Characterization of postoperative LASIK ectasia features on higher-order aberration excimer ablation maps

BACKGROUND

To characterize anterior corneal higher-order aberration (HOA) excimer ablation map patterns in postoperative LASIK ectasia (POE) and to examine correlations between newly identified corneal HOA ablation map features of POE and known topographic indices.

METHODS

Prospective multicenter non-interventional descriptive study. A total of 28 eyes from 22 POE patients were enrolled. The postoperative HOA ablation map was derived from Topolyzer Vario corneal imaging at the time of POE diagnosis. Features that recurred were identified and then analyzed. Correlations to Orbscan indices were studied.

RESULTS

An arrangement of two elliptical paracentral ablation islands, deep inferior and shallow superior, in direct mirror-like opposition to each other, were identified on all HOA maps. The paracentral islands were accompanied by peripheral ablation crescents. The deep paracentral inferior island 'hot spot' coincided with the topographical apical POE cone and was highly reproducible in angular position (249.3 ± 17.3°). There was significant variation in ablation depth (shallow superior island: 11.5 ± 6.9 μm and deep inferior island: 32.5 ± 18.8 μm). The superior crescents had high variability in depth (34.8 ± 18.9 μm). Strong correlations were found between the corneal irregularity index and the ablation depth difference between the deep and shallow paracentral islands (R = 0.96; P < 0.0001).

CONCLUSION

The corneal HOA excimer ablation map revealed a recurring, distinct, easily recognizable pattern in POE eyes. Validated Orbscan POE indices and HOA ablation map islands showed a strong correlation. It is possible to extract useful information from the corneal HOA ablation map, potentially making it suitable for diagnosing and monitoring POE although more studies are needed.

Artificial intelligence for detecting keratoconus

BACKGROUND

Keratoconus remains difficult to diagnose, especially in the early stages. It is a progressive disorder of the cornea that starts at a young age. Diagnosis is based on clinical examination and corneal imaging; though in the early stages, when there are no clinical signs, diagnosis depends on the interpretation of corneal imaging (e.g. topography and tomography) by trained cornea specialists. Using artificial intelligence (AI) to analyse the corneal images and detect cases of keratoconus could help prevent visual acuity loss and even corneal transplantation. However, a missed diagnosis in people seeking refractive surgery could lead to weakening of the cornea and keratoconus-like ectasia. There is a need for a reliable overview of the accuracy of AI for detecting keratoconus and the applicability of this automated method to the clinical setting.

OBJECTIVES

To assess the diagnostic accuracy of artificial intelligence (AI) algorithms for detecting keratoconus in people presenting with refractive errors, especially those whose vision can no longer be fully corrected with glasses, those seeking corneal refractive surgery, and those suspected of having keratoconus. AI could help ophthalmologists, optometrists, and other eye care professionals to make decisions on referral to cornea specialists. Secondary objectives To assess the following potential causes of heterogeneity in diagnostic performance across studies. • Different AI algorithms (e.g. neural networks, decision trees, support vector machines) • Index test methodology (preprocessing techniques, core AI method, and postprocessing techniques) • Sources of input to train algorithms (topography and tomography images from Placido disc system, Scheimpflug system, slit-scanning system, or optical coherence tomography (OCT); number of training and testing cases/images; label/endpoint variable used for training) • Study setting • Study design • Ethnicity, or geographic area as its proxy • Different index test positivity criteria provided by the topography or tomography device • Reference standard, topography or tomography, one or two cornea specialists • Definition of keratoconus • Mean age of participants • Recruitment of participants • Severity of keratoconus (clinically manifest or subclinical) SEARCH METHODS: We searched CENTRAL (which contains the Cochrane Eyes and Vision Trials Register), Ovid MEDLINE, Ovid Embase, OpenGrey, the ISRCTN registry, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP). There were no date or language restrictions in the electronic searches for trials. We last searched the electronic databases on 29 November 2022.

SELECTION CRITERIA

We included cross-sectional and diagnostic case-control studies that investigated AI for the diagnosis of keratoconus using topography, tomography, or both. We included studies that diagnosed manifest keratoconus, subclinical keratoconus, or both. The reference standard was the interpretation of topography or tomography images by at least two cornea specialists.

DATA COLLECTION AND ANALYSIS

Two review authors independently extracted the study data and assessed the quality of studies using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool. When an article contained multiple AI algorithms, we selected the algorithm with the highest Youden's index. We assessed the certainty of evidence using the GRADE approach.

MAIN RESULTS

We included 63 studies, published between 1994 and 2022, that developed and investigated the accuracy of AI for the diagnosis of keratoconus. There were three different units of analysis in the studies: eyes, participants, and images. Forty-four studies analysed 23,771 eyes, four studies analysed 3843 participants, and 15 studies analysed 38,832 images. Fifty-four articles evaluated the detection of manifest keratoconus, defined as a cornea that showed any clinical sign of keratoconus. The accuracy of AI seems almost perfect, with a summary sensitivity of 98.6% (95% confidence interval (CI) 97.6% to 99.1%) and a summary specificity of 98.3% (95% CI 97.4% to 98.9%). However, accuracy varied across studies and the certainty of the evidence was low. Twenty-eight articles evaluated the detection of subclinical keratoconus, although the definition of subclinical varied. We grouped subclinical keratoconus, forme fruste, and very asymmetrical eyes together. The tests showed good accuracy, with a summary sensitivity of 90.0% (95% CI 84.5% to 93.8%) and a summary specificity of 95.5% (95% CI 91.9% to 97.5%). However, the certainty of the evidence was very low for sensitivity and low for specificity. In both groups, we graded most studies at high risk of bias, with high applicability concerns, in the domain of patient selection, since most were case-control studies. Moreover, we graded the certainty of evidence as low to very low due to selection bias, inconsistency, and imprecision. We could not explain the heterogeneity between the studies. The sensitivity analyses based on study design, AI algorithm, imaging technique (topography versus tomography), and data source (parameters versus images) showed no differences in the results.

