Keratoconus Detection Based on a Single Scheimpflug Image

Parameters of a basic ophthalmic examination that can ensure proper timing of corneal crosslinking in patients with keratoconus

BACKGROUND

By the time patients with keratoconus are referred to specialists for treatment, the disease-related thinning of their corneas has already made them ineligible (< 400 μm) for corneal crosslinking (CXL).

PURPOSE

To find basic ophthalmic examination parameters that can ensure proper timing of referral for CXL.

METHODS

We reviewed cases referred to Toyama University Hospital for the treatment of keratoconus from August 2011 to May 2021 to identify the frequency of contraindication due to minimal corneal thickness (MCT) < 400 μm at first visit. We performed a receiver operator characteristic (ROC) analysis of basic exam parameters (uncorrected distance visual acuity, corrected distance visual acuity, corrected distance visual acuity with hard contact lens, sphericity, cylindricity, and/or corneal astigmatism) potentially predicting eligibility for CXL. For those with an area under the curve (AUC) > 0.8, we determined cut-off values and calculated sensitivity and specificity.

RESULTS

Analyses included 66 eyes of 38 Japanese patients aged 25.0 ± 7.1 yrs (range 12-38 yrs) (56 male eyes and 10 female eyes). Thirty percent of the patients had an MCT < 400 μm. The AUC for uncorrected distance visual acuity (UCDVA) was 0.85. A cut-off value of 1.22 (converted to decimal visual acuity: ≥ 0.06) yielded 87% sensitivity and 75% specificity. The AUC for corrected distance visual acuity (CDVA) was 0.90. A cut-off of 0.52 (converted to decimal visual acuity: ≥ 0.3) yielded 89% sensitivity and 75% specificity.

CONCLUSIONS

It is advisable to refer patients with keratoconus to a specialized facility for CXL when either of the following conditions is present: (i) UCDVA (decimal visual acuity) ≥ 0.06 or (ii) CDVA (decimal visual acuity) ≥ 0.3.

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.

Corneal tissue changes following short-term soft contact lens wear of different materials

PURPOSE

To study the effect of different soft contact lens (CL) materials during short-term wear on corneal tissue.

METHODS

Twenty-two healthy participants wore both silicone hydrogel (MyDay, CooperVision) and hydrogel soft CLs (Biomedics 1 day extra, CooperVision) for 8 h per lens. In each session, Scheimpflug images were captured before and immediately after CL removal. Images were analysed using the densitometry distribution analysis, a technique from which two parameters, α (corneal transparency) and β (corneal homogeneity), were estimated. In addition, the central corneal thickness changes after CL wear and the influence of the CL material on corneal transparency were evaluated.

RESULTS

The β parameter (homogeneity) increased by 5% after wearing both CL materials (paired t-test, p < 0.001). However, the α parameter (transparency) only increased in half of the participants. No material was found to be more determinant in causing the corneal densitometry changes. Statistically significant but not clinically relevant changes in corneal thickness were observed.

CONCLUSIONS

Biomarkers of corneal tissue integrity (α and β) were affected by short-term soft contact lens wear. The observed changes in corneal transparency and homogeneity were not clinically relevant but support the importance of participant-material biocompatibility.

Assessing and compensating for the confounding factors in Scheimpflug-based corneal densitometry

Scheimpflug-based corneal densitometry is a clinically verified method for assessing corneal transparency. Nevertheless, the estimates of corneal densitometry appear to be correlated with age and eye biometry parameters, such as the anterior chamber depth or the pupil size, and that ensues a convoluted conditional estimation problem, where it is difficult to interpret the results. This study aims at devising a methodology for compensating for such confounding factors by using, as a research platform, a commercially available Scheimpflug camera that allows exporting images in a dynamic fashion, allowing averaging the results from multiple acquisitions. Two approaches are considered, one based on appropriately normalizing the line densitometry signal and one based on image histogram equalization. Then, three parameters for describing corneal densitometry are derived including the mean value of backscatter and the scale and shape parameters of the Weibull distribution estimated in regions of interest encompassing parts of corneal stroma. The results show that, unlike the non-normalized measures, the proposed approaches lead to parameters that are not correlated with age nor the eye biometry.

