Parallel acquisition technique combined with deep learning reconstruction for shoulder joint MRI
10.13929/j.issn.1672-8475.2024.09.011
- VernacularTitle:并行采集技术联合深度学习重建用于肩关节MRI
- Author:
Qingyu YAN
1
;
Jingyi YANG
;
Lin WEN
Author Information
1. 四川省人民医院医学装备部,四川 成都 610072
- Keywords:
shoulder joint;
magnetic resonance imaging;
deep learning;
prospective studies;
image quality
- From:
Chinese Journal of Interventional Imaging and Therapy
2024;21(9):557-560
- CountryChina
- Language:Chinese
-
Abstract:
Objective To observe the value of parallel acquisition technique(PAT)combined with deep learning reconstruction(DLR)for shoulder joint MRI.Methods Totally 74 subjects who underwent shoulder joint MR scanning were prospectively enrolled.Shoulder joint MRI were acquired using standard sequence(group A),PAT accelerated sequence(group B)and PAT accelerated sequence combined with DLR(group C),respectively.Subjective evaluation of overall image quality,artifacts and clarity of anatomical structures(including articular cartilages,ligaments and bone marrow)were performed using a 5-point scale,and signal-to-noise ratio(SNR)and contrast-to-noise ratio(CNR)of articular cartilages were calculated for objective evaluation,which were then compared among groups.Results No significant difference of the subjective scores of overall image quality,artifacts nor clarity of anatomical structures was found between group A and C(all P＞0.05),which were all higher than those in group B(all P＜0.05).SNR and CNR of articular cartilages of shoulder joint MRI in group C,A,B decreased in order(all P＜0.05).Conclusion PAT combined with DLR could shorten scanning time of shoulder joint MRI,reduce image noise and improve image quality.