AUTHORS' CONCLUSIONS

AI appears to be a promising triage tool in ophthalmologic practice for diagnosing keratoconus. Test accuracy was very high for manifest keratoconus and slightly lower for subclinical keratoconus, indicating a higher chance of missing a diagnosis in people without clinical signs. This could lead to progression of keratoconus or an erroneous indication for refractive surgery, which would worsen the disease. We are unable to draw clear and reliable conclusions due to the high risk of bias, the unexplained heterogeneity of the results, and high applicability concerns, all of which reduced our confidence in the evidence. Greater standardization in future research would increase the quality of studies and improve comparability between studies.

Combinations of Scheimpflug tomography, ocular coherence tomography and air-puff tonometry improve the detection of keratoconus

PURPOSE

To determine whether combinations of devices with different measuring principles, supported by artificial intelligence (AI), can improve the diagnosis of keratoconus (KC).

METHODS

Scheimpflug tomography, spectral-domain optical coherence tomography (SD-OCT), and air-puff tonometry were performed in all eyes. The most relevant machine-derived parameters to diagnose KC were determined using feature selection. The normal and forme fruste KC (FFKC) eyes were divided into training and validation datasets. The selected features from a single device or different combinations of devices were used to develop models based on random forest (RF) or neural networks (NN) trained to distinguish FFKC from normal eyes. The accuracy was determined using receiver operating characteristic (ROC) curves, area under the curve (AUC), sensitivity, and specificity.

RESULTS

271 normal eyes, 84 FFKC eyes, 85 early KC eyes, and 159 advanced KC eyes were included. A total of 14 models were built. Air-puff tonometry had the highest AUC for detecting FFKC using a single device (AUC = 0.801). Among all two-device combinations, the highest AUC was accomplished using RF applied to selected features from SD-OCT and air-puff tonometry (AUC = 0.902), followed by the three-device combination with RF (AUC = 0.871) with the best accuracy.

CONCLUSION

Existing parameters can precisely diagnose early and advanced KC, but their diagnostic ability for FFKC could be optimized. Applying an AI algorithm to a combination of air-puff tonometry with Scheimpflug tomography or SD-OCT could improve FFKC diagnostic ability. The improvement in diagnostic ability by combining three devices is modest.

The Evolution of Diagnostics for Keratoconus: From Ophthalmometry to Biomechanics

PURPOSE

To enumerate the various diagnostic modalities used for keratoconus and their evolution over the past century.

METHODS

A comprehensive literature search including articles on diagnosis on keratoconus were searched on PUBMED and summarized in this review.

RESULTS

Initially diagnosed in later stages of the disease process through clinical signs and retinoscopy, the initial introduction of corneal topography devices like Placido disc, photokeratoscopy, keratometry and computer-assisted videokeratography helped in the earlier detection of keratoconus. The evolution of corneal tomography, initially with slit scanning devices and later with Scheimpflug imaging, has vastly improved the accuracy and detection of clinical and sub-clinical disease. Analyzing the alteration in corneal biomechanics further contributed to the earlier detection of keratoconus even before the tomographic changes became evident. Anterior segment optical coherence tomography has proven to be a helpful adjuvant in diagnosing keratoconus, especially with epithelial thickness mapping. Confocal microscopy has helped us understand the alterations at a cellular level in keratoconic corneas.

CONCLUSION

Thus, the collective contribution of the various investigative modalities have greatly enhanced earlier and accurate detection of keratoconus, thus reducing the disease morbidity.

Integration of Artificial Intelligence into the Approach for Diagnosis and Monitoring of Dry Eye Disease

Dry eye disease (DED) is one of the most common diseases worldwide that can lead to a significant impairment of quality of life. The diagnosis and treatment of the disease are often challenging because of the lack of correlation between the signs and symptoms, limited reliability of diagnostic tests, and absence of established consensus on the diagnostic criteria. The advancement of machine learning, particularly deep learning technology, has enabled the application of artificial intelligence (AI) in various anterior segment disorders, including DED. Currently, many studies have reported promising results of AI-based algorithms for the accurate diagnosis of DED and precise and reliable assessment of data obtained by imaging devices for DED. Thus, the integration of AI into clinical approaches for DED can enhance diagnostic and therapeutic performance. In this review, in addition to a brief summary of the application of AI in anterior segment diseases, we will provide an overview of studies regarding the application of AI in DED and discuss the recent advances in the integration of AI into the clinical approach for DED.

Impact of Degraded Optics on Monocular and Binocular Vision: Lessons from Recent Advances in Highly-Aberrated Eyes

PURPOSE

Optical imperfections of the eye, characterized by higher-order wavefront aberrations, are exaggerated in corneal disease (e.g., keratoconus) and iatrogeny (e.g., keratorefractive surgery for myopia correction, keratoplasty for optical clarity restoration). This article reviews the recent advances on this topic for a comprehensive understanding of how optical degradations in disease models impact retinal image quality and monocular and binocular visual performance.

