Early Detection Of Keratoconus Research Articles

Matlab based graphical user interface for the monitoring and early detection of keratoconus

The rapid and extensive growth in medical imaging modalities and their applications is creating a pressing need for computers and computing in image processing, visualization, archival, and analysis. In this article, a Matlab-based graphical user interface (GUI) program is proposed for the monitoring and early detection of keratoconus. The findings show the efficiency of the proposed to detect the early stage of keratoconus. The proposed neural network model produces accuracy, ranging from 96% to 92%. It considers, respectively, 2 classes (normal cornea and keratoconus) and 3 classes (keratoconus, suspected keratoconus or normal) which will increase to 99% with respect to the 2 classes of keratoconus and 94% to the 3 classes when combining topography parameters with OCT image corneal pachymetry measurements and clinical judgments. 
 The compatibility of the graphical interface components with common medical data and image analysis tools facilitates the involvement of the ophthalmologist in the digitization of the medical records, the image processing and the conception of multimodal artificial intelligence applications for medical imaging.

Deep learning in early keratoconus/corneal ectasia

Machine learning algorithms have been gaining tremendous relevance in the last decade in more or less all disciplines in medicine. In most cases supervised learning techniques are used to perform classification or graduation of diseases. Early detection of keratoconus in a screening environment is crucial as in this stage several therapeutic options such as collagen crosslinking or intracorneal ring implantation are available which could prevent further progression of the disease. In this talk we focus on new techniques of automatic identification of ectatic corneal diseases such as keratoconus in an early stage based on anterior segment optical coherence measurements with the Casia 2 tomographer. A neural network approach is implemented and fed with data of corneal front surface height and curvature, back surface height and curvature, and thickness data (in total 5 data layers) and the similarity of the measurement to normal or abnormal situation is provided. Performance data are shown for our training, validation, and test dataset.

Combined scheimpflug and optical coherence tomography imaging in screening of keratoconus in children with Down syndrome

Purpose The aim of this study was to screen for keratoconus (KC) in a sample of children with Down syndrome (DS) and to correlate the changes of combined Scheimpflug and optical coherence tomography imaging in these children and to compare it with an age-matched control group with healthy cornea. Patients and methods This was a prospective, case–control, analytic study that included 52 eyes of 26 children, who were divided into two groups: 13 with DS (group A) and 13 age-matched healthy controls (group B). Both groups were evaluated using anterior segment optical coherence tomography (AS-OCT) and Scheimpflug camera topographer (Sirius, CSO). Results KC-like features were detected in 61.5% of the patients in the DS group. Assessment of the corneal parameters showed statistically significant correlation between Sirius topography and AS-OCT epithelium and pachymetry maps parameters in the DS group. Conclusion Corneal features compatible with the diagnosis of KC were common findings in children with DS. Combined use of AS-OCT epithelial mapping and corneal topography is of high significance and can be used for screening and early detection of KC in patients with DS.

Transfer Learning in Keratoconus Classification

Early detection of keratoconus will provide more treatment choices, avoid heavy treatments, and help stop the rapid progression of the disease. Unlike traditional methods of keratoconus classification, this study presents a machine learning-based keratoconus classification approach, using transfer learning, applied on corneal topographic images. Classification is performed considering the three corneal classes already cited : normal, suspicious and keratoconus. Keratoconus classification is carried out using six pretrained convolutional neural networks (CNN) VGG16, InceptionV3, MobileNet, DenseNet201, Xception and EfficientNetB0. Each of these different classifiers is trained individually on five different datasets, generated from an original dataset of 2924 corneal topographic images. Original corneal topographic images have been subjected to a special preprocessing before their use by different models in the learning phase. Images of corneal maps are separated in five different datasets while removing noise and textual annotation from images. Most of models used in the classification allow good discrimination between normal cornea, suspicious and keratoconus one. Obtained results reached classification accuracy of 99.31% and 98.51% by DenseNet201 and VGG16 respectively. Obtained results indicate that transfer learning technique could well improve performance of keratoconus classification systems.

Prevalence of Keratoconus in a Population-Based Study in Syria.

Aim To determine the prevalence and associations of keratoconus (KC) in a university student population in Syria. Methods A prospective multicentre cross-sectional cohort study was conducted at two universities in Syria. Student volunteers were recruited from Tishreen University (Latakia governorate) and Damascus University (Damascus governorate). All participants underwent a comprehensive ocular examination. Placido/Scheimpflug-based corneal imaging using the Sirius (CSO, Florence. Italy), and a questionnaire to evaluate the baseline characteristics and medical history, as well as to highlight possible risk factors of KC. Univariate and bivariate analyses were performed. Results The estimated prevalence of KC among all subjects was 1.43% (n = 12). A strong association between eye rubbing and keratoconus was found (OR 9.33, 95% CI 2.94–29.63, P < 0.001). Damascus University participants had a higher prevalence of KC than Tishreen University. However, the difference was not statistically significant. Conclusion The prevalence of keratoconus in this Syrian student population was 1.43%. The results of this study demonstrate a high prevalence of keratoconus in the study population. Early detection of keratoconus through screening may yield benefits in preventing devastating sequelae of KC in populations with a high prevalence.

