Related Experiment Video
Updated: Jun 27, 2025
Development and Evaluation of 3D-Printed Cardiovascular Phantoms for Interventional Planning and Training
Published on: January 18, 2021
Feasibility of Artificial Intelligence Constrained Compressed SENSE Accelerated 3D Isotropic T1 VISTA Sequence For
Yue Ma1, Mengmeng Wang1, Yuting Qiao1
1Department of Radiology, The First Hospital of Jilin University, Changchun 130000, PR China.
Rationale And Objectives:
Investigate the feasibility of using deep learning-based accelerated 3D T1-weighted volumetric isotropic turbo spin-echo acquisition (VISTA) for vessel wall magnetic resonance imaging (VW-MRI), compared to traditional Compressed SENSE and optimize acceleration factor (AF) to obtain high-quality clinical images.
Methods:
40 patients with atherosclerotic plaques in the intracranial or carotid artery were prospectively enrolled in our study from October 1, 2022 to October 31, 2023 underwent high-resolution vessel wall imaging on a 3.0 T MR system using variable Compressed SENSE (CS) AFs and reconstructed by an optimized artificial intelligence constrained Compressed SENSE (CS-AI). Images were reconstructed through both traditional CS and optimized CS-AI. Two radiologists qualitatively assessed the image quality scores of CS and CS-AI across different segments and quantitatively evaluated SNR (signal-to-noise ratio) and CNR (contrast-to-noise ratio) metrics. Paired t-tests, ANOVA, and Friedman tests analyzed image quality metrics. Written informed consent was obtained from all patients in this study.
Results:
CS-AI groups demonstrated good image quality scores compared to reference scans until AF up to 12 (P < 0.05). The CS-AI 10 protocol provided the best images in the lumen of both normal and lesion sites (P < 0.05). The plaque SNR was significantly higher in CS-AI groups compared to CS groups until the AF increased to 12 (P < 0.05). CS-AI protocols had higher CNR compared to CS with whichever AF on both pre-and post-contrast T1WI (P < 0.05), The CNR was highest in the CS-AI 10 protocol on pre-contrast T1WI and in CS-AI 12 on post-contrast T1WI (P < 0.05).
Conclusion:
The study demonstrated the feasibility of using CS-AI technology to diagnose arteriosclerotic vascular disease with 3D T1 VISTA sequences. The image quality and diagnostic efficiency of CS-AI images were comparable or better than traditional CS images. Higher AFs are feasible and have potential for use in VW-MRI. The determination of standardized AFs for clinical scanning protocol is expected to help for empirical evaluation of new imaging technology.
