A new approach to accelerate readout segmented EPI with compressed sensing.

Abstract

High resolution diffusion-weighted imaging is limited by susceptibility-induced distortions and relaxation-induced blurring. Segmented acquisition techniques can address these limitations at the expense of a prolonged scan time. If segmentation is performed along the readout direction, e.g., in RESOLVE (readout segmentation of long and variable echo-trains), scan time can be reduced by readout (RO) partial Fourier methods, or simultaneous multi-slice (SMS) methods. In this paper, we present a new approach to additionally accelerate the image acquisition called variable segment (VASE) RESOLVE. To avoid discontinuities at the boundaries of the segments, the phase evolution and therefore the effective echo-spacing needs to be adjusted. To achieve this, we use higher undersampling factors in the outer parts of k-space. Simultaneously we increase the width of the outer segments resulting in an increase of the echo-spacing. Because of this variation, we introduce a kind of randomization to the sampling scheme. This enables the use of compressed sensing reconstruction techniques, which results in improved image quality compared to standard parallel imaging methods. The RMS errors for the VASE RESOLVE acquisitions were lower compared to the standard reconstructions. The VASE RESOLVE in vivo images show a higher apparent signal to noise ratio. VASE RESOLVE is a new approach to further decrease the acquisition time of RO segmented acquisitions. Compared to RESOLVE with SMS, VASE RESOLVE additionally reduces the acquisition time by a factor of 2.