METHODS

Published literature over the last decade on retinal image quality and/or monocular and binocular visual functions with corneal irregularity was reviewed based on their relevance to the current topic, study population and strength of study design. The literature was summarized into four themes: 1) wavefront errors and retinal image quality of highly aberrated eyes, 2) monocular and binocular vision loss consequent to degraded optics and visual strategies to optimize performance, 3) impact of optical correction modalities on visual performance and 4) implications for clinical management of patients.

RESULTS

Across the 46 articles reviewed, the results clearly indicated that an increase in higher-order aberrations across these conditions had a significant negative impact on the patient's retinal image quality, and monocular and binocular visual functions. Interocular differences in retinal image quality deteriorated visual performance more than an overall worsening of image quality bilaterally. Minimizing optical degradation using rigid contact lenses and adaptive optics technology significantly improves retinal image quality and monocular and binocular vision, but performance remains sub-optimal relative to age-similar healthy controls.

CONCLUSION

Corneal disease and iatrogeny are useful models to understand the impact of optical degradation on retinal image quality and visual performance. Clinical management will greatly benefit from equalizing retinal image quality of both eyes of these patients. Future studies that deepen our understanding of the structure-function relation in these conditions are desirable for advancing vision science in this area and for developing novel clinical management strategies.

Keratoconus Classifier for Smartphone-based Corneal Topographer

Keratoconus is a severe eye disease that leads to deformation of the cornea. It impacts people aged 10-25 years and is the leading cause of blindness in that demography. Corneal topography is the gold standard for keratoconus diag-nosis. It is a non-invasive process performed using expensive and bulky medical devices called corneal topographers. This makes it inaccessible to large populations, especially in the Global South. Low-cost smartphone-based corneal topographers, such as SmartKC, have been proposed to make keratoconus diagnosis accessible. Similar to medical-grade topographers, SmartKC outputs curvature heatmaps and quantitative metrics that need to be evaluated by doctors for keratoconus diagnosis. An auto-matic scheme for evaluation of these heatmaps and quantitative values can play a crucial role in screening keratoconus in areas where doctors are not available. In this work, we propose a dual-head convolutional neural network (CNN) for classifying keratoconus on the heatmaps generated by SmartKC. Since SmartKC is a new device and only had a small dataset (114 sam-ples), we developed a 2-stage transfer learning strategy-using historical data collected from a medical-grade topographer and a subset of SmartKC data-to satisfactorily train our network. This, combined with our domain-specific data augmentations, achieved a sensitivity of 91.3% and a specificity of 94.2%.

Quantitative interocular comparison of total corneal surface area and corneal diameter in patients with highly asymmetric keratoconus

Keratoconus is a progressive corneal disorder which is frequently asymmetric. The aetiology of keratoconus remains unclear, and the concept of keratoconus as an ectatic disorder has been challenged recently. We carried out a retrospective study in 160 eyes of 80 patients, to evaluate and compare interocular differences in corneal diameter and surface area in patients with unilateral or highly asymmetric keratoconus (UHAKC). Calculations were performed using raw topographic elevation data derived from topographic measurements using Orbscan II, and we extrapolated surface areas up to measured corneal diameter. We also evaluated inter-eye correlation, and correlation between corneal surface area, corneal diameter and keratoconus severity. Our results showed a statistically significant but not clinically important greater corneal diameter (12.14 mm and 12.17 mm; p = 0.04), and corneal surface area (paired t-test, p < 0.0001; p = 0.0009 respectively) in more affected eyes. Inter-eye comparison revealed corneal diameter, anterior chamber depth, and corneal surface area were strongly correlated between eyes. Corneal surface area remained strongly correlated, and Bland–Altman analysis also showed strong inter-ocular agreement. Our results show that in patients with UHAKC the interocular difference in corneal diameter and corneal surface area is clinically insignificant, and are consistent with a redistribution, rather than increase, of corneal surface area with keratoconus progression.

Bilateral Differential Topography-A Novel Topographic Algorithm for Keratoconus and Ectatic Disease Screening

Purpose: The purpose of this study was to establish a novel bilateral differential topographic algorithm and assess its efficacy for screening of keratoconus and corneal ectasia before corneal refractive surgery.

Methods: One hundred and sixty-one consecutive patients (115 men and 46 women, aged 22.8 ± 6.8 years) with keratoconus, including clinical keratoconus, subclinical keratoconus, forme fruste keratoconus (FFK), and corneal ectasia (KC group) and one hundred and seventy-four consecutive patients (97 men and 77 women, aged 25.1 ± 6.7 years) with ametropia (control group) visiting the Eye and ENT hospital of Fudan University from June 2018 to April 2021 were included. Bilateral differential keratometry, elevation, and pachymetry topographies were composed based on raw topographic data obtained by a Scheimpflug imaging anterior segment analyzer. Key bilateral differential characteristic parameters were calculated. SPSS 20 (SPSS Inc., IBM) was used for statistical analyses and the receiver operating characteristic (ROC) curves were used to determine the diagnostic efficacies.

