Adaptive Diffusion Priors for Accelerated MRI Reconstruction
Alper Güngör Address: Department of Electrical and Electronics Engineering, Bilkent University, Ankara 06800, Turkey Address: National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara 06800, Turkey Address: ASELSAN Research Center, Ankara 06200, Turkey Address: denotes equal contribution Salman UH Dar Address: Department of Electrical and Electronics Engineering, Bilkent University, Ankara 06800, Turkey Address: National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara 06800, Turkey Address: denotes equal contribution Şaban Öztürk Address: Department of Electrical and Electronics Engineering, Bilkent University, Ankara 06800, Turkey Address: National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara 06800, Turkey Address: Department of Electrical and Electronics Engineering, Amasya University, Amasya 05100, Turkey Address: denotes equal contribution Yilmaz Korkmaz Address: Department of Electrical and Electronics Engineering, Bilkent University, Ankara 06800, Turkey Address: National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara 06800, Turkey Hasan A Bedel Address: Department of Electrical and Electronics Engineering, Bilkent University, Ankara 06800, Turkey Address: National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara 06800, Turkey Gokberk Elmas Address: Department of Electrical and Electronics Engineering, Bilkent University, Ankara 06800, Turkey Address: National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara 06800, Turkey Muzaffer Ozbey Address: Department of Electrical and Electronics Engineering, Bilkent University, Ankara 06800, Turkey Address: National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara 06800, Turkey Tolga Çukur Corresponding author: Corresponding author, e-mail: Address: Department of Electrical and Electronics Engineering, Bilkent University, Ankara 06800, Turkey Address: National Magnetic Resonance Research Center (UMRAM), Bilkent University, Ankara 06800, Turkey Address: Neuroscience Program, Bilkent University, Ankara 06800, Turkey
Abstract
Deep MRI reconstruction is commonly performed with conditional models that de-alias undersampled acquisitions to recover images consistent with fully-sampled data. Since conditional models are trained with knowledge of the imaging operator, they can show poor generalization across variable operators. Unconditional models instead learn generative image priors decoupled from the operator to improve reliability against domain shifts related to the imaging operator. Recent diffusion models are particularly promising given their high sample fidelity. Nevertheless, inference with a static image prior can perform suboptimally. Here we propose the first adaptive diffusion prior for MRI reconstruction, AdaDiff, to improve performance and reliability against domain shifts. AdaDiff leverages an efficient diffusion prior trained via adversarial mapping over large reverse diffusion steps. A two-phase reconstruction is executed following training: a rapid-diffusion phase that produces an initial reconstruction with the trained prior, and an adaptation phase that further refines the result by updating the prior to minimize data-consistency loss. Demonstrations on multi-contrast brain MRI clearly
原文 arXiv:2207.05876;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2207.05876v3