Likelihood-Scheduled Score-Based Generative Modeling for Fully 3D PET Image Reconstruction
This addresses practical bottlenecks in medical imaging for PET reconstruction, offering faster and more robust 3D image generation, though it is incremental as it builds on existing SGM approaches.
The paper tackles slow reconstruction, hyperparameter tuning, and slice inconsistency in 3D PET image reconstruction using score-based generative models (SGMs), proposing a method that matches the SGM's reverse diffusion likelihood to maximum-likelihood expectation maximization. Results show matching or improved NRMSE and SSIM over state-of-the-art SGM methods while reducing reconstruction time and hyperparameter needs, with a first-ever implementation on real 3D PET data.
Medical image reconstruction with pre-trained score-based generative models (SGMs) has advantages over other existing state-of-the-art deep-learned reconstruction methods, including improved resilience to different scanner setups and advanced image distribution modeling. SGM-based reconstruction has recently been applied to simulated positron emission tomography (PET) datasets, showing improved contrast recovery for out-of-distribution lesions relative to the state-of-the-art. However, existing methods for SGM-based reconstruction from PET data suffer from slow reconstruction, burdensome hyperparameter tuning and slice inconsistency effects (in 3D). In this work, we propose a practical methodology for fully 3D reconstruction that accelerates reconstruction and reduces the number of critical hyperparameters by matching the likelihood of an SGM's reverse diffusion process to a current iterate of the maximum-likelihood expectation maximization algorithm. Using the example of low-count reconstruction from simulated [$^{18}$F]DPA-714 datasets, we show our methodology can match or improve on the NRMSE and SSIM of existing state-of-the-art SGM-based PET reconstruction while reducing reconstruction time and the need for hyperparameter tuning. We evaluate our methodology against state-of-the-art supervised and conventional reconstruction algorithms. Finally, we demonstrate a first-ever implementation of SGM-based reconstruction for real 3D PET data, specifically [$^{18}$F]DPA-714 data, where we integrate perpendicular pre-trained SGMs to eliminate slice inconsistency issues.