Skip to main content

Table 1 Dataset summary

From: Automated detection of myopic maculopathy from color fundus photographs using deep convolutional neural networks

Data source Subject characteristics Grading distribution
No. of images No. of eyes No. of individuals Age
Mean (SD)
Female no./total individuals (%) Camera No. of gradable images/total images (%) No. of no MM/total gradable images (%) No. of TF/total gradable images (%) No. of PM/total gradable images (%)
Development dataset
 Qingdao Eye Hospital North Branch of Shandong First Medical University 15,857 15,857 8156 50.8 (11.3) 3809 (46.7) Canon, Syseye 14,356 (90.5) 11,856 (82.6) 996 (6.9) 1504 (10.5)
 Rongcheng Eye Hospital 15,683 15,683 7919 52.1 (15.4) 3857 (48.7) Zeiss 13,876 (88.5) 10,647 (76.7) 1749 (12.6) 1480 (10.7)
 Qilu Hospital of Shandong University (Qingdao) 4915 4915 2507 55.2 (13.2) 1243 (49.6) Canon 4393 (89.4) 2987 (68.0) 620 (14.1) 786 (17.9)
 No. 971 Hospital of the People’s Liberation Army 4139 4139 2184 59.8 (12.9) 1262 (57.8) Topcon 3890 (94.0) 1896 (48.7) 902 (23.2) 1092 (28.1)
External testing datasets
 Shandong Eye Hospital of Shandong First Medical University 7869 7869 4024 54.5 (15.5) 1898 (47.2) Topcon 7077 (89.9) 4884 (69.0) 1687 (23.8) 506 (7.2)
 Qingdao Eye Hospital of Shandong First Medical University 8685 8685 4440 49.2 (17.5) 2264 (51.0) Kowa 7909 (91.1) 6508 (82.3) 747 (9.4) 654 (8.3)
  1. MM = myopic maculopathy; TF = tessellated fundus; PM = pathologic myopia; SD = standard deviation