目的 探讨一种基于深度学习的视网膜毛细血管无灌注区(NPA)智能识别定量分析系统在视网膜中央静脉阻塞(CRVO)缺血分型诊断中的应用价值.方法 回顾性病例系列研究.选取2017年1月至2018年12月经山西省眼科医院门诊确诊为CRVO并行荧光素眼底血管造影(FFA)检查的343例343眼患者纳入本研究.随访并记录患者自发病起1年内是否发生新生血管并发症.应用人工智能诊断系统计算CRVO患者后极部55°视野范围NPA面积,受试者工作特征曲线(ROC)分析该NPA面积对CRVO发生新生血管并发症的诊断价值.结果 343例CRVO患眼中有26眼发生了新生血管并发症,发病率7.58％.依据NPA面积判断CRVO患者是否发生新生血管并发症的曲线下面积为0.889(95％CI:0.799～0.978,P<0.001).最佳截断值为20.997视盘面积(DA),灵敏度为0.808,26眼新生血管并发症患者中有21眼NPA值大于该值;特异度为0.946,317眼无并发症患者中有300眼NPA值小于该值.结论 基于深度学习的NPA智能识别定量分析系统可为CRVO分型诊断提供决策依据.55°视野范围后极部NPA>20 DA可作为CRVO缺血分型的阈值标准.

Objective:To investigate the effect of phacoemulsification and intraocular lens implantation on diabetic retinopathy (DR) in type 2 diabetes.Methods:This was a retrospective case series study. A total of 272 eyes of 136 cases with type 2 diabetes mellitus who underwent phacoemulsification and intraocular lens implantation from Jan. 2018 to Jul. 2020 in Shanxi Eye Hospital were enrolled in this study. There were 130 eyes of 65 cases without DR and 142 eyes of 71 cases with DR. The progress of fundus lesions before and after operation was compared and the comparison was also analyzed between the surgical eyes and the contralateral eyes.Results:No fundus lesion was found in patients without DR at 6 months after operation. The progressive fundus lesions were found in 23 eyes of DR patients, including 21 eyes (29.58%, 21/71) in the surgical eye group and 2 eyes (2.82%, 2/71) in the contralateral eye group. The risk of developing DR in the surgical eye group was significantly higher than that in the contralateral eye group ( Z=-3.391, P<0.001). The correlation analysis of Spearman showed that the degree of DR before operation was significantly related to the progress rate of DR. Conclusion:Phacoemulsification and intraocular lens implantation can aggravate the progress of fundus lesions in DR patients. The degree of DR is the risk factor for the development of DR.

Objective:To evaluate the application of ultra-wide-field fundus imaging in the diagnosis of traumatic chorioretinopathy.Methods:This was a retrospective case series study. A total of 34 eyes of 32 cases with traumatic chororetinopathy were examined by using the ultra-wide-field fundus imaging system from Oct. 2019 to Oct. 2020 in Shanxi Eye Hospital. The characteristics of ultra-wide-field fundus color photography and autofluorescence imaging were observed.Results:Ultra-wide-field fundus photography and autofluorescence imaging showed characteristic changes. There were many types of lesions in the same eye, including retinal concussion, retinal contusion, retinal hemorrhage, choroidal rupture, retinal detachment and retinal tear.Conclusion:The ultra-wide-field color photography and autofluorescence examination play an important role in the diagnosis and follow-up observation of traumatic chorioretinopathy. It is more conductive to understand the damage degree and recovery of retinal and choroidal anatomical structure and function after ocular contusion, and estimate the prognosis of the disease.