Results: Mann-Whitney tests detected that the front keratometry, front elevation, corneal pachymetry, and back elevation maximal, mean, and standard deviation values within a 1.5-mm radius of the bilateral differential topography were all significantly higher in the KC group than in the control group (all p-values <0.001). The front keratometry mean (ΔFKmean) and standard deviation (ΔFKsd) and the front elevation standard deviation (ΔFEsd) and maximal (ΔFEmax) values within a 1.5-mm radius of the bilateral differential topography yielded the four highest accuracies (area under the ROC curve = 0.985, 0.985, 0.984, and 0.983, respectively) for discriminating KC cases (including FFK cases) from normal cases. Cut-off values of 0.75 diopters (D) for the ΔFKmean, 0.67 D for the ΔFKsd, 2.9 μm for the ΔFEsd, and 14.6 μm for the ΔFEmax had the highest sensitivities (95.7, 95.0, 96.9, and 95.0%, respectively) and specificities (96.0, 97.7, 94.8, and 95.4%, respectively).

Conclusion: Bilateral differential topographic parameters may be efficient for the early detection of keratoconus and corneal ectasia secondary to corneal refractive surgery. This bilateral differential topographic algorithm may complement conventional diagnostic models by improving the sensitivity and specificity of screening for early keratoconus and ectasia before corneal refractive surgeries.

Machine Learning Algorithms to Detect Subclinical Keratoconus: Systematic Review

BACKGROUND

Keratoconus is a disorder characterized by progressive thinning and distortion of the cornea. If detected at an early stage, corneal collagen cross-linking can prevent disease progression and further visual loss. Although advanced forms are easily detected, reliable identification of subclinical disease can be problematic. Several different machine learning algorithms have been used to improve the detection of subclinical keratoconus based on the analysis of multiple types of clinical measures, such as corneal imaging, aberrometry, or biomechanical measurements.

OBJECTIVE

The aim of this study is to survey and critically evaluate the literature on the algorithmic detection of subclinical keratoconus and equivalent definitions.

METHODS

For this systematic review, we performed a structured search of the following databases: MEDLINE, Embase, and Web of Science and Cochrane Library from January 1, 2010, to October 31, 2020. We included all full-text studies that have used algorithms for the detection of subclinical keratoconus and excluded studies that did not perform validation. This systematic review followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) recommendations.

RESULTS

We compared the measured parameters and the design of the machine learning algorithms reported in 26 papers that met the inclusion criteria. All salient information required for detailed comparison, including diagnostic criteria, demographic data, sample size, acquisition system, validation details, parameter inputs, machine learning algorithm, and key results are reported in this study.

CONCLUSIONS

Machine learning has the potential to improve the detection of subclinical keratoconus or early keratoconus in routine ophthalmic practice. Currently, there is no consensus regarding the corneal parameters that should be included for assessment and the optimal design for the machine learning algorithm. We have identified avenues for further research to improve early detection and stratification of patients for early treatment to prevent disease progression.

Optical Quality in Keratoconus Is Associated With Corneal Biomechanics

PURPOSE

To evaluate the correlations between corneal biomechanical indices from dynamic Scheimpflug assessment and optical quality assessed as higher-order aberrations (HOAs) using a Hartmann-Shack ocular wavefront sensor in patients with keratoconus (KC).

METHODS

In this prospective, observational case series, the eyes with KC or KC suspect (KCS) from Osaka University Hospital, Osaka, Japan, were analyzed. Corneal biomechanical assessment was performed using Corvis ST (Oculus Optikgeräte GmbH, Wetzlar, Germany), and ocular wavefront aberrations were measured using the KR-1W (Topcon Corp, Tokyo, Japan). Correlations between the biomechanical indices and ocular HOAs were assessed. Corneal biomechanical indices included the deformation amplitude ratio within 2 mm, integrated radius, stiffness parameter at the first applanation, and the linear Corvis Biomechanical Index. Wavefront data of the central 4-mm region were expanded up to the sixth order of Zernike polynomials. The magnitudes of trefoil, coma, tetrafoil, secondary astigmatism, and spherical aberration were calculated by Zernike vector analysis and then used as ocular HOA parameters along with total HOAs.

RESULTS

Thirty-four KC eyes and 37 KCS eyes were included. KC eyes showed significant correlations between ocular HOAs and biomechanics, whereas there were few significant correlations in KCS eyes. In KC eyes, deformation amplitude ratio within 2 mm, integrated radius, and Corvis Biomechanical Index beta showed stronger correlations with coma among the wavefront parameters.

CONCLUSIONS

Corneal biomechanical indices correlated with ocular HOAs in patients with KC. In particular, there was a strong association with the increase in coma caused by inferosuperior asymmetry of the shape of the cornea in patients with KC.

Artificial intelligence applications in different imaging modalities for corneal topography

Interpretation of topographical maps used to detect corneal ectasias requires a high level of expertise. Several artificial intelligence (AI) technologies have attempted to interpret topographic maps. The purpose of this study is to provide a review of AI algorithms in corneal topography from the perspectives of an eye care professional, a biomedical engineer, and a data scientist. A systematic literature review using Web of Science, Pubmed, and Google Scholar was performed from 2010 to 2020 on themes regarding imaging modalities, their parameters, purpose, and conclusions and their samples and performance related to AI in corneal topography. We provide a comprehensive summary of advances in corneal imaging and its applications in AI. Combined metrics from the Dual Scheimpflug and Placido device could be a good starting point to try AI models in corneal imaging systems. The range of area under the receiving operating curve for AI in keratoconus detection and classification was from 0.87 to 1, sensitivity was from 0.89 to 1, and specificity was from 0.82 to 1. A combination of different types of AI applications to corneal ectasia diagnosis is recommended.

