基于深度学习压缩感知技术在子宫T2WI中的对比研究

Application of Compressed Sensing Deep Learning-based Reconstruction in Uterine T2WI MR Imaging

目的:使用深度学习压缩感知技术行子宫T2WI,通过对综合质量的评估,探讨其在临床应用中的可行性.方法:选取临床女性盆腔检查患者80例,分别应用常规并行采集(PI)和深度学习卷积神经网络压缩感知(CNN-CS)技术行T2WI,各采集40例.通过对运动伪影和组织边界清晰度评分进行评估,以及对子宫内膜、肌层与结合带对比度进行比较.结果:CNN-CS扫描的T2WI图像质量总体评分显著高于PI法(2.75±0.44 vs 2.35±0.53,P<0.05);CNN-CS组子宫内膜、肌层与结合带对比度均优于常规PI组(0.74±0.07 vs 0.60±0.11,P<0.001;0.53±0.11 vs 0.44±0.10,P<0.05);CNN-CS组成像时间小于常规PI组.结论:与常规PI技术成像对比,基于深度学习的CNN-CS技术对子宫T2WI能够减少伪影的影响并提高组织图像对比度,可优化图像质量并减少成像时间.

Purpose:To improve the image quality of female uterus T2WI with deep learning compressed sensing technique,and to explore its feasibility in clinical application.Methods:Eighty female patients undergoing pelvic examination were selected.T2WI was performed using conventional parallel acquisition(PI)and deep learning convolutional neural network compressed sensing(CNN-CS)technology,respectively,with 40 cases collected each.Image motion artifact and boundary sharpness of uterine tissue were scored.The contrast of endometrium,myometrium to junctional zone were compared.Results:The overall score of T2WI image quality of CNN-CS scanning was significantly higher than that of PI technique(2.75±0.44 vs 2.35±0.53,P<0.05);the contrast of endometrium,myometrium to the junctional zone in CNN-CS group was better than that in traditional PI group(0.74±0.07 vs 0.60±0.11,P<0.001;0.53±0.11 vs 0.44±0.10,P<0.05);The acquisition time of the CNN-CS group was shorter than that of the conventional PI group.Conclusion:Compared with conventional PI technique,the CNN-CS technique based on deep learning can be used to reduce the influence of respiratory artifacts and improve tissue image contrast,which can optimize image quality and reduce acquisition time.