Automated early detection of diabetic retinopathy.
Automated early detection of diabetic retinopathy.
复制标题
DOI：
10.1016/j.ophtha.2010.03.046
复制
发表时间：
2010-06
期刊：
影响因子：
13.7
通讯作者：
Quellec G
中科院分区：
文献类型：
--
作者：
Abràmoff MD;Reinhardt JM;Russell SR;Folk JC;Mahajan VB;Niemeijer M;Quellec G
为了比较自动糖尿病视网膜病变 (DR) 检测的性能，使用赢得 2009 年视网膜病变在线挑战赛 (“Challenge2009”) 的算法与目前在大型计算机辅助早期 DR 检测项目 EyeCheck 中使用的算法。诊断测试或技术的评估。眼底摄影集由每只眼睛的两张眼底图像组成，根据 16,670 名之前未诊断为 DR 的糖尿病患者的 16,670 次就诊进行评估。每次就诊的眼底照片均由一位视网膜专家进行分析； 16,770 组中的 793 组被归类为包含超过最低 DR（转诊阈值）。两种算法检测器的结果分别应用于数据集，并通过标准统计测量进行比较。受试者工作特征曲线 (AUC) 下的面积，用于衡量 DR 检测的敏感性和特异性。一致性很高，“Eyecheck”算法检测到的包含超过最小 DR 的检查的 AUC 为 0.839，“Challenge2009”的 AUC 为 0.821，统计上不显着（z 得分 1.91）。如果任一算法组合检测 DR，则检测的 AUC 为 0.86，与理论上预期的最大值相同。在灵敏度为 90% 时，“EyeCheck”算法的特异性为 47.7%，“Challenge2009”算法的特异性为 43.6%。 DR 检测算法似乎正在成熟，检测性能的进一步改进无法与最佳临床实践区分开来，因为竞争性算法开发的性能现已达到人类阅读器内部变异性极限。迫切需要对规模更大、定义明确但更多样化的糖尿病患者群体进行额外的验证研究，预计对数百万糖尿病患者进行经济高效的 DR 早期检测，以便对那些患有早期而非晚期 DR 的患者进行分类。
To compare the performance of automated diabetic retinopathy (DR) detection, using the algorithm that won the 2009 Retinopathy Online Challenge Competition in 2009, (‘Challenge2009’) against that of the one currently used in EyeCheck, a large computer-aided early DR detection project. Evaluation of diagnostic test or technology. Fundus photographic sets, consisting of two fundus images from each eye, were evaluated from 16,670 patient visits of 16,670 people with diabetes who had not previously been diagnosed with DR. The fundus photographic set from each visit was analyzed by a single retinal expert; 793 of the 16,770 sets were classified as containing more than minimal DR (threshold for referral). The outcomes of the two algorithmic detectors were applied separately to the dataset and compared by standard statistical measures. The area under the Receiver Operating Characteristic curve (AUC), a measure of the sensitivity and specificity of DR detection. Agreement was high, and exams containing more than minimal DR were detected with an AUC of 0.839 by the ‘Eyecheck’ algorithm and an AUC of 0.821 for ‘Challenge2009’, a statistically non-significant difference (z-score 1.91). If either of the algorithms detected DR in combination, AUC for detection was 0.86, the same as the theoretically expected maximum. At 90% sensitivity, the specificity of the ‘EyeCheck’ algorithm was 47.7% and the ‘Challenge2009’ algorithm, 43.6%. DR detection algorithms appear to be maturing, and further improvements in detection performance cannot be differentiated from best clinical practices, because the performance of competitive algorithm development has now reached the human intra-reader variability limit. Additional validation studies on larger, well-defined, but more diverse populations of patients with diabetes are urgently needed, anticipating cost-effective early detection of DR in millions of people with diabetes to triage those patients who need further care at a time when they have early rather than advanced DR.
登录
查看更多内容
影响因子：
13.7
作者：
Bresnick, GH;Mukamel, DB;Cole, DR
通讯作者：
Cole, DR
影响因子：
10.6
作者：
Niemeijer, Meindert;van Ginneken, Bram;Abramoff, Michael D.
通讯作者：
Abramoff, Michael D.
影响因子：
4.2
作者：
Lin, DY;Blumenkranz, MS;Grosvenor, DM
通讯作者：
Grosvenor, DM
影响因子：
4.1
作者：
MILTON, RC;GANLEY, JP;LYNK, RH
通讯作者：
LYNK, RH
影响因子：
4.4
作者：
Larsen, N;Godt, J;Larsen, M
通讯作者：
Larsen, M