Fast method finds best MRI image from noisy data quickly
D-MOD: Diffusion mode estimation for accelerated MRI
Machine Learning
Summary
MRI scans often take a long time to reconstruct high-quality images because current methods generate many random guesses and average them. The authors created D-MOD, a new approach that finds the most likely image directly, saving a lot of time. It uses a smart algorithm that gradually reduces noise and combines data to zoom in on the best final picture. This means doctors can get reliable MRI images much faster without losing accuracy.
What this means in practice
- •For medical imaging developers: Produce faster MRI image reconstructions by directly estimating the most probable image rather than averaging multiple noisy samples.
- •For radiology department it teams: Integrate a deterministic algorithm that accelerates MRI post-processing, reducing exam turnaround times with maintained image quality.
Authors
Md Kawsar Ahmed, Mathews Jacob
Abstract
Diffusion models are powerful priors for accelerated MRI reconstruction, but most current methods are posterior samplers, which return a random draw from the posterior. A reproducible and high-quality reconstruction often requires averaging many samples, resulting in a high computational cost. We introduce D-MOD, a fast algorithm that instead provides a posterior mode. It uses a minorize-maximize algorithm on a lower bound of the smoothed log-posterior, which provably increases the objective. Each iteration draws S noisy samples of the current estimate at the current noise level, denoises them with a pretrained diffusion model, and combines their average with the measured k-space through a conjugate-gradient (CG) solve. We anneal the noise levels, as the distance between the bound and its target decays quadratically with noise. On fastMRI brain (T1, T2, FLAIR), a single D-MOD reconstruction exceeds, in PSNR, the 10-sample finite-MMSE estimates of DPS, DAPS and SNORE-P and stays within 0.31 dB of the strongest baselines (DiffPIR, DDS). This approach thus provides a fast deterministic reconstruction algorithm (11-73 times faster than MMSE estimates), which is useful in a clinical setting.