ECLARE: Efficient cross-planar learning for anisotropic resolution enhancement

2025-03-14

AI Summary

Anisotropic 2D clinical MR acquisitions—characterized by thick slice thickness and inter-slice gaps—degrade the performance of 3D analysis. To address this, we propose a self-supervised super-resolution framework that requires no external data, uniquely integrating slice profile estimation, inter-slice gap compensation, in-domain adaptive learning, and arbitrary non-integer upsampling—thereby eliminating train-test domain shift. Our method leverages cross-plane self-supervision within the same volumetric scan, incorporating differentiable slice profile modeling, anti-aliasing upsampling networks, and B-spline prior constraints. On simulated data, it significantly improves signal reconstruction fidelity (PSNR/SSIM) and enhances downstream segmentation and registration accuracy. On real clinical data, it achieves superior visual quality compared to all baselines—including SMORE and conventional interpolation methods—demonstrating robust generalizability and practical utility for clinical 3D reconstruction.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Computer Vision: Segmentation
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

In clinical imaging, magnetic resonance (MR) image volumes are often acquired as stacks of 2D slices, permitting decreased scan times, improved signal-to-noise ratio, and image contrasts unique to 2D MR pulse sequences. While this is sufficient for clinical evaluation, automated algorithms designed for 3D analysis perform sub-optimally on 2D-acquired scans, especially those with thick slices and gaps between slices. Super-resolution (SR) methods aim to address this problem, but previous methods do not address all of the following: slice profile shape estimation, slice gap, domain shift, and non-integer / arbitrary upsampling factors. In this paper, we propose ECLARE (Efficient Cross-planar Learning for Anisotropic Resolution Enhancement), a self-SR method that addresses each of these factors. ECLARE estimates the slice profile from the 2D-acquired multi-slice MR volume, trains a network to learn the mapping from low-resolution to high-resolution in-plane patches from the same volume, and performs SR with anti-aliasing. We compared ECLARE to cubic B-spline interpolation, SMORE, and other contemporary SR methods. We used realistic and representative simulations so that quantitative performance against a ground truth could be computed, and ECLARE outperformed all other methods in both signal recovery and downstream tasks. On real data for which there is no ground truth, ECLARE demonstrated qualitative superiority over other methods as well. Importantly, as ECLARE does not use external training data it cannot suffer from domain shift between training and testing. Our code is open-source and available at https://www.github.com/sremedios/eclare.

Problem

• Enhances anisotropic resolution in 2D-acquired MR images.
• Addresses slice profile, gap, domain shift, and upsampling issues.
• Improves signal recovery and downstream task performance.

Innovation

• Self-supervised SR method for MR images
• Estimates slice profile and performs anti-aliasing
• No external training data, avoids domain shift