Differentiating highly asymmetric keratoconus eyes using a combined Scheimpflug/Placido device

PURPOSE

To determine the ability to differentiate between normal eyes and clinically unaffected eyes of patients with highly asymmetric keratoconus (AKC) using a Scheimpflug/Placido device.

SETTING

Tel Aviv Sourasky Medical Center and Enaim Medical Center, Israel.

DESIGN

Retrospective case-control.

METHODS

Imaging from a combined Scheimpflug/Placido device (Sirius, C.S.O.) was obtained from 26 clinically unaffected eyes of patients with frank keratoconus in the fellow eye, and 166 eyes from 166 patients with bilaterally normal corneal examinations that underwent uneventful corneal refractive surgery with at least 1 year of follow-up. Receiver operating characteristic curves were produced to calculate the area under the curve, sensitivity, and specificity of 60 metrics, and finally a logistic regression modeling was used to determine optimal variables to differentiate populations.

RESULTS

The most predictive individual metric able to differentiate between 26 eyes in the case group to 166 eye in the control group was the posterior wall inferior-superior (I-S) ratio, with an receiver operating characteristics (ROC) of 0.862. A combination model of 4 metrics (posterior wall I-S ratio in the central 3 mm, thinnest pachymetry coordinate on the x horizontal axis, posterior asymmetry and asphericity index, corneal volume) yielded an ROC of 0.936, with a sensitivity/specificity pair of 92.3%/87%. Variables related to elevation were not found significant.

CONCLUSIONS

Using a combination of metrics from a combined Scheimpflug/Placido device, a practical model for discrimination between clinically normal eyes of patients with highly AKC and normal eyes was constructed. Variables related to pachymetry and posterior cornea asymmetry were the most impactful.

Optical Coherence Tomography-Guided Femtosecond LASIK in the Setting of Corneal Scarring

Purpose

To report the outcomes of femtosecond-assisted laser in situ keratomileusis (LASIK) in patients with previous corneal scarring using optical coherence tomography (OCT) imaging to determine flap depths.

Methods

The charts of 11 eyes of 9 patients with previous off-visual axis corneal scarring that underwent femtosecond LASIK using OCT guidance for flap depth determination were retrospectively reviewed at a single private practice institution. The baseline characteristics, intraoperative findings and postoperative outcomes were analyzed.

Results

All 11 eyes underwent femtosecond laser flap creation and LASIK without any significant intraoperative complications. Uncorrected visual acuity improved postoperatively (p<0.0001) and remained stable at 3 months follow-up. None of the subjects lost any lines of best spectacle corrected visual acuity or developed any flap complications during the postoperative period.

Conclusion

The OCT-guided femtosecond laser technique described in this report can provide a safe and effective method to deliver LASIK in the setting of previous corneal scarring. Future investigations are required to further validate the findings in this study.

Application of Artificial Intelligence in the Diagnosis and Management of Corneal Diseases

Diagnosis and treatment planning in ophthalmology heavily depend on clinical examination and advanced imaging modalities, which can be time-consuming and carry the risk of human error. Artificial intelligence (AI) and deep learning (DL) are being used in different fields of ophthalmology and in particular, when running diagnostics and predicting outcomes of anterior segment surgeries. This review will evaluate the recent developments in AI for diagnostics, surgical interventions, and prognosis of corneal diseases. It also provides a brief overview of the newer AI dependent modalities in corneal diseases.

Keratoconus Features on Corneal Higher-Order Aberration Ablation Maps: Proof-of-Concept of a New Diagnostic Modality

PURPOSE

To assess the potential application of corneal higher-order aberration (HOA) excimer ablation map imaging in identifying reproducible keratoconus (KC) features and to explore if newly derived map metrics correlate to Pentacam KC indices.

METHODS

Case series of 12 eyes with KC ≥ grade 2. Topolyzer Vario corneal imaging with its resultant HOA ablation map islands were analyzed for their centroid, distance from center, angular position, orientation, sphericity, diameter, area, and maximal ablation depth. Correlations to Pentacam indices were studied.

RESULTS

All eyes showed recurrent features with an arrangement of two elliptical paracentral ablation islands, one deep inferotemporal and one shallow superonasal, in direct mirror-like opposition to each other. These were always accompanied by superior peripheral ablation crescents. The two paracentral islands had highly reproducible distance from center (1.2 ± 0.1 mm and 1.3 ± 0. 2 mm) and angular positions (246.8 ± 15.9° and 76.7 ± 7.7°), with greater variation in ablation depth (68.3 ± 33.2 µm and 17.6 ± 12.1 µm). Distance from center of the peripheral superior crescents was highly reproducible (3.3 ± 0.1 mm), with a larger range of depth (74.5 ± 37.2 µm). The deep paracentral inferotemporal island "hot spot" was coincident with the topographical apical cone. Strong correlations were found between the depth of the inferotemporal island and Pentacam indices of posterior radius curvature (PRC: R = -0.74) and Belin/Ambrosio enhanced ectasia total deviation (BAD-D: R = 0.71).

CONCLUSION

The corneal HOA ablation map revealed a recurring, distinct, easily recognizable pattern in KC eyes. There was a strong correlation between the depth of novel HOA ablation map metrics and validated Pentacam KC indices. Novel information can be extracted from the corneal HOA ablation map giving it the potential to be a new modality to diagnose and grade KC.

Subclinical keratoconus detection with three-dimensional (3-D) morphogeometric and volumetric analysis

PURPOSE

To assess the efficacy of morphogeometric and volumetric characterization of the cornea based on three-dimensional (3-D) modelling in diagnosis of subclinical keratoconus (KC).

