Computer-aided diagnosis of keratoconus through VAE-augmented images using deep learning.

Detecting clinical keratoconus (KCN) poses a challenging and time-consuming task. During the diagnostic process, ophthalmologists are required to review demographic and clinical ophthalmic examinations in order to make an accurate diagnosis. This study aims to develop and evaluate the accuracy of deep convolutional neural network (CNN) models for the detection of keratoconus (KCN) using corneal topographic maps. We retrospectively collected 1758 corneal images (978 normal and 780 keratoconus) from 1010 subjects of the KCN group with clinically evident keratoconus and the normal group with regular astigmatism. To expand the dataset, we developed a model using Variational Auto Encoder (VAE) to generate and augment images, resulting in a dataset of 4000 samples. Four deep learning models were used to extract and identify deep corneal features of original and synthesized images. We demonstrated that the utilization of synthesized images during training process increased classification performance. The overall average accuracy of the deep learning models ranged from 99% for VGG16 to 95% for EfficientNet-B0. All CNN models exhibited sensitivity and specificity above 0.94, with the VGG16 model achieving an AUC of 0.99. The customized CNN model achieved satisfactory results with an accuracy and AUC of 0.97 at a much faster processing speed compared to other models. In conclusion, the DL models showed high accuracy in screening for keratoconus based on corneal topography images. This is a development toward the potential clinical implementation of a more enhanced computer-aided diagnosis (CAD) system for KCN detection, which would aid ophthalmologists in validating the clinical decision and carrying out prompt and precise KCN treatment.

Changes in corneal biomechanical parameters in keratoconus eyes with various severities after corneal cross-linking (CXL): A comparative study.

OBJECTIVES: To compare changes in corneal biomechanical parameters one year after corneal cross-linking (CXL) in keratoconus (KCN) eyes of different severities. METHODS: Seventy-five eyes with mild, moderate, and severe grades of KCN (n = 24, 31, and 20 eyes, respectively) that were treated with CXL, based upon the standard Dresden protocol, were included. The corneal biomechanical assessment was performed using Corvis ST and Ocular Response Analyzer (ORA). Changes in Corvis's dynamic corneal response (DCR) parameters and ORA's derived parameters (corneal hysteresis (CH), and corneal resistance factor (CRF)) were assessed whilst the corneal thickness and intraocular pressure were considered as covariates. RESULTS: There was no statistically significant difference in the corneal biomechanical parameters obtained using both devices after surgery separately in different KCN grades, except for the deformation amplitude (DA) in the severe KCN group (P = 0.017). Changes in the classic parameters of the highest concavity phase of Corvis ST (peak distance, radius, and DA) were more positive and in the newer parameters (integrated inverse radius (IIR), deformation amplitude ratio (DAR)) more negative in the severe group compared to the other groups. Also, the mean change in CH (P = 0.710), and CRF (P = 0.565), showed a negative shift in higher grades of KCN; however, there was no significant difference in the mean changes of all parameters between different groups. (P > 0.05). CONCLUSIONS: Similar changes in the Corvis ST and ORA parameters in mild, moderate, and severe KCN indicate biomechanical stability and the effective role of CXL in stopping the progressive nature of keratoconus in eyes of varying severities one year after CXL.

Comparative analysis of two different types of intracorneal implants in keratoconus: A corneal tomographic study.

OBJECTIVE: To evaluate changes in visual acuity and corneal tomographic outcomes at 6 months after femtosecond-laser assisted implantation of two different types of intracorneal implants in keratoconus. METHODS: A total of 39 keratoconus eyes implanted with two segments KeraRing (n = 22 eyes) or MyoRing (n = 17 eyes) were enrolled. Tomographic data (Pentacam system, Oculus) were analyzed and correlated with enhancement in uncorrected (UDVA) and corrected distance visual acuity (CDVA). The tomographic indices were front maximum keratometry (Kmax), corneal asphericity (Q-value) on both surfaces, average pachymetric progression indices (PPI), maximum Ambrosio relational thickness (ARTmax), Belin-Ambrósio enhanced ectasia total deviation index (BAD-D), index of surface variance (ISV), vertical asymmetry (IVA), height asymmetry (IHA) and height decentration (IHD). RESULTS: LogMAR UDVA and CDVA improved 2.1 (p = 0.003) and 0.7 (p = 0.074) lines with KeraRing, and 8 and 2.5 lines with MyoRing (p = 0.001). The highest and lowest mean differences in the tomographic indices between both groups were related to ISV and IHD, respectively. Changes in all indices differed significantly between two groups except for changes in front corneal astigmatism, ARTmax, ISV, IVA, IHD and IHA (p > 0.05). Correlation of changes in CDVA with changes in other parameters was statistically significant only for IHD in the KeraRing group, while changes in in UDVA were significantly correlated with changes in spherical equivalent, back Q-value, ISV, IVA, and IHA only in the MyoRing group. CONCLUSION: Both implants promote corneal shape regularization and an enhancement in UDVA in keratoconus. A considerable flattening effect and reduction in prolateness in the front corneal surface were observed with MyoRing.

Anatomical and Visual Effects of the MyoRing Implantation Measured by the ABCD Keratoconus Grading System.