Objective:The aim of this study was to systematically evaluate the effect of intraoperative combined with internal limiting membrane (ILM) removal during the treatment of rhegmatogenous retinal detachment (RRD) with vitrectomy (PPV).Methods:The Pubmed, Embase, Cochrane Librara, Clinical Trials.gov and CNKI databases were searched. The search time was from the self-built database to November 2020. All clinical controlled trials of PPV combined with ILM peeling and without ILM peeling were collected for the treatment of RRD. Two researchers independently screened the literatures and extracted postoperative macular membrane (ERM) incidence, best corrected visual acuity (BCVA), RRD recurrence rate, ellipsoid zone defect incidence, and retinal nerve fiber layer separation (DONFL) incidence. The Q test and I2 statistics were used to evaluate the heterogeneity of the data. When the heterogeneity was large (I2>50%), the random effects model was used to analysis; otherwise, the fixed effects model. Binary variables was represented by risk ratio (RR) and its 95% confidence interval (CI) ; continuous variables was represented by difference (MD) and its 95%CI.Results:893 literatures were collected after a preliminary searching. After excluding duplicate documents, articles irrelevant to the study, and documents that did not meet the outcome indicators, and finally 11 literatures were included. Among of them, a total of 987 eyes were included, of which 514 eyes were combined with ILM removal and 473 eyes without ILM removal during the operation. The original studies had high homogeneity, and the results were robust. The heterogeneity of the incidence of ERM among patients in each study was relatively small (I2=26%). The postoperative ERM rate of patients in PPV combined with ILM peeling group was significantly lower than that of the non-ILM peeling group. After the fixed effects model test, the difference between two groups was statistically significant (RR=0.11, 95%CI: 0.07~0.18, P<0.05). The heterogeneity of BCVA of patients in each study was big (I2=86%). After the random effects model analysis, there was no significant difference in BCVA between two groups (MD=0.08, 95%CI: -0.05~0.21, P>0.05). The heterogeneity of RRD recurrence rate among patients in four studies was big (I2=73%). The postoperative RRD recurrence rate in the combined ILM peeling group was lower than that in the non-ILM peeling group. After the fixed effects model analysis, the difference between two groups was statistically significant (RR=0.48, 95%CI: 0.30~0.76, P<0.05). The incidence of ellipsoidal zone defects among patients in four studies was quite heterogeneous (I2 =73%). After the fixed effects model analysis, there was no significant difference in the incidence of postoperative ellipsoidal band defects between two groups (RR=0.99, 95%CI: 0.38~2.53, P>0.05). The incidence of postoperative DONFL among patients in five studies was relatively heterogeneous (I2=0). The incidence of DONFL in the combined ILM peeling group was higher than that in the non-ILM peeling group. After the fixed effects model analysis, the difference between two groups was statistically significant (RR=39.08, 95%CI: 11.35~134.60, P<0.05).Conclusions:ILM peeling during PPV operation could prevent the formation of ERM after RRD and reduce the recurrence rate of RRD after RRD; however, PPV combined with ILM removal had the disadvantage for vision and also has potential harm to the macula.

脉络膜裂伤是眼部受到外力损伤比如钝挫伤等所致的视网膜色素上皮(RPE)层、Bruch膜和脉络膜毛细血管层发生裂伤.可位于眼底视盘周围、黄斑区及其周围.病灶形态一般呈弧形、条形裂纹样,中部略宽,色泽呈黄白色或淡黄色,周围可伴有出血,部分病灶可继发CNV. 其形成可能与RPE局部断裂合并脉络膜血管损伤、视网膜色素上皮细胞过度增生有关[1].

多发性一过性白点综合征(MEWD)好发于青年女性,具体机制不明,Lages等［1］认为是由于脉络膜毛细血管的炎症引起.该病最早在1984 年Jampol 等［2］首次提出.因该病起病急,有自限性,病程短且变化快,尤其是病情轻微的患者临床上容易漏诊及误诊.

