目的:针对眼科超声影像检测及诊断中模型复杂度高、部署难度大以及准确度通常无法满足临床需求的问题,提出一种基于改进YOLOv5s的眼科超声影像病灶区域检测方法。方法:首先,建立包含星状玻璃体变性、视网膜脱离、玻璃体积血、玻璃体后...目的:针对眼科超声影像检测及诊断中模型复杂度高、部署难度大以及准确度通常无法满足临床需求的问题,提出一种基于改进YOLOv5s的眼科超声影像病灶区域检测方法。方法:首先,建立包含星状玻璃体变性、视网膜脱离、玻璃体积血、玻璃体后脱离、后巩膜葡萄肿5种眼科疾病图像的眼科超声图像数据集。其次,以YOLOv5s为基础,引入轻量级网络MobileNet对原主干特征提取网络CSPDarkNet进行替换,构建YOLOv5s-MobileNetV2模型。再次,通过平均精度均值(mean average precision,mAP)、参数量、每秒检测帧数等评估模型对眼科超声影像中病灶区域的检测性能。最后,基于PyQt5设计眼科超声影像智能检测软件。结果:YOLOv5s-MobileNetV2模型在测试集上的mAP、参数量、每秒检测帧数分别为97.73%、4.61×10^(6)、47帧/s。与YOLOv5s相比,YOLOv5s-MobileNetV2模型的mAP提升了0.22%,参数量减少了34.98%,具有更佳的实时性能。设计的眼科超声影像智能检测软件具有良好的人机交互能力,提升了YOLOv5s-MobileNetV2模型的临床适用性。结论:基于改进YOLOv5s的眼科超声影像病灶区域检测方法在实现轻量化的同时具有较好的检测性能,能够准确检测眼科病灶区域,满足眼科疾病临床诊断需求。展开更多
Photoacoustic imaging has many advantages in ophthalmic application including high-resolution,requirement of no exogenous contrast agent,and noninvasive acquisition of both morphologic and functional information.Howev...Photoacoustic imaging has many advantages in ophthalmic application including high-resolution,requirement of no exogenous contrast agent,and noninvasive acquisition of both morphologic and functional information.However,due to the limited depth of focus of the imaging method and large curvature of the eye,it remains a challenge to obtain high quality vascular image of entire anterior segment.Here,we proposed a new method to achieve high quality imaging of anterior segment.The new method applied a curvature imaging strategy based on only one time scanning,and hence is time efficient and more suitable for ophthalmic imaging compared to previously reported methods using similar strategy.A custom-built photoacoustic imaging system was adapted for ophthalmic application and a customized image processing method was developed to quantitatively analyze both morphologic and functional information in vasculature of the anterior segment.The results showed that the new method improved the image quality of anterior segment significantly compared to that of conventional high resolution photoacoustic imaging.More importantly,we applied the new method to study ophthalmic disease in an in vivo mouse model for the first time.The results verified the suitability and advantages of the new method for imaging the entire anterior segment and the numerous potentials of applying it in ophthalmic imaging in future.展开更多
AIM:To summarize the application of deep learning in detecting ophthalmic disease with ultrawide-field fundus images and analyze the advantages,limitations,and possible solutions common to all tasks.METHODS:We searche...AIM:To summarize the application of deep learning in detecting ophthalmic disease with ultrawide-field fundus images and analyze the advantages,limitations,and possible solutions common to all tasks.METHODS:We searched three academic databases,including PubMed,Web of Science,and Ovid,with the date of August 2022.We matched and screened according to the target keywords and publication year and retrieved a total of 4358 research papers according to the keywords,of which 23 studies were retrieved on applying deep learning in diagnosing ophthalmic disease with ultrawide-field images.RESULTS:Deep learning in ultrawide-field images can detect various ophthalmic diseases and achieve great performance,including diabetic retinopathy,glaucoma,age-related macular degeneration,retinal vein occlusions,retinal detachment,and other peripheral retinal diseases.Compared to fundus images,the ultrawide-field fundus scanning laser ophthalmoscopy enables the capture of the ocular fundus up to 200°in a single exposure,which can observe more areas of the retina.CONCLUSION:The combination of ultrawide-field fundus images and artificial intelligence will achieve great performance in diagnosing multiple ophthalmic diseases in the future.展开更多
