Search papers, labs, and topics across Lattice.
This paper tackles the challenge of MRI reconstruction from highly undersampled data by framing it as autoregressive next-acceleration-scale prediction in a discrete latent space. They leverage discrete priors and introduce on-policy privileged information distillation, where a teacher model trained on fully sampled data guides a student model trained on its own rollouts. Experiments on the fastMRI benchmark demonstrate improved reconstruction performance across diverse sampling patterns under extreme undersampling.
Achieve sharper MRI reconstructions from extremely sparse measurements by recasting the problem as discrete autoregressive prediction, sidestepping the blurring inherent in continuous pixel-domain approaches.
MRI reconstruction is an inherently ill-posed inverse problem, since incomplete measurements admit many plausible solutions. This ambiguity becomes more severe under high acceleration, where pixel-domain continuous predictors tend to average over feasible reconstructions and suppress high-frequency anatomy. We address this limitation by moving reconstruction to discrete multi-scale latent space and posing it as autoregressive next-acceleration-scale prediction. Leveraging discrete priors proven effective in visual autoregressive modeling, our method restricts the solution to compact sequences of codebook tokens, enabling sharp reconstructions even from extremely sparse measurements. This discrete autoregressive formulation also aligns naturally with modern large language model post-training techniques. Building on this observation, we introduce on-policy privileged information distillation for visual autoregressive modeling, where a teacher is provided training only privileged context that is unavailable at inference, in our case fully sampled acquisitions, and supervises a student trained on its own rollouts, leading to consistent reconstruction gains. Through extensive experiments on the fastMRI benchmark, we show that our approach delivers improved reconstruction performance across diverse sampling patterns under extreme undersampling. Project website is https://yilmazkorkmaz1.github.io/discrete-mri-reconstruction-opd/{here}.