目的 观察糖尿病黄斑缺血(diabetic macular ischemia,DMI)患者黄斑区脉络膜厚度特征.方法 回顾性病例对照研究.经FFA检查确诊的糖尿病视网膜病变(diabetic retinopathy,DR)不伴糖尿病黄斑水肿患者80例102眼纳入本研究.其中DR伴DMI患者37例47眼为DMI组,DR不伴DMI患者43例55眼为NDMI组;选择同期年龄、性别相匹配的经散瞳检查明确眼底无明显DR的2型糖尿病患者33例50眼作为对照组(NDR组).所有患者均利用EDI-OCT测量黄斑中心凹下脉络膜厚度(subfoveal choroidal thick-ness,SFCT).采用单因素方差分析及LSD-t检验比较三组间SFCT差异.结果 三组间性别及年龄比较,差异均无统计学意义(χ2=1.412,P=0.494;F=0.784,P=0.459).EDI-OCT检查结果显示,DMI组、NDMI组及NDR组SFCT分别为(357.13±101.41)μm、(296.00±95.88)μm、(342.68±80.76)μm,三组相比差异有统计学意义(F=6.137,P=0.003).组间两两比较结果显示,与NDMI组比较,DMI组SFCT明显增厚,差异有统计学意义(t=3.308,P=0.001);与NDR组比较,NDMI组SFCT明显变薄,差异有统计学意义(t=2.568,P=0.011);DMI组与NDR组SFCT比较,差异无统计学意义(t=0.765,P=0.446).结论 DR患者SFCT变化与是否存在DMI有关,DMI患眼较不伴DMI患眼SF-CT明显增厚.

PurposeRetinal vein occlusion (RVO) is the second most common cause of vision loss after diabetic retinopathy due to retinal vascular disease. Retinal nonperfusion (RNP), identified on fluorescein angiograms (FA) and appearing as hypofluorescence regions, is one of the most significant characteristics of RVO. Quantification of RNP is crucial for assessing the severity and progression of RVO. However, in current clinical practice, it is mostly conducted manually, which is time‐consuming, subjective, and error‐prone. The purpose of this study is to develop fully automated methods for segmentation of RNP using convolutional neural networks (CNNs).MethodsFA images from 161 patients were analyzed, and RNP areas were annotated by three independent physicians. The optimal method to use multi‐physicians’ labeled data to train the CNNs was evaluated. An adaptive histogram‐based data augmentation method was utilized to boost the CNN performance. CNN methods based on context encoder module were developed for automated segmentation of RNP and compared with existing state‐of‐the‐art methods.ResultsThe proposed methods achieved excellent agreements with physicians for segmentation of RNP in FA images. The CNN performance can be improved significantly by the proposed adaptive histogram‐based data augmentation method. Using the averaged labels from physicians to train the CNNs achieved the best consensus with all physicians, with a mean accuracy of 0.883±0.166 with fivefold cross‐validation.ConclusionsWe reported CNN methods to segment RNP in RVO in FA images. Our work can help improve clinical workflow, and can be useful for further investigating the association between RNP and retinal disease progression, as well as for evaluating the optimal treatments for the management of RVO.

Objective To observe the changes ofmacular structure and choroidal capillary blood flow density in patients with acute central serous chorioretinopathy (CSC).Methods Prospective cross-sectional study.A total of 24 eyes of 24 patients with monocular acute CSC (case group) diagnosed by clinical examination from Shanxi Eye Hospital during January and March 2018 were included in the study.The eyes (24 eyes) and contralateral eyes (24 eyes) of the patients in the case group were set to CSC group and contralateral eye group,respectively.Twenty-one eyes of 21 healthy volunteers with age and gender matching were selected as normal control group.The macular structure of the eyes were observed by OCT and OCT angiography (OCTA),and the blood vessel density of choroidal capillary layer in the circular area of the macular area with a radius of 1 mm was measured.The paired t-test was used to compare the differences in blood flow density in the choroidal capillaries between the three groups.Results The results of OCT showed that the serous neuroepithelial detachment in the macular area was observed in all eyes of the CSC group,with or without RPE detachment being 20 or 4 eyes,respectively.Of the 24 eyes in the contralateral eye group,13 eyes (54.2％) had thick choroidal RPE lesions (PPE).There was no abnormality in the retina and choroidal structure in the macular area of the normal control group.The results of OCTA showed that the blood flow density of choroidal capillaries in the CSC group,the contralateral eye group and the normal control group were 1.759 ± 0.132,1.924± 0.463,and 1.940± 0.033,respectively.Compared with the eyes of the contralateral eye group and the normal control group,the blood flow density of choroidal capillaries in the CSC group was significantly lower (t=6.611,6.474;P=0.000,0.000).There was no significant difference in the blood flow density of choroidal capillary layer between the contralateral eye group and the normal control group (t=1.328,P ＞ 0.05).In the contralateral eye group,there was no significant difference in the blood flow density of choroidal capillary layer between PPE eyes and no RPE eyes (t=0.806,P＞0.05).Conclusions There is 54.2％ of the contralateral eyes in the monocular acute CSC patients with PPE.The choroidal capillary layer blood flow density is lower than that of the contralateral and normal eyes.