Suspect glaucoma detection from corneal densitometry supported by machine learning

PURPOSE

To discriminate suspect glaucomatous from control eyes using corneal densitometry based on the statistical modeling of the pixel intensity distribution of Scheimpflug images.

METHODS

Twenty-four participants (10 suspect glaucomatous and 14 control eyes) were included in this retrospective study. Corneal biomechanics was assessed with the commercial Scheimpflug camera Corvis ST (Oculus). Sets of 140 images acquired per measurement were exported for further analysis. After corneal segmentation, pixel intensities corresponding to different corneal depths were statistically modeled for each image, from which corneal densitometry in the form of parameters α (brightness) and β (homogeneity) was derived. After data pre-processing, parameters α and β were input to six supervised machine learning algorithms that were trained, tested, and compared.

RESULTS

There exists a statistically significant difference in α and β parameters between suspect glaucomatous and control eyes (both, P < 0.05/N, Bonferroni). From the implemented supervised machine learning algorithms, the K-nearest neighbors (K-NN) algorithm reached 83.93% accuracy to discriminate between groups only using corneal densitometry parameters (α and β).

CONCLUSION

Densitometry of the anterior cornea including epithelium on its own has the potential to serve as a clinical tool for early glaucoma risk assessment.

Age-related changes in dynamic corneal backscatter observed in Scheimpflug imaging

Corneal visualization Scheimpflug technology (Corvis ST) has the potential to indirectly provide information on corneal structure via the analysis of statistical properties of the backscatter. The aim of this work was to ascertain whether there are age-related changes in the dynamics of corneal backscatter during an air-puffed induced corneal deformation. Retrospective data from Corvis ST measurements of 151 young subjects (19-30 years) and 82 older subjects (50-87 years) were considered. Each measurement consisted of 140 frames (sampling frequency: 4330 fps). For every frame the cornea was first segmented, then regions of interest, encompassing temporal, central and nasal parts of cornea were selected, to which the parameters of Weibull distribution (scale and shape) were fitted, leading to time series of the estimated parameters. Apparent differences were found between the parameters of Weibull distribution between the two considered groups that manifest themselves mostly in the nasal region of the cornea. However, those differences cannot be attributed to the age alone. For this, a normalization method is proposed that leads to a much better separation between the groups in all considered regions. Clinical Relevance- The parameters of the corneal backscat-ter are widely used to asses corneal clarity (so-called corneal densitometry). Recently the parameters of Weibull distribution fitted to the corneal backscatter data have been used to support diagnosis of keratoconus. This work contributes to the assessment of corneal clarity by identifying the apparent age-related differences in those parameters when dynamic raw data is considered highlighting the need for such parameters to be appropriately normalized. Further it shown that the shape parameter of Weibull distribution unlike the scale parameter carries that information already for the raw non-normalized data.

Influence of eye tilt on corneal densitometry

PURPOSE

To investigate whether Pentacam densitometry readings are affected by corneal tilt.

METHODS

In a prospective study, the right eyes of 86 healthy participants aged 42.8 ± 20.0 years (range 18-79 years) were imaged using Scheimpflug tomography. Elevation maps were exported to calculate corneal tilt using custom-made software, and densitometry readings were acquired directly from the corneal densitometry analysis add-on to the standard software Oculus Pentacam HR. Simple mediation analysis was applied to study age as a confounding factor in the correlation between corneal tilt and corneal densitometry.

RESULTS

Corneal tilt and corneal densitometry are not independent from one another because age is significantly correlated with both corneal tilt (r = 0.50, p < 0.001) and corneal densitometry (r = 0.91, p < 0.001). Only 3.8% of the correlation between tilt and densitometry operates directly, while the remaining 96.2% depends on age.