文摘目的:针对眼科超声影像检测及诊断中模型复杂度高、部署难度大以及准确度通常无法满足临床需求的问题,提出一种基于改进YOLOv5s的眼科超声影像病灶区域检测方法。方法:首先,建立包含星状玻璃体变性、视网膜脱离、玻璃体积血、玻璃体后脱离、后巩膜葡萄肿5种眼科疾病图像的眼科超声图像数据集。其次,以YOLOv5s为基础,引入轻量级网络MobileNet对原主干特征提取网络CSPDarkNet进行替换,构建YOLOv5s-MobileNetV2模型。再次,通过平均精度均值(mean average precision,mAP)、参数量、每秒检测帧数等评估模型对眼科超声影像中病灶区域的检测性能。最后,基于PyQt5设计眼科超声影像智能检测软件。结果:YOLOv5s-MobileNetV2模型在测试集上的mAP、参数量、每秒检测帧数分别为97.73%、4.61×10^(6)、47帧/s。与YOLOv5s相比,YOLOv5s-MobileNetV2模型的mAP提升了0.22%,参数量减少了34.98%,具有更佳的实时性能。设计的眼科超声影像智能检测软件具有良好的人机交互能力,提升了YOLOv5s-MobileNetV2模型的临床适用性。结论:基于改进YOLOv5s的眼科超声影像病灶区域检测方法在实现轻量化的同时具有较好的检测性能,能够准确检测眼科病灶区域,满足眼科疾病临床诊断需求。
基金financial supports from the National Natural Science Foundation of China(NSFC)(Grants No.91739117,31570952,81873919,81371662 and 81927807)Shenzhen Science and Technology Innovation(Grant No.JCYJ20170413153129570)+1 种基金Beijing Natural Science Foundation of China(Grant No.3122010)。
文摘Photoacoustic imaging has many advantages in ophthalmic application including high-resolution,requirement of no exogenous contrast agent,and noninvasive acquisition of both morphologic and functional information.However,due to the limited depth of focus of the imaging method and large curvature of the eye,it remains a challenge to obtain high quality vascular image of entire anterior segment.Here,we proposed a new method to achieve high quality imaging of anterior segment.The new method applied a curvature imaging strategy based on only one time scanning,and hence is time efficient and more suitable for ophthalmic imaging compared to previously reported methods using similar strategy.A custom-built photoacoustic imaging system was adapted for ophthalmic application and a customized image processing method was developed to quantitatively analyze both morphologic and functional information in vasculature of the anterior segment.The results showed that the new method improved the image quality of anterior segment significantly compared to that of conventional high resolution photoacoustic imaging.More importantly,we applied the new method to study ophthalmic disease in an in vivo mouse model for the first time.The results verified the suitability and advantages of the new method for imaging the entire anterior segment and the numerous potentials of applying it in ophthalmic imaging in future.
基金Supported by 1.3.5 Project for Disciplines of Excellence,West China Hospital,Sichuan University(No.ZYJC21025).
文摘AIM:To summarize the application of deep learning in detecting ophthalmic disease with ultrawide-field fundus images and analyze the advantages,limitations,and possible solutions common to all tasks.METHODS:We searched three academic databases,including PubMed,Web of Science,and Ovid,with the date of August 2022.We matched and screened according to the target keywords and publication year and retrieved a total of 4358 research papers according to the keywords,of which 23 studies were retrieved on applying deep learning in diagnosing ophthalmic disease with ultrawide-field images.RESULTS:Deep learning in ultrawide-field images can detect various ophthalmic diseases and achieve great performance,including diabetic retinopathy,glaucoma,age-related macular degeneration,retinal vein occlusions,retinal detachment,and other peripheral retinal diseases.Compared to fundus images,the ultrawide-field fundus scanning laser ophthalmoscopy enables the capture of the ocular fundus up to 200°in a single exposure,which can observe more areas of the retina.CONCLUSION:The combination of ultrawide-field fundus images and artificial intelligence will achieve great performance in diagnosing multiple ophthalmic diseases in the future.
