Automatic off-resonance correction in spiral imaging with piecewise linear autofocus

Abstract

Off-resonance generates blurring artifacts in spiral images. Applications that often utilize spiral trajectories, such as fine-resolution imaging and rapid scanning, typically preclude the measurement of accurate field maps needed for effective off-resonance correction. Automatic deblurring, or autofocus, algorithms have been developed to estimate the field map directly from the corrupted data prior to off-resonance correction, eliminating the need for field map measurements. These algorithms rely in whole or in part on optimizing an objective function, and suffer from problems related to the accurate minimization and utility of the function. Here, a new method is presented to correct off-resonance blurring automatically without an objective function using a piecewise linear framework. Local linear field maps are estimated with a combination of k-space spectral analysis and mapdrift, an image feature-based correlation technique, for subsequent piecewise linear deblurring. This approach enables field map estimation without optimization, provides accurate off-resonance correction, is suitable for low signal-to-noise ratio and fine-resolution applications, and does not require access to the raw data. Deblurred images from fine-resolution spiral scans of a phantom and healthy volunteers at 3T show that the proposed method can be superior to conventional autofocus and comparable to field map-based correction.