Compressed Sensing and Parallel Imaging for Double Hepatic Arterial Phase Acquisition in Gadoxetate-Enhanced Dynamic Liver Magnetic Resonance Imaging

Abstract

The aim of this study was to investigate the utility of a combined compressed sensing and parallel imaging (C-SENSE) technique for single breath-hold, double arterial phase (AP) examinations in gadoxetate-enhanced magnetic resonance imaging (MRI) of the liver.

MATERIALS AND METHODS: We retrospectively reviewed single breath-hold, double AP images obtained by using a C-SENSE technique for gadoxetate-enhanced dynamic liver MRI in a total of 127 patients (89 men and 38 women; mean age, 62.6 ± 7.5 [range, 29-87] years). For qualitative analysis, 3 readers independently scored the timing of the AP images, degree of artifacts, and overall image quality on both the first and second AP images (AP1 and AP2, respectively). The combined scores of AP1 and AP2 (AP1 + AP2) were determined by using the better scores from the 2 sets. Focal lesion detectability was assessed for 124 lesions with arterial enhancement on AP1 and AP2, and on simultaneous review of both AP1 and AP2. Then, in 62 patients whose previous gadoxetate-enhanced single AP images were available, AP timing and overall image quality were compared between single and double AP images. Wilcoxon signed rank test was performed for each comparison. Fleiss kappa value was calculated for analysis of interreader agreement.

RESULTS: Optimal AP timing was achieved in 86% of AP1, 65% of AP2, and 90% of AP1 + AP2 images; results were significantly better for AP1 and AP1 + AP2 images than for AP2 images (P < 0.001 for both comparisons). Respiratory motion artifacts were negligible in 73% of the AP1 + AP2 images, which was significantly better than the corresponding values for the AP1 (61%, P < 0.001) or AP2 (50%, P < 0.001) images. Overall image quality was significantly better for AP1 + AP2 (excellent in 54%) than for AP1 (49%, P < 0.001) or AP2 (39%, P < 0.001) images. Lesion detectability was comparable between AP1 and AP2 images and was significantly better on AP1 + AP2. Comparison of single and double AP imaging techniques showed better AP timing (P = 0.004) and fewer respiratory motion artifacts (P < 0.001) for AP1 + AP2 than for the single AP images.

CONCLUSIONS: The C-SENSE technique may facilitate single breath-hold, double AP imaging with optimal timing and reduced respiratory motion artifacts in gadoxetate-enhanced dynamic MRI of the liver.