Optimal MRI Undersampling Patterns for Ultimate Benefit of Medical Vision Tasks

TL;DR

This paper introduces a novel approach to MRI undersampling that optimizes sampling patterns based on downstream medical vision task performance rather than image quality, leading to improved diagnostic outcomes.

Contribution

It proposes a new method to find optimal undersampling patterns tailored to specific medical analysis tasks, improving efficiency and accuracy in MRI acceleration.

Findings

Up to 12% improvement in Dice score for segmentation at 16x acceleration.

Validated on three medical datasets with significant performance gains.

Introduced a universal iterative gradient sampling routine for various tasks.

Abstract

To accelerate MRI, the field of compressed sensing is traditionally concerned with optimizing the image quality after a partial undersampling of the measurable $\textit{k}$-space. In our work, we propose to change the focus from the quality of the reconstructed image to the quality of the downstream image analysis outcome. Specifically, we propose to optimize the patterns according to how well a sought-after pathology could be detected or localized in the reconstructed images. We find the optimal undersampling patterns in $\textit{k}$-space that maximize target value functions of interest in commonplace medical vision problems (reconstruction, segmentation, and classification) and propose a new iterative gradient sampling routine universally suitable for these tasks. We validate the proposed MRI acceleration paradigm on three classical medical datasets, demonstrating a noticeable improvement of the target metrics at the high acceleration factors (for the segmentation problem at $\times$16 acceleration, we report up to 12% improvement in Dice score over the other undersampling patterns).