METHODS

Cross-sectional study. Ninety-three eyes with subclinical KC with a best spectacle-corrected distance visual acuity ≥20/20 (grade zero KC according to the RETICS classification) and 109 control eyes were included. Computer-based 3-D corneal morphogeometric model was generated using raw topographic data. Distance-, area- and volume-based parameters were used for statistical analysis. Distance parameters included deviation of anterior (D_apexant )/posterior (D_apexpost ) apices and minimum thickness points (D_mctant , D_mctpost ) from corneal vertex, and D_apexant -D_apexpost difference. Areal variables were derived from anterior (A_ant ) and posterior (A_post ) corneal surfaces, sagittal plane passing through corneal apices (A_apexant , A_apexpost ) and thinnest point (A_mctant , A_mctpost ). Total corneal volume (V_total ) and volumetric distribution (with 0.1mm steps) centred to thinnest corneal point (VOL_mct ) and anterior (VOL_aap )/posterior (VOL_pap ) apices comprised the volume-based parameters.

RESULTS

In the subclinical KC group, all D values, D_apexant -D_apexpost difference, A_ant , A_post and A_apexant values were higher (p < 0.001), while A_apexpost , A_mctpost , V_total , VOL_mct , VOL_aap and VOL_pap values were lower when compared to the control group (p < 0.001). Regression analysis-based formula correctly classified 96.8% of the eyes with subclinical KC and 94.5% of the normal ones (p < 0.0001).

CONCLUSIONS

Eyes with subclinical KC seem to represent asymmetrically displaced anterior and posterior corneal apex, corneal thinning and volume loss. 3-D morphogeometric and volumetric parameters and differentiation formula can be incorporated into topography software to detect subclinical KC with high sensitivity and specificity in clinical practice.

Incorrect sleeping position and eye rubbing in patients with unilateral or highly asymmetric keratoconus: a case-control study

PURPOSE

To evaluate eye rubbing and sleeping position in patients with Unilateral or Highly Asymmetric Keratoconus (UHAKC).

METHODS

Case-control study of consecutive UHAKC patients diagnosed at the Rothschild Foundation. Controls were age- and sex-matched, randomly selected refractive surgery clinic patients. Patients self-administered questionnaires regarding their family history of keratoconus, eye rubbing, and sleeping habits. All the eyes underwent a comprehensive ocular examination. Logistic regression was used to analyze univariate and multivariate data to identify risk factors for keratoconus.

RESULTS

Thirty-three UHAKC patients and 64 controls were included. Univariate analyses showed that daytime eye rubbing [OR = 172.78], in the morning [OR = 24.3], or in eyes with the steepest keratometry [OR = 21.7] were significantly different between groups. Allergy [OR = 2.94], red eyes in the morning [OR = 6.36], and sleeping on stomach/sides [OR = 14.31] or on the same side as the steepest keratometry [OR = 94.72] were also significantly different. The multivariate model also showed statistical significance for most factors including daytime eye rubbing [OR = 134.96], in the morning [OR = 24.86], in the steepest eye [OR = 27.56], and sleeping on stomach/sides [OR = 65.02] or on the steepest side [OR = 144.02]. A univariate analysis in UHAKC group, comparing the worse and better eye, showed that eye rubbing [OR = 162.14] and sleeping position [OR = 99.74] were significantly (p < 0.001) associated with the worse eye.

CONCLUSION

Our data suggests that vigorous eye rubbing and incorrect sleeping position are associated with UHAKC. This is especially true in rubbing the most afflicted eye, and contributory sleep position, including positions placing pressure on the eye with the steepest keratometry.

Anatomical and Visual Outcomes after LASIK Performed in Myopic Eyes with the WaveLight® Refractive Suite (Alcon® Laboratories Inc., USA)

Purpose

To evaluate changes in corneal anatomy and quality of vision following LASIK refractive surgery for mild to high myopia using the WaveLight® Refractive Suite (Alcon® Laboratories Inc., USA). Setting. Rothschild Foundation, Paris, France.

Design

Prospective interventional case series.

Methods

We examined 60 myopic eyes (average SE −4.5 D, from −9.3 to −0.75 D) of 30 patients from 21.3 to 38.7 years old. Pachymetry, keratometry, Q factor, corneal aberrations, visual acuity (VA), contrast sensitivity, dry eye assessment, and quality of vision were measured preoperatively, one day (D1), and 1, 3, and 6 months postoperatively.

Results

6 months postoperatively, keratometry became flatter, and the Q factor became more oblate (from −0.18 ± 0.08 to +0.19 ± 0.06). Pachymetry decreased by 117.9 ± 62.2 µm at D1 and increased by 37.87 ± 32.6 µm between D1 and M6. Refraction was emmetropic at D1 and remained stable thereafter. Six months after surgery, VA was slightly but nonsignificantly improved (<0.05 log MAR), whereas contrast sensitivity remained unchanged. Quality of vision was not affected by surgery and was more related to dry eye symptoms than to corneal HOAs (r² = 0.49; p < 0.001 vs. r² = 0.03; p < 0.001).

Conclusions

LASIK surgery for moderate to high myopia, performed with the WaveLight® Refractive Suite, showed good postoperative outcomes, with demonstrated safety, predictability, efficiency, and stability. This is probably due to well-controlled spherical aberration and the use of large optical zones. Besides, we can assume that the patients' quality of vision depends more on the postoperative dry eye disease generated by the laser than on the induced HOAs.

