Susceptibility artifact correction for sub-millimeter fMRI using inverse phase encoding registration and T1 weighted regularization

Background: Functional magnetic resonance imaging (fMRI) enables non-invasive examination of both the structure and the function of the human brain. The prevalence of high spatial-resolution (sub-millimeter) fMRI has triggered new research on the intra-cortex, such as cortical columns and cortical layers. At present, echo-planar imaging (EPI) is used exclusively to acquire fMRI data; however, susceptibility artifacts are unavoidable. These distortions are especially severe in high spatial-resolution images and can lead to misrepresentation of brain function in fMRI experiments. New method: This paper presents a new method for correcting susceptibility artifacts by combining a T1-weighted (T) image and inverse phase-encoding (PE) based registration. The latter uses two EPI images acquired using identical sequences but with inverse-PE directions. In the proposed method, the T image is used to regularize the registration, and to select the regularization parameters automatically. The motivation is that the T image is considered to reflect the anatomical structure of the brain. Results: Our proposed method is evaluated on two sub-millimeter EPI-fMRI datasets, acquired using 3T and 7T scanners. Experiments show that the proposed method provides improved corrections that are well-aligned to the T image. Comparison with existing methods: The proposed method provides more robust and sharper corrections and runs faster compared with two other state-of-the-art inverse-PE based correction methods, i.e. HySCO and TOPUP. Conclusions: The proposed correction method used the T image as a reference in the inverse-PE registration. Results show its promising performance. Our proposed method is timely, as sub-millimeter fMRI has become increasingly popular