Skip to main content
. 2020 Sep 25;9(2):53. doi: 10.1167/tvst.9.2.53

Table 4.

Keratoconus Probability Prediction of Subclinical Keratoconus Casesa by VGG16 and Comparing With Topographic, Tomographic, and Tonometry Indexes

TMS Indexes (Topography) Pentacam (Tomography) Corvis (Tonometry)
Case No. Probability of keratoconus (%) KCI KSI BAD-D CBI
1 97% 15.7% 35% 0.23 0
2 89% 24.2% 37.5% 0.22 0
3 89% 29.3% 29.4% 0.5 0.07
4 81% 0% 0% 2.75 0.67
5 73% 23.8% 31.7% 3.03 0.08
6 67% 34.3% 22.3% 0.19 0
7 64% 0% 0% 0.5 0
8 55% 0% 0% 2.1 0.02
9 44% 41.2% 37.2% 0.36 0
10 37% 0% 19.3% 0.64 0
11 36% 26.5% 31.7% 1.81 0.82
12 20% 0% 0% 0.94 0
13 14% 0% 15.3% 0.49 0.06
14 12% 0% 0% 2.82 1
15 10% 0% 0% 0.92 0
16 7% 0% 0% 1.83 0.03
17 4% 0% 20.8% 1.84 0.05
18 4% 0% 0% 1.27 0.05
19 3% 0% 0% 1.57 0.07
20 3% 0% 0% 0.91 0
21 2% 0% 15.5% 1.61 0.25
22 1% 0% 20% 2.37 0
23 1% 0% 17.9% 2.11 0
24 1% 0% 17.4% 1.62 0.01
25 1% 0% 15.8% 2.38 0
26 0% 0% 0% 2.01 0.9
27 0% 0% 0% 1.79 0.71
28 0% 0% 17.9% 1.87 0.23

aThe ground truth of subclinical keratoconus was based on topographic pattern interpreted by clinicians.