Clinical application of single breath-holding left ventricle artificial intelligence assisted compression sensing T 2-weighted dark blood imaging sequence / 中华放射学杂志

ABSTRACT

Objective:To explore the feasibility of cardiac T 2 weighted dark blood sequence (T 2W-DB) based on artificial intelligence assisted compression-sensing(ACS) in clinical cardiac MR examination, compared with the conventional cardiac fast spin-echo T 2W-DB sequence. Methods:A total of 38 patients referred for cardiac MR examination in Tongji Hospital were enrolled prospectively from August to December 2021. The conventional T 2W-DB scan and the single-shot ACS T 2W-DB scan were acquired at continuous short-axial slices covering the whole left ventricle in all patients, and the acquisition time of each scan was recorded. The image quality of the two sequences was evaluated by the objective quantitative parameters and the subjective scoring methods, respectively. The signal to noise ratio (SNR), the contrast to noise ratio of the interventricular septum to blood pool (CNR), and the sharpness of the images were calculated. The subjective scoring was to evaluate the overall image quality, the effect of blood pool suppression, the visibility of right ventricular free wall, left ventricular free wall, and interventricular septum with a 5-point Likert scale. The intraclass correlation coefficient and Kendall W were calculated to evaluate the interobserver agreement of the objective quantitative parameters and subjective scoring. The paired t-test or Wilcoxon test was used to compare the difference in acquisition time, objective quantitative parameters and subjective scoring between the conventional T 2W-DB and the single-shot ACS T 2W-DB. Results:The inter-observer agreement between the single-shot ACS T 2W-DB and conventional T 2W-DB was good in all the objective quantitative parameters and subjective scoring of image quality ( P<0.05 for all). Compared with conventional T 2W-DB, acquisition time of single-shot ACS T 2W-DB was remarkably shortened [(85.8±14.7) s vs. (16.9±3.0) s, t=35.42, P<0.001]. Compared with SNR (66.4±29.0) and CNR(61.8±28.6) of conventional T 2W-DB, single-shot ACS T 2W-DB had significantly higher SNR(110.8±36.8, t=-8.13, P?0.001) and CNR(88.2±31.1, t=-5.89, P?0.001). Compared with conventional T 2W-DB, single-shot ACS T 2W-DB had better blood pool inhibition effect (4.6±0.6 vs. 4.7±0.5, Z=-2.64, P=0.008). There was no significant difference in overall image quality, visibility of right ventricular free wall, left ventricular free wall, and interventricular septum between the two sequences( P>0.05 for all). Conclusions:In cardiac MR examination, compared with the conventional T 2W-DB sequence, the single-shot ACS T 2W-DB sequence can significantly shorten the acquisition time and obtain better image quality.