目的：探究光学相干断层血管造影(OCTA)参数在糖尿病视网膜病变(DR)诊断中的价值，并为糖尿病肾病(DN)患者提供更灵敏的OCTA筛查参数，以便在疾病早期诊断DR的并发。

方法：纳入2022/2023就诊于中山大学附属第七医院眼科的患者200例，其中包括95例首次诊断为DR的患者和105例未诊断为DR的患者，所有患者均进行OCTA检查，收集人口统计数据和肾功能参数。经质量检查后，测量患者3 mm×3 mm和6 mm×6 mm窗口的中央凹无血管区面积、血管密度(VD)和灌注密度(PD)指标情况。

结果：利用随机森林和多因素回归方法建立一个基于12个变量(年龄、血糖、血压、血脂、糖化血红蛋白、尿素、糖尿病持续时间、HUA、DN、CMT)的DR诊断模型。添加特定的OCTA参数(6 mm×6 mm窗口中的外层血管密度，AUC=0.837 vs 0.819，P=0.03)增强了现有DR诊断模型的有效性。在对DN患者的研究中，6 mm×6 mm窗口中的参数(内层血管密度； 外层血管密度； 全血管密度； 外层灌注密度； 全灌注密度)提高了DR的诊断效果。

结论：6 mm×6 mm窗口中外层血管密度参数的加入可以提高传统DR诊断模型的效果。同时，与3 mm×3 mm窗口相比，聚焦于6 mm×6 mm窗口中的微血管参数对诊断DN患者并发DR更有意义。

METHODS: A total of 200 patients who treated in the ophthalmology department of the Seventh Affiliated Hospital, Sun Yat-sen University from 2022 to 2023 were included, including 95 first-diagnosed DR patients and 105 patients without DR, and all patients underwent OCTA examination and a collection of demographics and renal function parameters. After a quality check, automated measurements of the foveal avascular zone area, vessel density(VD), and perfusion density(PD)of both 3 mm×3 mm and 6 mm×6 mm windows were obtained.

RESULTS: Using random forest and multivariate Logistic regression methods, we developed a diagnostic model for DR based on 12 variables(age, FBG, SBP, DBP, HbA1c, ALT, ALP, urea/Scr, DM duration, HUA, DN, and CMT). Adding specific OCTA parameters enhanced the efficacy of the existing diagnostic model for DR(outer vessel density in 6 mm×6 mm window, AUC=0.837 vs 0.819, P=0.03). In the study of DN patients, the parameters in the 6 mm×6 mm window improved the diagnostic efficacy of DR(inner VD; outer VD; full VD; outer PD; full PD).

CONCLUSION:The outer VD in the 6 mm×6 mm window can enhance the efficacy of the traditional DR diagnostic model. Meanwhile, compared with the 3 mm×3 mm window, the microvascular parameters in the 6 mm× 6 mm window focusing on DN patients can be more sensitive to diagnosing the occurrence of DR.