Ultra low field mri improved for better pediatric brain scans
A Dual-Stream Regulated Reconstruction and Segmentation Network with Hierarchical Artifact-Prior Modeling for Ultra-Low-Field Pediatric Neuroimaging
Computer Vision and Pattern Recognition
Summary
Low-strength MRI scans of children's brains often produce blurry images with weak boundaries and artifacts, making it hard to assess and segment brain structures. The authors designed a combined computer model that enhances these MRI images, identifies different brain parts, and detects image artifacts all in one go. They use clever techniques like two linked processing streams and artifact-aware features to get clearer images and better segmentations. Their method also deals with limited labeled data by using brain atlases to improve accuracy.
What this means in practice
- •For medical imaging technicians: Use the dual-stream network to enhance low-field pediatric MRI scans for clearer brain images and reliable anatomical segmentation in clinical settings.
- •For neuroimaging developers: Integrate artifact-aware reconstruction and segmentation into imaging software to improve automated quality control and analysis of ultra-low-field MRI data.
Authors
Bahram Jafrasteh, Leo Milecki, Qingyu Zhao
Abstract
Automated quality assessment, enhancement, and segmentation of multiple structures in $0.064\,\mathrm{T}$ ultra-low-field pediatric MRI are limited by a low signal-to-noise ratio, weak anatomical boundaries, and frequent artifacts. We present a unified framework for the LISA 2026 Challenge that performs all three tasks together within one inference pipeline. A network with two coupled streams, built on a 3D U-Net, first reconstructs an enhanced uLF volume and then combines the original and enhanced images for subcortical segmentation. To improve boundary stability, we add an auxiliary class covering brain tissue outside the target structures, derived from whole brain masks. A head conditioned on an artifact graph predicts the seven artifact ratings from reconstruction residuals and frozen segmentation features. We address the scarcity of dense annotations using diffeomorphic registration from atlas to target for label propagation and to regularize anatomical reconstruction. We report validation results across all three tasks.