The measurement of stratum corneum (SC) thickness from in-vivo Raman water concentration depth profiles is gaining in popularity and appeal due to the availability and ease of use of in-vivo confocal Raman measurement...The measurement of stratum corneum (SC) thickness from in-vivo Raman water concentration depth profiles is gaining in popularity and appeal due to the availability and ease of use of in-vivo confocal Raman measurement systems. The foundation of these measurements relies on high-quality confocal Raman spectroscopy of skin and the robust numerical analysis of water profiles, which allow for accurate determination of SC thickness. These measurements are useful for studying intrinsic skin hydration profiles at different body sites and for determining hydration properties of skin related to topically applied materials. While the use of high-quality in-vivo Raman instrumentation has become routine and its use for SC thickness measurement widely reported, there is lack of agreement as to the best method of computing SC thickness values from Raman water profiles. Several methods have been proposed and are currently in use for such computations, but none of these methods has been critically evaluated. The work reported in this paper describes a new method for the determination of stratum corneum thickness from in-vivo confocal Raman water profiles. The method represents a consensus approach to the problem, which was found necessary to apply in order to properly model and quantify the large diversity of water profile types encountered in typical in-vivo Raman water measurement. The methodology is evaluated for performance using three criteria: 1) frequency of minimum fitting error on modeling to a standard numerical function;2) frequency of minimum model error for consensus vs. individual SC thickness values;and 3) correlation with reflectance confocal microscopy (RCM) values for SC thickness. The correlation study shows this approach to be a reasonable replacement for the more tedious and time-consuming RCM method with R2 = 0.68 and RMS error = 3.7 microns over the three body sites tested (cheek, forearm and leg).展开更多
Purpose:Assessing the quality of the ocular surface by in vivo scanning laser confocal microscopy(IVCM)in primary open angle glaucoma(POAG)patients treated by Xen 45 Gel Stent,medical therapy and trabeculectomy.Method...Purpose:Assessing the quality of the ocular surface by in vivo scanning laser confocal microscopy(IVCM)in primary open angle glaucoma(POAG)patients treated by Xen 45 Gel Stent,medical therapy and trabeculectomy.Methods:Retrospective,single-center,single-masked,comparative study including 60 eyes of 30 patients(mean age 61.16±10 years)affected by POAG.Eyes were divided into 3 groups:Group 1 eyes underwent the Xen 45 Gel Stent procedure,Group 2 eyes were under medical therapy,Group 3 eyes were surgically treated by trabeculectomy.All patients underwent HRT II IVCM analysis of cornea,limbus,conjunctiva,sub-tenionian space and sclera.Results:The Xen 45 Gel stent,if properly positioned in the sub-conjunctival space preserves goblet cells and limits ocular surface inflammation.Regular corneal epithelial cells with micro-cysts,and normo-reflective sub-epithelial nerve plexus are documented by IVCM.In sub Tenon’s implants an alternative lamellar intra-scleral filtration is detectable.Combined surgical procedures show a noticeable number of inflammatory cells with rare micro-cysts.Post-trabeculectomy inflammatory reaction is more evident than Xen 45 Gel Stent associated surgical procedures,but less than medical therapy where a conspicuous presence of Langerhans cells,peri-neural infiltrates,marked loss of goblet cells and fibrosis is visible.Conclusion:Ocular surface inflammation was more notable in topical therapy than after trabeculectomy,which itself causes more inflammation than XEN Gel stents.展开更多
