Efficient Reconstruction of Free Breathing Under-Sampled Cardiac Cine MRI

TL;DR

This paper introduces a novel two-stage compressed sensing-based method for reconstructing free-breathing cardiac cine MRI images, effectively correcting respiratory motion artifacts and improving image quality from undersampled data.

Contribution

It presents a new motion correction and reconstruction algorithm that sorts data by respiratory state and iteratively refines images, outperforming existing CS-based methods.

Findings

Better reconstruction of respiratory motion corrected images

Quantitative improvements over traditional CS methods

Validated on simulated and clinical data

Abstract

Respiratory motion can cause strong blurring artifacts in the reconstructed image during MR acquisition. These artifacts become more prominent when use in the presence of undersampled data. Recently, compressed sensing (CS) is developed as an MR reconstruction technique, to recover good quality images from the compressive k-space samples. To maximize the benefits of CS in free breathing data, it is understandable to use CS with the motion corrected images. In this paper, we have developed a new CS based motion corrected image reconstruction technique. In this two-stage technique, we use similarity measure to sort the motion corrupted data into different respiratory states. Then, we use a new reconstruction algorithm, which iteratively performs reconstruction and motion correction. The performance of the proposed method is qualitatively and quantitively evaluated using simulated data and clinical data. Results depict that this method performs the better reconstruction of respiratory motion corrected cardiac cine images as compared to the CS based reconstruction method.