Smart decision support system for keratoconus severity staging using corneal curvature and thinnest pachymetry indices

Table 6 Comparison with state-of-the-art KC severity staging techniques (as of 2018)

Authors	Year	Model used	Imaging device used	Dataset used		Performance metrics (%)		Input data type	Input feature-set
Authors	Year	Model used	Imaging device used	Feature set	Subjects used	Accuracy	Other metrics	Input data type	Input feature-set
Yousefi et al. [95]	2018	Unsupervised ML	SS-1000 CASIA (OCT)	420	3156	n/a	Sen: 97.7; Spe: 94.1-	IM	8
Lavric et al. [40]	2020	SVM + 24 others	SS-1000 CASIA (OCT)	8	3151	94	Sen: 89.5; Spe: 96	CP	8
Cao et al. [96]	2020	RF + 7 others	Pentacam + SS-1000 CASIA (OCT)	11	88	97	Sen: 94.0; Spe: 90.0	CP	11
Issaarti et al. [97]	2020	FNN	Pentacam	49	812	Suspect KC: 74 Other stages: 97.6 (ave)	Suspect KC: AUC: 87.0; Sen: 85.2; Spe: 70.0	CP	n/a
Hallett et al. [98]	2020	MLP + Unsupervised	Pentacam	29	124	MLP: 73 Unsupervised: 80	AUC: 89	CP	29
Aatila et al. [99]	2021	RF	SS-1000 CASIA (OCT)	446	3162	4-classes: 95	Sen: 98: Pre: 98	CP	10
Malyugin et al. [100]	2021	QDA	Pentacam	490	47,419	97	AUC: 95	CP	7
Lavric et al. [101]	2021	SVM + others	Pentacam	38	5881	n/a	AUC: 5-classes: 88	CP	3
Kamiya et al. [102]	2021	CNN	Placido topographer	n/a	179	87.78 (ave)	AUC 93.68 (ave)	IM	n/a
Shetty et al. [103]	2021	RF	Pentacam	n/a	366	91	AUC: 93; Sen: 89; Spe: 81	CP	10
Priya et al. [104]	2022	SVM	SS-1000 CASIA (OCT)	447	3164	90	Spe: 97.7; Pre: 94.1	CP	447 (all features)
Proposed system	2023	RF + 2 others	Pentacam	79	644	5-classes: 98.62 (ave)	5-classes: Sen: 98.62; Pre: 98.70; F1: 98.66; F2: 98.64	CP	3

CNN = convolutional neural network; DT = decision tree; FNN = feedforward neural network; MLP = multilayer perception neural network; QDA = quadratic discriminant analysis; RF = random forest; SVM = support vector machine; AUC = area under the curve; F1 = F1-score; F2 = F2-score; Sen = sensitivity (or recall); Spe = specificity; Pre = precision; CP = corneal parameters; n/a = not applicable; ave = average; IM = image (topography, tomography or Placido); OCT = optical coherence tomography

ISSN: 2326-0254