基于深度学习的甲状腺超声图像修复算法的初步研究  

Preliminary study on thyroid ultrasound image restoration algorithm based on deep learning

作　　者：张敏[1] 倪驰明 温佳恒 邓子烨 徐海珊[1] 楼海亚[1] 潘美[1] 李强[1] 周凌 张传菊[1] 凌玉 王娇妮 陈娟萍 王高昂 李世岩[1] Zhang Min;Ni Chiming;Wen Jiaheng;Deng Ziye;Xu Haishan;Lou Haiya;Pan Mei;Li Qiang;Zhou Ling;Zhang Chuanju;Ling Yu;Wang Jiaoni;Chen Juanping;Wang Gaoang;Li Shiyan(Department of Ultrasound in Medicine,Sir Run Run Shaw Hospital,Zhejiang University School of Medicine,Hangzhou 310016,China;Zhejiang University-University of Illinois at Urbana-Champaign Institute,Haining 314400,China)

机构地区：[1]浙江大学医学院附属邵逸夫医院超声医学科,杭州310016 [2]浙江大学伊利诺伊大学厄巴纳香槟校区联合学院,海宁314400

出　　处：《中华超声影像学杂志》2023年第6期515-522,共8页Chinese Journal of Ultrasonography

基　　金：浙江省自然科学基金(LY20H180005)。

摘　　要：目的探讨基于深度学习对被遮挡的甲状腺超声图像进行修复的可行性。方法回顾性收集自2020年1月至2021年10月于浙江大学医学院附属邵逸夫医院采集的甲状腺结节图像共358张,对图片进行随机遮挡后,使用DeepFillv2方法对上述图像被遮挡部分进行修复,比较修复前后图像的灰度值差异。邀请6位不同年资医师(主任医师、主治医师、住院医师各2位)比较修复前后图片是否存在形态差异,比较不同医师组判断的正确率及对图像差异的检出率。根据2020甲状腺结节超声恶性危险分层中国指南(C-TIRADS)对甲状腺结节图像进行超声特征提取(垂直位、实性、极低回声、可疑微钙化、边缘模糊及彗星尾伪像),比较修复前后图像中甲状腺结节超声特征的一致性。结果图片修复前后的灰度值均方误差范围为0.274~0.522,各组医师的正确率及检出率之间差异具有统计学意义(均P<0.001),总体正确率为51.95%,总体检出率为1.79%,其中主任医师与住院医师组内的正确率及检出率差异同样具有统计学意义(均P<0.001)。图像修复前后图像中各超声特征的一致性均高于70%,实性、彗星尾伪像的一致性均高于90%。结论基于深度学习的甲状腺超声图像修复算法可以有效修复被遮挡的甲状腺超声图像,同时可以保留甲状腺超声图像特征,有望扩大深度学习图像数据库规模,促进深度学习在超声领域的发展。Objective To explore the feasibility of deep learning-based restoration of obscured thyroid ultrasound images.Methods A total of 358 images of thyroid nodules were retropectively collected from January 2020 to October 2021 at Sir Run Run Shaw Hospital,Zhejiang University School of Medicine,and the images were randomly masked and restored using DeepFillv2.The difference in grey values between the images before and after restoration was compared,and 6 sonographers(2 chief physicians,2 attending physicians,2 residents)were invited to compare the rate of correctness of judgement and detection of image discrepancies.The ultrasound features of thyroid nodules(solid composition,microcalcifications,markedly hypoechoic,ill-defined or irregular margins,or extrathyroidal extensions,vertical orientation and comet-tail artifact)were extracted according to the Chinese Thyroid Imaging Reporting and Data System(C-TIRADS).The consistency of ultrasound features of thyroid nodules before and after restoration were compared.Results The mean squared error of the images before and after restoration ranged from 0.274 to 0.522,and there were significant differences in the rate of correctness of judgement and detection of image discrepancies between physicians of different groups(all P<0.001).The overall accuracy rate was 51.95%,the overall detection rate was 1.79%,there were significant differences also within the chief physicians and resident groups(all P<0.001).The agreement rate of all ultrasound features of the nodules before and after image restoration was higher than 70%,over 90%agreement rate for features such as solid composition and comet-tail artifact.Conclusions The algorithm can effectively repair obscured thyroid ultrasound images while preserving image features,which is expected to expand the deep learning image database,and promote the development of deep learning in the field of ultrasound images.

关 键 词：超声检查 甲状腺结节 图像修复 深度学习 

分 类 号：R445.1[医药卫生—影像医学与核医学] R581[医药卫生—诊断学]