|
| Author, year | Device | Significant indices or algorithms | Comparison eyes | Results |
|
| Li et al. 2012 [7] | SD-OCT
6 mm D | RMSV: root mean square of variation
RMSPD: root mean square pattern deviation | 76 NL
35 KCN | All indices were positive in KCN
RMSPD has the greatest AUC |
|
| Silverman et al. 2014 [104] | VHF-US
10 mm | 6 variables, including 4 ETM variables analyzed by linear discriminant analysis (LDA)
Neural network (NN) analysis | 130 NL
74 KCN | LDA: 100% AUC
NN: 100% AUC |
|
| Temstet et al. 2015 [8] | SD-OCT
6 mm | Location and CET of thinnest | 42 NL
36 FFK
32 severe
KCN | Inferior thinnest CET in 91.3%
Thinnest central CET was thinner in FFK compared to NL |
|
| Catalan et al. 2016 [105] | SD-OCT
6 mm | Combination of 7 ETM and other pachymetry variables | 104 NL
22 FFK
22 KCN | No good discrimination of individual ETM variables
A combination of ETM and pachymetry had good discrimination power |
|
| Tang et al. 2016 [106] | SD-OCT
6 mm D | Epi-PSD (pattern standard deviation)
Ant. ectasia index
Warpage index | 22 NL
31 KCN
11 CLW
8 FFK | Epi-PSD: NL <4.1% vs. KCN, CLW, FFK >4.1%
Ant. ectasia index: KCN > FFK > CLW > NL
Warpage index: positive in CLW vs. negative in KCN, FFK |
|
| Li et al. 2016 [107] | SD-OCT
6 mm D | Epi-PSD
Corneal PSD
Stromal PSD | 83 NL
50 S-KCN
1 FFK | All 3 parameters were successful
Epi-PSD had the greatest AUC |
|
| Xu et al. 2016 [55] | UHR-OCT | EEI: epithelium ectasia index
BEI: Bowman’s layer ectasia index
EEI-max: maximum epithelium ectasia index
BEI-max: maximum BL ectasia index | 81 NL
37 KCN
32 FFK | EEI-max and BEI-max had the highest power of discrimination |
|
| Silverman et al. 2017 [108] | VHF-US and Scheimpflug device | 3 variables of VHF-US including 2 ETM variables and 4 variables of Scheimpflug imaging | 111 NL
30 KCN | 97.3% specificity
100% sensitivity |
|
| Hwang et al. 2018 [109] | SD-OCT 6 mm D and Scheimpflug imaging | Two ETM parameters (ETM SD and greater min-max) and 11 other OCT pachymetry and Scheimpflug indices | 60 NL
30 FFK | None of the individual ETM parameters showed a good discrimination power
Combination of parameters: 100% sensitivity and 100% specificity |
|
| Vega-Strada et al. 2019 [46] | SD-OCT+
Placido disc | Thinner 3 mm central
Greater S-I (8 mm)ratio
Greater S-I (6 mm)ratio | 60 NL
107 KCN | Combined 3 parameters: AUC: 0.92 |
|
| Pircher et al. 2019 [54] | UHR-OCT | BLTM
Minimum BL thickness
R1ET (thinnest to thickest ET) | 20 NL
47 KCN | Moth-like change in BLTM
Min BLT: AUC: 0.983
R1ET: AUC: 0.926 |
|
| Pavlatos et al. 2020 [110] | SD-OCT | Coincident thinning index | 82 NL
133 KCN, S-KCN, FFK | CTN: KCN > S-KCN > FFK > NL |
|
| Yang et al. 2020 [111] | SD-OCT | 2-step decision tree | 54 NL
19 FFK
11 S-KCN
91 KCN | NL: 100% specificity
KCN: 97.8% sensitivity
S-KCN: 100% sensitivity
FFK: 73.3% sensitivity |
|
| Pavlatos et al. 2021 [112] | SD-OCT
6 mm | Epi-MI (modulation index)
Epi-PSD | 32 NL
20 CLW
89 KCN
16 S-KCN
26 FFK | Epi-PSD: NL<4.1%, KCN, CLW, FFK> 4.1%
Epi-MI: NL=CLW << FFK < S-KCN < KCN |
|
| Toprak et al. 2021 [113] | SD-OCT+
Placido disc | E/S (epithelium to stroma) ratio | 66 NL
27 FFK
55 S-KCN | ETM parameters failed to show a good discrimination power between FFK or S-KCN and normal |
|
| Yücekul et al. 2022 [44] | SD-OCT | 2-step decision tree | 172 NL
20 S-KCN
172 KCN | 100% sensitivity, 100% specificity in KCN
90.4% sensitivity in S-KCN |
|
| Shi et al. 2022 [114] | UHR-OCT and Scheimpflug imaging | Epithelial pattern variation (EPV)
Analysis: neural network
Logistic regression | 50 NL
33 S-KCN
38 KCN | EPV: best discrimination power
Combination OCT and Scheimpflug: AUC about 0.9 |
|