Unsupervised learning for large-scale corneal topography clustering

Machine learning algorithms have recently shown their precision and potential in many different use cases and fields of medicine. Most of the algorithms used are supervised and need a large quantity of labeled data to achieve high accuracy. Also, most applications of machine learning in medicine are attempts to mimic or exceed human diagnostic capabilities but little work has been done to show the power of these algorithms to help collect and pre-process a large amount of data. In this study we show how unsupervised learning can extract and sort usable data from large unlabeled datasets with minimal human intervention. Our digital examination tools used in clinical practice store such databases and are largely under-exploited. We applied unsupervised algorithms to corneal topography examinations which remains the gold standard test for diagnosis and follow-up of many corneal diseases and refractive surgery screening. We could extract 7019 usable examinations which were automatically sorted in 3 common diagnoses (Normal, Keratoconus and History of Refractive Surgery) from an unlabeled database with an overall accuracy of 96.5%. Similar methods could be used on any form of digital examination database and greatly speed up the data collection process and yield to the elaboration of stronger supervised models.

Keratoconus Screening Based on Deep Learning Approach of Corneal Topography

Purpose

To develop and compare deep learning (DL) algorithms to detect keratoconus on the basis of corneal topography and validate with visualization methods.

Methods

We retrospectively collected corneal topographies of the study group with clinically manifested keratoconus and the control group with regular astigmatism. All images were divided into training and test datasets. We adopted three convolutional neural network (CNN) models for learning. The test dataset was applied to analyze the performance of the three models. In addition, for better discrimination and understanding, we displayed the pixel-wise discriminative features and class-discriminative heat map of diopter images for visualization.

Results

Overall, 170 keratoconus, 28 subclinical keratoconus and 156 normal topographic pictures were collected. The convergence of accuracy and loss for the training and test datasets after training revealed no overfitting in all three CNN models. The sensitivity and specificity of all CNN models were over 0.90, and the area under the receiver operating characteristic curve reached 0.995 in the ResNet152 model. The pixel-wise discriminative features and the heat map of the prediction layer in the VGG16 model both revealed it focused on the largest gradient difference of topographic maps, which was corresponding to the diagnostic clues of ophthalmologists. The subclinical keratoconus was positively predicted with our model and also correlated with topographic indexes.

Conclusions

The DL models had fair accuracy for keratoconus screening based on corneal topographic images. The visualization mentioned in the current study revealed that the model focused on the appropriate region for diagnosis and rendered clinical explainability of deep learning more acceptable.

Translational Relevance

These high accuracy CNN models can aid ophthalmologists in keratoconus screening with color-coded corneal topography maps.

Machine learning helps improve diagnostic ability of subclinical keratoconus using Scheimpflug and OCT imaging modalities

Purpose

To develop an automated classification system using a machine learning classifier to distinguish clinically unaffected eyes in patients with keratoconus from a normal control population based on a combination of Scheimpflug camera images and ultra-high-resolution optical coherence tomography (UHR-OCT) imaging data.

Methods

A total of 121 eyes from 121 participants were classified by 2 cornea experts into 3 groups: normal (50 eyes), with keratoconus (38 eyes) or with subclinical keratoconus (33 eyes). All eyes were imaged with a Scheimpflug camera and UHR-OCT. Corneal morphological features were extracted from the imaging data. A neural network was used to train a model based on these features to distinguish the eyes with subclinical keratoconus from normal eyes. Fisher’s score was used to rank the differentiable power of each feature. The receiver operating characteristic (ROC) curves were calculated to obtain the area under the ROC curves (AUCs).

Results

The developed classification model used to combine all features from the Scheimpflug camera and UHR-OCT dramatically improved the differentiable power to discriminate between normal eyes and eyes with subclinical keratoconus (AUC = 0.93). The variation in the thickness profile within each individual in the corneal epithelium extracted from UHR-OCT imaging ranked the highest in differentiating eyes with subclinical keratoconus from normal eyes.

Conclusion

The automated classification system using machine learning based on the combination of Scheimpflug camera data and UHR-OCT imaging data showed excellent performance in discriminating eyes with subclinical keratoconus from normal eyes. The epithelial features extracted from the OCT images were the most valuable in the discrimination process. This classification system has the potential to improve the differentiable power of subclinical keratoconus and the efficiency of keratoconus screening.

Artificial intelligence for anterior segment diseases: Emerging applications in ophthalmology

With the advancement of computational power, refinement of learning algorithms and architectures, and availability of big data, artificial intelligence (AI) technology, particularly with machine learning and deep learning, is paving the way for 'intelligent' healthcare systems. AI-related research in ophthalmology previously focused on the screening and diagnosis of posterior segment diseases, particularly diabetic retinopathy, age-related macular degeneration and glaucoma. There is now emerging evidence demonstrating the application of AI to the diagnosis and management of a variety of anterior segment conditions. In this review, we provide an overview of AI applications to the anterior segment addressing keratoconus, infectious keratitis, refractive surgery, corneal transplant, adult and paediatric cataracts, angle-closure glaucoma and iris tumour, and highlight important clinical considerations for adoption of AI technologies, potential integration with telemedicine and future directions.

Evaluation of the percentage tissue altered as a risk factor for developing post-laser in situ keratomileusis ectasia

PURPOSE

To assess the currently recommended percentage tissue altered (PTA) metric for its ability to screen for ectasia after laser in situ keratomileusis (LASIK).