CONCLUSIONS

Corneal tilt plays a role in corneal densitometry readings, even though the interaction is strongly influenced by age. Age is a well-known factor in densitometry readings that should be taken into consideration when interpreting Scheimpflug densitometry.

Keratoconus: An updated review

Keratoconus is a bilateral and asymmetric disease which results in progressive thinning and steeping of the cornea leading to irregular astigmatism and decreased visual acuity. Traditionally, the condition has been described as a noninflammatory disease; however, more recently it has been associated with ocular inflammation. Keratoconus normally develops in the second and third decades of life and progresses until the fourth decade. The condition affects all ethnicities and both sexes. The prevalence and incidence rates of keratoconus have been estimated to be between 0.2 and 4,790 per 100,000 persons and 1.5 and 25 cases per 100,000 persons/year, respectively, with highest rates typically occurring in 20- to 30-year-olds and Middle Eastern and Asian ethnicities. Progressive stromal thinning, rupture of the anterior limiting membrane, and subsequent ectasia of the central/paracentral cornea are the most commonly observed histopathological findings. A family history of keratoconus, eye rubbing, eczema, asthma, and allergy are risk factors for developing keratoconus. Detecting keratoconus in its earliest stages remains a challenge. Corneal topography is the primary diagnostic tool for keratoconus detection. In incipient cases, however, the use of a single parameter to diagnose keratoconus is insufficient, and in addition to corneal topography, corneal pachymetry and higher order aberration data are now commonly used. Keratoconus severity and progression may be classified based on morphological features and disease evolution, ocular signs, and index-based systems. Keratoconus treatment varies depending on disease severity and progression. Mild cases are typically treated with spectacles, moderate cases with contact lenses, while severe cases that cannot be managed with scleral contact lenses may require corneal surgery. Mild to moderate cases of progressive keratoconus may also be treated surgically, most commonly with corneal cross-linking. This article provides an updated review on the definition, epidemiology, histopathology, aetiology and pathogenesis, clinical features, detection, classification, and management and treatment strategies for keratoconus.

Age-Related Corneal Transparency Changes Evaluated With an Alternative Method to Corneal Densitometry

PURPOSE

To compare densitometry distribution analysis (DDA), a platform-independent method to assess corneal transparency, with traditional corneal densitometry.

METHODS

A total of 196 healthy participants aged 43.3 ± 18.0 years (range 18-79 years) were recruited for assessment. All participants were assessed using the corneal densitometry analysis add-on to the standard software of the Oculus Pentacam HR. In addition, the Scheimpflug image corresponding to the horizontal meridian of each participant was exported for further analysis. For each image, corneal pixel intensities were statistically modeled. The estimated output parameters, α and β, were compared with the corresponding densitometry values. The analysis was performed considering 3 concentric areas and 3 layers defined at fixed corneal depths. To demonstrate the platform independence of the DDA method, a randomly selected subset of 80 participants also had their eye measured with Oculus Corvis ST.

RESULTS

α and β were found to be well correlated with densitometry, especially α (overall cornea; r = 0.89, P < 0.001), independent of the corneal region investigated. Changes in α, β, and corneal densitometry were correlated with age.

CONCLUSIONS

In this work, we presented the relationship of DDA with age and traditional corneal densitometry. The α and β parameters, the output of DDA, are platform independent and can be used to investigate corneal clarity objectively.

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.

Detection of Subclinical Keratoconus With a Validated Alternative Method to Corneal Densitometry

Purpose

To enhance the current standards of subclinical keratoconus screening based on the statistical modeling of the pixel intensity distribution of Scheimpflug images.

Methods

Scheimpflug corneal tomographies corresponding to 25 corneal meridians of 60 participants were retrospectively collected and divided into three groups: controls (20 eyes), subclinical keratoconus (20 eyes), and clinical keratoconus (20 eyes). Only right eyes were selected. After corneal segmentation, pixel intensities of the stromal tissue were statistically modeled using a Weibull probability density function from which parameter α (pixel brightness) was derived. Further, data were transformed to polar coordinates, smoothed, and interpolated to build a map of the corneal α parameter. The discriminative power of the method was analyzed using receiver operating characteristic curves.