荧光物质被某种波长的光照射后会激发出比入射光波长更长的光. 眼底荧光血管造影(FFA)应用此原理将荧光素钠经静脉注入人体并随全身血液循环到达眼底,应用特殊的眼底照相机激发其产生荧光,并拍摄记录其在视网膜血管动态循环情况.

视网膜格子样变性是视网膜周边部退行性变的一种表现,发病机制不详[1].以往对格子样变性的认识和研究大都为形态学和组织学方面[2-4].眼底自身荧光(AF)作为一种快速无创的眼底检查,可以反映活体内视网膜色素上皮(RPE)功能和代谢的改变[5].

Objective To observe the laser spots created by subthreshold micropulse laser (577 nm) photocoagulation for macular edema with a multi-wavelength photographic technique (F-10).Methods It was a prospective case-series study.Seventeen patients (19 eyes) with macular edema diagnosed by fundus fluorescein angiography (FFA) and optical coherence tomography (OCT) were included.Eight patients (10 eyes) were diagnosed with diabetic macular edema (DME),5 patients (5 eyes) with macular edema attributable to central retinal vein occlusion (CRVO) and 4 patients (4 eyes) with macular edema attributable to branch retinal vein occlusion (BRVO).First,the threshold energy (P) was tested by continuous wave laser photocoagulation,then with the detection of the F-10 in retro mode,19 eyes underwent subthreshold micropulse laser photocoagulation.The laser energy was based on P.Fundus color images and autofluorescence were obtained before and immediately after subthreshold micropulse laser photocoagulation.Changes in central macular thickness (CMT) and best corrected visual acuity (BCVA) before treatment and 1 month after treatment were compared with a paired t test and signed rank sum test,respectively.Results Immediately after treatment,in retro mode images,laser spots appeared under laser energy of 122.4％ P in 6 eyes,153.1％ P in 8 eyes and 183.7％ P in 5 eyes.There were no changes in fundus color images or autofluorecence in any of the eyes.After 1 month of treatment,CMT had significantly decreased compared to that before treatment (t=4.238,P＜0.01),BCVA showed no significant difference between pre- and post-treatment.Conclusion Retro mode imaging is useful to detect the invisible laser spots created by subthreshold micropulse laser photocoagulation,and can help to select the appropriate laser energy for treatment.

Objective To compare the efficacy of subthreshold micropulse 577nm laser and traditional laser for diabetic macular edema (DME).Methods In a total of 28 cases (44 eyes) with DME who underwent routine ocular examination,FFA and OCT were enrolled in this study.The eyes were randomized into 2 groups for subthreshold micropulse 577nm laser (21 eyes) or traditional laser (23 eyes).The changes of best corrected visual acuity (BCVA) and central macular thickhess (CMT) at 1-2 months,3-5 months follow-up between 2 groups were compared.Results Before the treatment,at 1-2 months and 3-5 months follow-up,the changes of BCVA between micropulse group and traditional group did not vary significantly (Z=0.155,-0.624,and -0.258; P =0.877,0.532,and 0.817).The differences in CMT between 2 groups did not vary significantly before the treatment (Z=-0.329,P =0.742).The changes of CMT at 1-2 months,3-5 months follow-up was significantly in micropulse group (Z=-2.086 and-2.223,P =0.037 and 0.026) and traditional group (Z=-4.091 and -3.621,P ＜ 0.001 and P ＜ 0.001) respectively.But the reduction of CMT in traditional group was significantly higher than micropulse group (Z=-2.221 and -2.756; P =0.026 and 0.006).Conclusions Both subthreshold micropulse 577nm laser and traditional laser for DME are effective.Both of them can maintain visual acuity in the short time.The latter is more effective in reducing the CMT in the short time.