Value of deep learning reconstruction in high-resolution T2-weighted imaging of the uterus
10.3969/j.issn.1671-8348.2025.10.020
- VernacularTitle:深度学习重建技术在子宫高分辨T2WI中的应用价值
- Author:
Jing PAN
1
;
Rui JIN
;
Zhigang CHU
;
Renqiang YU
Author Information
1. 重庆医科大学附属第一医院放射科,重庆 400016
- Keywords:
deep learning;
cervical cancer;
high-resolution imaging;
reconstruction technology;
image quality;
MRI
- From:
Chongqing Medicine
2025;54(10):2357-2360
- CountryChina
- Language:Chinese
-
Abstract:
Objective To evaluate the application of deep learning reconstruction(DLR)in high-resolu-tion T2WI of the uterus and compare it with traditional reconstruction method.Methods A total of 45 pa-tients diagnosed with cervical cancer and undergoing pelvic MRI scans at the hospital from May to August 2024 were prospectively included in the study.DLR technology was used to reconstruct high-resolution T2WI images,which were then compared with high-resolution T2WI images obtained using traditional reconstruc-tion techniques.Likert-type scale was employed for subjective quality evaluation of artifacts and tissue con-trast in high-resolution T2WI images,while relative contrast(RC)between the lesion area and uterine myo-metrium was used for objective quality assessment of the images.Results The artifact score of high-resolu-tion T2WI images obtained using DLR technology showed no significant difference compared to traditional re-construction method(4.22±0.42 vs.4.16±0.37,P=0.18).However,the tissue contrast score was signifi-cantly higher than that of traditional reconstruction methods(4.38±0.49 vs.3.98±0.26,P＜0.001).The RC of high-resolution T2WI images obtained using DLR technology was superior to that of traditional recon-struction methods(0.74±0.06 vs.0.71±0.05,P＜0.001).Conclusion DLR demonstrates significant ad-vantages in high-resolution uterine T2WI.Although it shows no significant difference in artifact suppression compared to traditional methods,it improves tissue contrast and enhances lesion visualization.
