephalus, a significant pediatric neurological condition. While computed tomography (CT) provides high-quality volumetric assessment, the associated ionizing radiation poses risks, especially for children. Low-field magnetic resonance imaging (LF-MRI) offers a safer, more accessible alternative, particularly in resource-constrained settings. However, its lower resolution and an increased likelihood for structural distortions complicate accurate segmentation. This study aims to demonstrate that reliable volumetric measurements can be obtained from LF-MRI, comparable to CT, enabling safer and more frequent monitoring of hydrocephalic infants. Approach We propose EnSegNet-Cross , a cross-modality guided enhancement-aware segmentation network for brain volume analysis using LF-MRI. The framework relies on high-fidelity CT data during training but requires only LF-MRI at inference. At the heart of this innovation, lies a novel cross modal topological penalty to minimize discrepancies between predicted LF-MRI and CT structures. A central contribution is the integration of a 3D topological loss based on persistent homology, which penalizes topological discrepancies in CSF regions (specifically CSF holes formed by enclosed brain parenchyma) between CT and LF-MRI segmentations. This embedding of structural priors facilitates generalization across heterogeneous clinical cases while obviating the need for CT data at inference time, leading to more anatomically coherent and topologically faithful segmentations. Main Results On a curated cohort of hydrocephalic infants with paired LF-MRI and CT scans, including infectious and non-infectious causes, EnSegNet-Cross consistently outperformed state-of-the-art machine learning alternatives by achieving the highest Dice Score (0.8532 ± 0.03) and Volume Score (0.9318 ± 0.03), and robustly handled challenging cases with confounding factors (Dice 0.8340 ± 0.03, Volume 0.9111 ± 0.05). Leveraging CT-derived topological priors allowed EnSegNet-Cross to succeed in anatomically complex scenarios where conventional models fail. Significance EnSegNet-Cross offers a reliable, interpretable solution for brain–CSF segmentation, as demonstrated in complex hydrocephalus cases. This study demonstrates that high-fidelity volumetric estimates can be achieved using only LF-MRI, facilitating frequent, radiation-free monitoring. By bridging the fidelity gap between low-quality LF-MRI and high-resolution CT through clinically grounded enhancement and topological supervision, EnSegNet-Cross provides a robust clinical tool for brain volumetric analysis in hydrocephalus infants using LF-MRI.