OBJECTIVE: To assess the anatomical changes produced by implantation of the complete intracorneal ring (MyoRing; DIOPTEX, GmbH, Linz, Austria) on the different corneal anatomical layers as measured by the ABCD keratoconus staging system 6 months after operation. METHODS: Seventeen eyes of 17 keratoconic patients implanted with MyoRing using the femtosecond laser were assessed preoperatively and postoperatively. Distance uncorrected visual acuity and distance corrected visual acuity (DCVA), refraction, and Scheimpflug tomography with Pentacam HR were determined for each patient. The 4 ABCD parameters were recorded, which include the anterior and posterior radius of curvature from a 3.0-mm optical zone centered on the thinnest point (TP) (A & B, respectively), the thinnest corneal thickness (C), and DCVA (D). RESULTS: Statistically significant improvement (preoperatively vs. postoperatively) in the DCVA (0.40±0.21 vs. 0.67±0.20 in decimal notation, P=0.001) associated with significant flattening effect was observed in both anterior (6.02±0.40 vs. 7.18±0.54 mm, P<0.001) and posterior (4.49±0.41 vs. 4.66±0.40 mm, P=0.001) radius of curvature. There was no significant change in the corneal thickness at the TP (P=0.981). The largest magnitude of change occurred on the anterior surface as measured by the "A" parameter. The average ABCD keratoconus staging before MyoRing implantation was A4B4C2D2 and changed to A1B4C2D1 6 months after ring implantation. CONCLUSION: Despite the placement of MyoRings in the posterior one-third of the cornea, the greatest changes in curvature occurred on the anterior corneal surface. The ABCD classification and staging system allows for a differential assessment of the different anatomical layers and may prove useful in better understanding of the geometric (structural) and functional changes after MyoRing implantation, as well as with other corneal procedures.

Predictors of Successful Outcome following Intrastromal Corneal Ring Segments Implantation.

Purpose: To assess the preoperative visual, refractive, corneal topo/tomographic, aberrometric and biomechanical parameters as predictive factors of a successful outcome 6 months following intrastromal corneal ring segments implantation. Methods: Sixty-eight keratoconus eyes implanted with Keraring using femtosecond laser technology were assessed. The preoperative assessed parameters included uncorrected and corrected distance visual acuity (UDVA & CDVA), refraction, placido-disk based topography using TMS-4, Scheimplfug tomography using Pentacam HR, corneal biomechanical assessments using Ocular Response Analyzer (ORA) and the wavefront analysis using i-Trace aberrometer. Other variables were type of astigmatism based on orientation of the steep meridian, keratoconus staging based on the Amsler-Krumeich classification and the difference between the axes of refractive astigmatism, topographic astigmatism and comatic aberration based on a difference less or more than 30°. The success criterion was defined based on CDVA, a post-operative CDVA improvement at least two lines were considered as a success and otherwise were recognized as a failure following Keraring implantation. Results: Only UDVA, coincidence of the most elevated points on the front and back corneal surfaces and the difference between UDVA and CDVA showed significant difference between the eyes with successful outcomes and those with unsuccessful results (P < 0.05). Although corneal curvature and astigmatism were higher and corneal thickness was lower in the unsuccessful group, differences were not statistically significant. Conclusion: It is expected that the greater difference between the preoperative uncorrected and corrected distance visual acuity (Δ UDVA-CDVA) and more coincidence of the most elevated points in the two corneal surfaces on the elevation maps increase the rate of successful outcome following the Keraring implantation. Abbreviations: ICRS: intrastromal corneal ring segments; UDVA: uncorrected distance visual acuity; CDVA: corrected distance visual acuity; logMAR: logarithm of minimum angle of resolution; SE: spherical equivalent; IOP: intra-ocular pressure; Km: mean keratometry; CA: corneal astigmatism; WTR: with-the-rule; ATR: against-the-rule; OBL: oblique; SB: symmetric bow-tie; AB: asymmetric bow-tie; IS: inferior steepening; SRAX: skewed radial axis; Q: asphericity (Q-value); CCT: central corneal thickness; CTP: corneal thinnest point; ORA: ocular response analyzer; CH: corneal hysteresis; CRF: corneal resistance factor; HOAs: higher-orders aberrations; RMS: root mean square; SD: standard deviation; CI: confidence interval.

Changes in the ABCD Keratoconus Grade After Intracorneal Ring Segment Implantation.

PURPOSE: To assess the changes in the ABCD keratoconus staging system 6 months after intracorneal ring segment implantation. METHODS: Fifty eyes of 50 patients with keratoconus who were implanted with the Keraring (Mediphacos, Belo Horizonte, Brazil) using the femtosecond laser were assessed. Preoperative and postoperative assessments included determination of distance uncorrected visual acuity and distance corrected visual acuity (DCVA), refraction, and Scheimpflug tomography with the Pentacam HR. In the ABCD keratoconus staging system, the elements A, B, C, and D stand for anterior and posterior radii of curvature in a 3.0-mm zone centered on the thinnest point (TP), corneal thickness at the TP, and DCVA, respectively. RESULTS: Keraring implantation produced significant flattening changes (preoperatively vs. postoperatively) in the anterior (6.60 ± 0.48 vs. 7.22 ± 0.57 mm, P < 0.001) and posterior (4.99 ± 0.47 vs. 5.16 ± 0.53 mm, P = 0.002) radii of curvature of the 3-mm zone centered on the corneal TP associated with a statistically significant improvement in the DCVA (0.56 ± 0.24 vs. 0.70 ± 0.22 in the decimal notation, P = 0.001) with no significant change in the corneal thickness at the TP (P = 0.285). The most changes occurred in element A of the ABCD keratoconus classification. Also, a 1-stage change was observed for element B, whereas elements C and D did not show changes in their postoperative stages after ring implantation. ABCD keratoconus staging before ring implantation was A2B3C2D1 and changed to A1B2C2D1 6 months after surgery. CONCLUSIONS: The ABCD staging system provides a more comprehensive guide that better illustrates the structural changes and visual acuity as one aspect of visual function after the implantation of the intrastromal corneal ring segments.