Score-Based Generative Models for PET Image Reconstruction

TL;DR

This paper introduces PET-specific adaptations of score-based generative models to improve image reconstruction, addressing challenges like Poisson noise and dynamic range, validated through extensive in-silico experiments showing robustness and potential.

Contribution

It develops novel PET-specific score-based generative models, including guided reconstruction with MRI, for both 2D and 3D PET, demonstrating improved robustness and applicability.

Findings

Effective reconstruction on in-silico PET data without lesions

Robust performance on out-of-distribution data with lesions

Potential for enhanced PET imaging quality

Abstract

Score-based generative models have demonstrated highly promising results for medical image reconstruction tasks in magnetic resonance imaging or computed tomography. However, their application to Positron Emission Tomography (PET) is still largely unexplored. PET image reconstruction involves a variety of challenges, including Poisson noise with high variance and a wide dynamic range. To address these challenges, we propose several PET-specific adaptations of score-based generative models. The proposed framework is developed for both 2D and 3D PET. In addition, we provide an extension to guided reconstruction using magnetic resonance images. We validate the approach through extensive 2D and 3D $\textit{in-silico}$ experiments with a model trained on patient-realistic data without lesions, and evaluate on data without lesions as well as out-of-distribution data with lesions. This demonstrates the proposed method's robustness and significant potential for improved PET reconstruction.