Differentiable Score-Based Likelihoods: Learning CT Motion Compensation From Clean Images

TL;DR

This paper introduces a score-based likelihood model for CT images that enables effective motion artifact correction without requiring prior examples of motion-affected scans, improving robustness and accuracy.

Contribution

The authors develop a novel likelihood-based approach using score models for motion correction in CT, eliminating the need for motion-affected training data.

Findings

Achieves comparable results to state-of-the-art methods

Effectively reduces motion artifacts in CT images

Does not require a dataset of motion-affected samples

Abstract

Motion artifacts can compromise the diagnostic value of computed tomography (CT) images. Motion correction approaches require a per-scan estimation of patient-specific motion patterns. In this work, we train a score-based model to act as a probability density estimator for clean head CT images. Given the trained model, we quantify the deviation of a given motion-affected CT image from the ideal distribution through likelihood computation. We demonstrate that the likelihood can be utilized as a surrogate metric for motion artifact severity in the CT image facilitating the application of an iterative, gradient-based motion compensation algorithm. By optimizing the underlying motion parameters to maximize likelihood, our method effectively reduces motion artifacts, bringing the image closer to the distribution of motion-free scans. Our approach achieves comparable performance to state-of-the-art methods while eliminating the need for a representative data set of motion-affected samples. This is particularly advantageous in real-world applications, where patient motion patterns may exhibit unforeseen variability, ensuring robustness without implicit assumptions about recoverable motion types.