Browse Teach Curate

Physics-Constrained Unsupervised Estimation of High-Field MR Images from Low-Field Scans

An unsupervised method that recovers high-field-quality MRI from cheaper low-field scans by jointly estimating an unknown linear transformation and the target image via alternating minimization, regularized by the field dependence of tissue T1 relaxation. A student learns how a physics prior replaces paired training images and registration in cross-field MRI enhancement.

Rosie Text

Estimation of 3T MR images from 1.5T images regularized with Physics based Constraint ⋆ Prabhjot

This MICCAI 2023 paper (a conference paper, not lecture notes despite the Springer LNCS venue) proposes an unsupervised post-processing method to estimate high-field MR images (e.g. 3T) from low-fiel…