Table 1 Number of slides and cases in the entire dataset for this study.

From: Yet Another Automated Gleason Grading System (YAAGGS) by weakly supervised deep learning

Category

Discovery (HUMC)

Discovery (KUGH)

Validation (HUMC)

Validation (KUGH)

Gleason2019

Benign

3537 (490)

604 (109)

439 (64)

89 (15)

26

66.3%

45.4%

59.6%

44.5%

10.7%

Grade group 1

570 (311)

134 (67)

76 (42)

19 (10)

70

(Gleason score ≤ 6)

10.7%

10.1%

10.3%

9.5%

28.7%

Grade group 2

379 (190)

80 (55)

68 (33)

14 (9)

23

(Gleason score 7 = 3 + 4)

7.1%

6.0%

9.2%

7.0%

9.4%

Grade group 3

274 (142)

231 (87)

51 (23)

40 (12)

21

(Gleason score 7 = 4 + 3)

5.1%

17.4%

6.9%

20.0%

8.6%

Grade group 4

300 (132)

118 (57)

47 (18)

20 (10)

101

(Gleason score 8)

5.6%

8.9%

6.4%

10.0%

41.4%

Grade group 5

275 (90)

162 (43)

55 (14)

18 (6)

3

(Gleason score ≥ 9)

5.2%

12.2%

7.5%

9.0%

1.2%

Total

5,335 (543)

1,329 (146)

736 (78)

200 (21)

244

100%

100%

100%

100%

100%

  1. The ratio of each category is also presented per each dataset type and institution. As each case can contain multiple slides from different categories, the sum of the cases for each category does not coincide with the total cases. For Gleason 2019, the case id couldn’t be identified for each TMA image.