Generative-Model-Based Fully 3D PET Image Reconstruction by Conditional Diffusion Sampling
This work addresses low-dose or short-duration PET scanning for medical imaging, representing an incremental advance by applying an existing generative model to real 3D PET data for the first time.
The authors tackled 3D PET image reconstruction at very low counts (1% of original) using a score-based generative model (SGM), showing that their SGM-based reconstructions match full-dose reconstructions more closely and have lower variance than conventional OSEM and MAP-EM baselines.
Score-based generative models (SGMs) have recently shown promising results for image reconstruction on simulated positron emission tomography (PET) datasets. In this work we have developed and implemented practical methodology for 3D image reconstruction with SGMs, and perform (to our knowledge) the first SGM-based reconstruction of real fully 3D PET data. We train an SGM on full-count reference brain images, and extend methodology to allow SGM-based reconstructions at very low counts (1% of original, to simulate low-dose or short-duration scanning). We then perform reconstructions for multiple independent realisations of 1% count data, allowing us to analyse the bias and variance characteristics of the method. We sample from the learned posterior distribution of the generative algorithm to calculate uncertainty images for our reconstructions. We evaluate the method's performance on real full- and low-count PET data and compare with conventional OSEM and MAP-EM baselines, showing that our SGM-based low-count reconstructions match full-dose reconstructions more closely and in a bias-variance trade-off comparison, our SGM-reconstructed images have lower variance than existing baselines. Future work will compare to supervised deep-learned methods, with other avenues for investigation including how data conditioning affects the SGM's posterior distribution and the algorithm's performance with different tracers.