目的开发和评估基于深度学习算法的自动识别角膜共聚焦显微镜(in vivo confocal microscopy,IVCM)图片中角膜炎细胞的智能辅助诊断系统。方法纳入广西壮族自治区人民医院眼科感染性角膜炎患者IVCM图像。采用ResNet101卷积神经网络构建...目的开发和评估基于深度学习算法的自动识别角膜共聚焦显微镜(in vivo confocal microscopy,IVCM)图片中角膜炎细胞的智能辅助诊断系统。方法纳入广西壮族自治区人民医院眼科感染性角膜炎患者IVCM图像。采用ResNet101卷积神经网络构建智能模型。使用5倍交叉验证的方法对模型的效能进行检验,计算模型准确度、特异度和敏感度评估该智能辅助诊断系统的识别真菌菌丝、炎症细胞、活化的树突细胞的效能。结果该研究共纳入2105张图片,经交叉验证,该模型识别真菌菌丝的准确度为0.974,特异度为0.976,敏感度为0.971。识别炎症细胞的准确度为0.993,特异度为0.994,敏感度为0.990。识别活化的树突细胞的准确度为0.993,特异度为0.994,敏感度为0.990。结论该研究自主研发的基于深度学习算法的智能系统可有效地将共聚焦图片中的角膜炎异常细胞进行自动识别,在识别多种IVCM图像的角膜炎细胞中表现出良好的诊断效能。展开更多
文摘The measurement of stratum corneum (SC) thickness from in-vivo Raman water concentration depth profiles is gaining in popularity and appeal due to the availability and ease of use of in-vivo confocal Raman measurement systems. The foundation of these measurements relies on high-quality confocal Raman spectroscopy of skin and the robust numerical analysis of water profiles, which allow for accurate determination of SC thickness. These measurements are useful for studying intrinsic skin hydration profiles at different body sites and for determining hydration properties of skin related to topically applied materials. While the use of high-quality in-vivo Raman instrumentation has become routine and its use for SC thickness measurement widely reported, there is lack of agreement as to the best method of computing SC thickness values from Raman water profiles. Several methods have been proposed and are currently in use for such computations, but none of these methods has been critically evaluated. The work reported in this paper describes a new method for the determination of stratum corneum thickness from in-vivo confocal Raman water profiles. The method represents a consensus approach to the problem, which was found necessary to apply in order to properly model and quantify the large diversity of water profile types encountered in typical in-vivo Raman water measurement. The methodology is evaluated for performance using three criteria: 1) frequency of minimum fitting error on modeling to a standard numerical function;2) frequency of minimum model error for consensus vs. individual SC thickness values;and 3) correlation with reflectance confocal microscopy (RCM) values for SC thickness. The correlation study shows this approach to be a reasonable replacement for the more tedious and time-consuming RCM method with R2 = 0.68 and RMS error = 3.7 microns over the three body sites tested (cheek, forearm and leg).
文摘Purpose:Assessing the quality of the ocular surface by in vivo scanning laser confocal microscopy(IVCM)in primary open angle glaucoma(POAG)patients treated by Xen 45 Gel Stent,medical therapy and trabeculectomy.Methods:Retrospective,single-center,single-masked,comparative study including 60 eyes of 30 patients(mean age 61.16±10 years)affected by POAG.Eyes were divided into 3 groups:Group 1 eyes underwent the Xen 45 Gel Stent procedure,Group 2 eyes were under medical therapy,Group 3 eyes were surgically treated by trabeculectomy.All patients underwent HRT II IVCM analysis of cornea,limbus,conjunctiva,sub-tenionian space and sclera.Results:The Xen 45 Gel stent,if properly positioned in the sub-conjunctival space preserves goblet cells and limits ocular surface inflammation.Regular corneal epithelial cells with micro-cysts,and normo-reflective sub-epithelial nerve plexus are documented by IVCM.In sub Tenon’s implants an alternative lamellar intra-scleral filtration is detectable.Combined surgical procedures show a noticeable number of inflammatory cells with rare micro-cysts.Post-trabeculectomy inflammatory reaction is more evident than Xen 45 Gel Stent associated surgical procedures,but less than medical therapy where a conspicuous presence of Langerhans cells,peri-neural infiltrates,marked loss of goblet cells and fibrosis is visible.Conclusion:Ocular surface inflammation was more notable in topical therapy than after trabeculectomy,which itself causes more inflammation than XEN Gel stents.
文摘目的开发和评估基于深度学习算法的自动识别角膜共聚焦显微镜(in vivo confocal microscopy,IVCM)图片中角膜炎细胞的智能辅助诊断系统。方法纳入广西壮族自治区人民医院眼科感染性角膜炎患者IVCM图像。采用ResNet101卷积神经网络构建智能模型。使用5倍交叉验证的方法对模型的效能进行检验,计算模型准确度、特异度和敏感度评估该智能辅助诊断系统的识别真菌菌丝、炎症细胞、活化的树突细胞的效能。结果该研究共纳入2105张图片,经交叉验证,该模型识别真菌菌丝的准确度为0.974,特异度为0.976,敏感度为0.971。识别炎症细胞的准确度为0.993,特异度为0.994,敏感度为0.990。识别活化的树突细胞的准确度为0.993,特异度为0.994,敏感度为0.990。结论该研究自主研发的基于深度学习算法的智能系统可有效地将共聚焦图片中的角膜炎异常细胞进行自动识别,在识别多种IVCM图像的角膜炎细胞中表现出良好的诊断效能。
