Self-Auditing Residual Drifting for Pathology-Preserving Accelerated Knee MRI

2026-07-02

AI Summary

Accelerated knee MRI reconstruction often blurs subtle lesions or introduces imperceptible artifacts, compromising diagnostic reliability. This work proposes SA-RDM-DC, the first method to integrate a generative residual drift paradigm into accelerated MRI: starting from zero-filled reconstruction, it learns a physically constrained drift field to correct residuals between the initial image and missing k-space data, simultaneously producing a reconstructed image, a dense error map, and a slice-level risk score in a single inference pass. By combining frequency-aware supervision, data consistency optimization, and multi-coil k-space modeling, the method achieves superior SSIM over existing baselines on the fastMRI knee dataset while maintaining sub-second inference speed. It effectively preserves critical structures such as menisci and reduces prediction instability. Its built-in self-auditing mechanism accurately flags high-error samples and demonstrates robust cross-protocol generalization on the SKM-TEA dataset.

Abstract

Accelerated magnetic resonance imaging reduces acquisition time, but reconstruction from undersampled k-space can blur diagnostically relevant structures or introduce failures that are not captured by global image metrics. We propose SA-RDM-DC, a Self-Auditing Residual generative Drifting Model with Data Consistency for accelerated knee MRI. The method adapts the newly proposed generative drifting paradigm to accelerated MRI by training a physics-conditioned drift field from the zero-filled reconstruction toward the fully sampled residual correction. It predicts image- and missing-k-space residual corrections, enforces data consistency with acquired k-space, uses frequency-aware and residual drifting supervision to recover fine detail, and produces dense error maps and slice-level risk scores in the same inference pass. We evaluate SA-RDM-DC on multi-coil fastMRI knee data at acceleration factors of 4, 8, and 12, with fastMRI+ pathology annotations for region-level and classifier-based task preservation, and on SKM-TEA for zero-shot and fine-tuned protocol-shift evaluation. Compared with zero-filled reconstruction, UNet-image-SENSE, DC-UNet, Score-Diffusion, ELF-Diff, SENSE-VarNet, and MoDL baselines, SA-RDM-DC achieves the highest SSIM across fastMRI acceleration factors while retaining subsecond per-slice inference and avoiding the long sampling time of iterative diffusion baselines. In pathology-aware analysis, SA-RDM-DC preserves lesion-region structural fidelity and reduces meniscus prediction instability. Its self-auditing scores strongly identify high-error reconstructions on fastMRI and partially transfer as a selective-review signal under SKM-TEA protocol shift. These results support reconstruction evaluation that jointly considers image fidelity, pathology preservation, runtime, and case-specific reliability.

Problem

• accelerated MRI
• pathology preservation
• reconstruction error
• k-space undersampling
• diagnostic fidelity

Innovation

• generative drifting
• self-auditing
• data consistency
• accelerated MRI
• residual correction