Results

The proposed platform-independent method achieved a higher performance in discriminating subclinical keratoconus from control eyes (90.0% sensitivity, 95.0% specificity, 0.97 area under the curve [AUC]) than the standard method (Belin-Ambrósio enhanced ectasia display), which uses only corneal morphometry (85.0% sensitivity, 85.0% specificity, 0.80 AUC).

Conclusions

Analysis of light backscatter at the cornea successfully discriminates subclinical keratoconus from control eyes, upgrading the results previously reported in the literature.

Translational Relevance

The proposed methodology has the potential to support clinicians in the detection of keratoconus before showing clinical signs.

Scheimpflug-Based Analysis of the Reflectivity of the Cornea in Marfan Syndrome

Purpose

We sought to investigate corneal reflectivity in Marfan syndrome (MFS) on the basis of Scheimpflug light intensity distribution.

Methods

In a retrospective case-control analysis, the left eyes of 40 MFS patients and 40 age- and refraction-matched healthy controls were investigated. Patients with MFS meeting the Ghent II diagnostic criteria and with genetic confirmation of disease were included. Exclusion criteria were the following: coexisting corneal, conjunctival, or scleral pathology; use of medication known to affect corneal transparency; history of ocular surgery; and insufficient data. Scheimpflug tomography images were exported to analyze corneal transparency in different corneal layers and regions. Each corneal image was automatically segmented, after which the corresponding pixel intensities in the defined regions of interest were statistically modeled using a Weibull probability density function from which parameters α (transparency) and β (homogeneity) were derived.

Results

The cornea in MFS showed significantly higher light reflectivity (overall cornea, α = 71 ± 17 arbitrary units (a.u.)) than in the control group (overall cornea, α = 59 ± 15 a.u.) (t test, P = 0.003). The α parameter was significantly higher in MFS eyes in all examined layers and regions (P < 0.05), whereas the β parameter showed no statistical difference between MFS and controls (P > 0.05). The difference in α did not correlate with ocular biometric properties (corneal thickness and curvature) or ectopia lentis (P > 0.05).

Conclusions

The cornea in MFS shows significantly higher reflectivity than healthy controls with similar levels of homogeneity.

Translational Relevance

The proposed methodology detects corneal reflectivity changes in MFS not available from regular slit-lamp examination.

Corneal tissue properties following scleral lens wear using Scheimpflug imaging

Purpose

To investigate the effect of short‐term scleral lens wear on the corneal stroma at a macroscopic (thickness) and microscopic (within tissue) level, including regional variations.

Methods

Fourteen young, healthy participants wore a rotationally symmetric, 16.5 mm diameter, scleral lens for 8 h. Scheimpflug images were captured before, and immediately after, lens wear, and also on a second day (without lens wear) to quantify natural corneal diurnal variations. After corneal segmentation, pixel intensities of the stromal tissue were statistically modelled using a Weibull probability density function from which parameters α and β were derived.

Results

Both α and β parameters increased significantly following scleral lens wear (by 5.7 ± 10% and 6.5 ± 6.5%, respectively, both p < 0.01). Corneal thickness also increased slightly following lens wear (mean increase 0.49 ± 1.77%, p = 0.01); however, the change in α and β parameters did not correlate with the magnitude of corneal swelling. On the control day, small but significant corneal thinning was observed (−0.82 ± 1.1%, p = 0.03), while α and β parameters remained stable. Both microparameters varied significantly across the cornea, with α decreasing (−15.4 ± 0.7%) and β increasing towards the periphery (+4.4 ± 2.6%) (both p < 0.001).

Conclusion

Corneal microparameters α and β varied regionally across the cornea and displayed a statistically significant increase following short‐term scleral lens wear, but remained stable between morning and evening measurements taken during a control day without lens wear. These corneal microparameters may be a useful metric to quantify subclinical corneal changes associated with low level hypoxia.