SETTING

Gavin Herbert Eye Institute, University of California, Irvine, California, USA, and Rothschild Foundation, Paris, France.

DESIGN

Retrospective case series.

METHODS

The study used a LASIK database created by 1 surgeon for LASIK cases with normal preoperative topography that had a minimum follow-up of 24 months with complete preoperative and intraoperative data to permit the calculation of PTA values to detect eyes at risk for developing ectasia.

RESULTS

Of the eyes, 593 eyes had complete data and met the inclusion criteria. Based on measured flap thickness, 126 eyes (21%) had a PTA value of 40% or more (mean 44) and a percentage of that flap thickness accounted for the PTA (mean 66.7%; range 34% to 92%). The mean attempted laser ablation was 79.8 μm ± 29.2 (SD), and the mean residual bed thickness was 304.4 ± 29.2 μm (range 212 to 369 μm). No eye developed ectasia over a mean follow-up of 30 months.

CONCLUSIONS

The current PTA calculation when applied to a LASIK population with normal preoperative topography and flap thickness measured with ultrasound did not predict the risk for ectasia. Differences between study populations and assumptions might have accounted for the different outcomes obtained in the initially published PTA study.

Analysis of cases and accuracy of 3 risk scoring systems in predicting ectasia after laser in situ keratomileusis

PURPOSE

To identify risk factors for ectasia after laser in situ keratomileusis (LASIK) by comparing the accuracy of the Ectasia Risk Score System (ERSS), Screening Corneal Objective Risk of Ectasia (SCORE) Analyzer, and percentage of tissue altered (PTA) in predicting the occurrence of ectasia.

SETTING

Multiple centers in 8 countries.

DESIGN

Retrospective case series.

METHODS

Previously unpublished post-LASIK ectasia cases were analyzed. Consecutive patients who had LASIK performed at least 5 years previously with no resultant ectasia were used as controls. Axial maps from preoperative Orbscan IIz topographies were analyzed in a masked fashion, and examination files tested with the SCORE Analyzer. The PTA values and ERSS scores were generated using available preoperative and perioperative data. Only eyes with subjectively identified normal preoperative topography were tested with the PTA. Threshold values for the SCORE, ERSS, and PTA were more than or equal to 0, 4, and 40, respectively.

RESULTS

Ectasia occurred in 31 eyes (22 patients); 79 eyes (44 patients) were used as controls. In all eyes, the sensitivity and specificity for predicting ectasia, respectively, were 67.7% and 79.7% for the ERSS and 64.5% and 100% for the SCORE. In eyes with normal topography (ectasia group, 12 eyes; controls, 64 eyes), the PTA yielded sensitivity of 33.3% and specificity of 85.9%. The area under the receiver operating characteristic curve was highest for SCORE (0.911) followed by the ERSS (0.844) and PTA (0.557).

CONCLUSIONS

The SCORE was most predictive of ectasia, achieving the best specificity; the ERSS had the best sensitivity. Further studies are required to validate the PTA as a screening metric for ectasia.

A Review of Machine Learning Techniques for Keratoconus Detection and Refractive Surgery Screening

Various machine learning techniques have been developed for keratoconus detection and refractive surgery screening. These techniques utilize inputs from a range of corneal imaging devices and are built with automated decision trees, support vector machines, and various types of neural networks. In general, these techniques demonstrate very good differentiation of normal and keratoconic eyes, as well as good differentiation of normal and form fruste keratoconus. However, it is difficult to directly compare these studies, as keratoconus represents a wide spectrum of disease. More importantly, no public dataset exists for research purposes. Despite these challenges, machine learning in keratoconus detection and refractive surgery screening is a burgeoning field of study, with significant potential for continued advancement as imaging devices and techniques become more sophisticated.

Distinguishing Highly Asymmetric Keratoconus Eyes Using Dual Scheimpflug/Placido Analysis

PURPOSE

To identify the best metrics or combination of metrics that provide the highest predictive power between normal eyes and the clinically unaffected eye of patients with highly asymmetric keratoconus using data from a Dual Scheimpflug/Placido device.

DESIGN

Retrospective case-control study.

METHODS

Combined Dual Scheimpflug/Placido imaging was obtained from the Galilei G4 device (Ziemer Ophthalmic Systems AG, Port, Switzerland) in 31 clinically unaffected eyes with highly asymmetric keratoconus and 178 eyes from 178 patients with bilaterally normal corneal examinations that underwent uneventful LASIK with at least 1 year follow-up. Receiver operating characteristic (ROC) curves were generated to determine area under the curve (AUC), sensitivity, and specificity for 87 metrics, and logistic regression modeling was used to determine optimal variable combinations.

RESULTS

No individual metric achieved an AUC greater than 0.79. A combined model consisting of 9 metrics yielded an AUC of 0.96, with 90.3% sensitivity and 92.6% specificity. Among those 9 metrics included, 5 related to corneal pachymetry: Opposite Sector Index and Anterior Height BFS Z from the anterior surface, Asphericity and Asymmetry Index, Posterior Height BFS Z, and Posterior Height BFS X from the posterior surface. The strongest variable in the model was the thinnest point location on the horizontal (x) axis.

CONCLUSION

While individual metrics performed poorly, using a combination of metrics from the combined Dual Scheimpflug/Placido device provided a useful model for differentiating normal corneas from the clinically normal eyes of patients with highly asymmetric keratoconus. Pachymetry values were the most impactful metrics.