Detection of subclinical keratoconus using a novel combined tomographic and biomechanical model based on an automated decision tree

Early detection of keratoconus is a crucial factor in monitoring its progression and making the decision to perform refractive surgery. The aim of this study was to use the decision tree technique in the classification and prediction of subclinical keratoconus (SKC). A total of 194 eyes (including 105 normal eyes and 89 with SKC) were included in the double-center retrospective study. Data were separately used for training and validation databases. The baseline variables were derived from tomography and biomechanical imaging. The decision tree models were generated using Chi-square automatic interaction detection (CHAID) and classification and regression tree (CART) algorithms based on the training database. The discriminating rules of the CART model selected metrics of the Belin/Ambrósio deviation (BAD-D), stiffness parameter at first applanation (SPA1), back eccentricity (Becc), and maximum pachymetric progression index in that order; On the other hand, the CHAID model selected BAD-D, deformation amplitude ratio, SPA1, and Becc. Further, the CART model allowed for discrimination between normal and SKC eyes with 92.2% accuracy, which was higher than that of the CHAID model (88.3%), BAD-D (82.0%), Corvis biomechanical index (CBI, 77.3%), and tomographic and biomechanical index (TBI, 78.1%). The discriminating performance of the CART model was validated with 92.4% accuracy, while the CHAID model was validated with 86.4% accuracy in the validation database. Thus, the CART model using tomography and biomechanical imaging was an excellent model for SKC screening and provided easy-to-understand discriminating rules.

Keratoconus Classification with Convolutional Neural Networks Using Segmentation and Index Quantification of Eye Topography Images by Particle Swarm Optimisation.

In keratoconus, the cornea assumes a conical shape due to its thinning and protrusion. Early detection of keratoconus is vital in preventing vision loss or costly repairs. In corneal topography maps, curvature and steepness can be distinguished by the colour scales, with warm colours representing curved steep areas and cold colours representing flat areas. With the advent of machine learning algorithms like convolutional neural networks (CNN), the identification and classification of keratoconus from these topography maps have been made faster and more accurate. The classification and grading of keratoconus depend on the colour scales used. Artefacts and minimal variations in the corneal shape, in mild or developing keratoconus, are not represented clearly in the image gradients. Segmentation of the maps needs to be carried out for identifying the severity of the keratoconus as well as for identifying the changes in the severity. In this paper, we are considering the use of particle swarm optimisation and its modifications for segmenting the topography image. Pretrained CNN models are then trained with the dataset and tested. Results show that the performance of the system in terms of accuracy is 95.9% compared to 93%, 95.3%, and 84% available in the literature for a 3-class classification that involved mild keratoconus or forme fruste keratoconus.

Characteristics of pediatric keratoconus and the role of corneal topography in early diagnosis: A prospective study

Purpose: The purpose of this study is to describe the clinical and topographical features of keratoconus in children and to emphasize the role of corneal topography in early diagnosis of keratoconus in eyes with astigmatism. Materials and Methods: It was a prospective study done to screen children below 18 years with astigmatism of 1.50 Diopter or greater for presence of keratoconus. Ophthalmic evaluation included retinoscopy, cycloplegic refraction, and detailed slit lamp examination for clinical diagnosis of keratoconus. All eyes underwent corneal topography and tomography. Eyes were followed up for progression with both refraction and corneal topographic imaging. Progressive cases later underwent Accelerated Collagen Crosslinking (ACXL). Results: 700 eyes of 350 patients with astigmatism were screened for keratoconus. 44 eyes of 28 patients were diagnosed with keratoconus. Mean age of the diagnosed children was 11.7 years (6–18 years), 18 boys and 10 girls. Overall, most common clinical association was allergic conjunctivitis. Twenty-five eyes (57%) had clinical evidence of keratoconus while other 19 eyes (43%) required corneal topography to establish the diagnosis. Based on the Amsler-Muckenheim tomographical classification, 22 of 44 eyes (50%) were grouped in prestage level followed by 13 eyes (29.5%) in Stage 1, 5 eyes (11.4%) in Stage 2, and 4 eyes (9.1%) in Stage 3. Thirty-five eyes of 21 patients showed progression and underwent ACXL. Conclusion: Early detection of keratoconus in children with astigmatism is of utmost importance to avoid visual impairment and surgical intervention. Even in the absence of clinical signs, corneal topography and tomography should be performed as a screening tool to rule out keratoconus in these children.

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.

Diagnosis of Keratoconus with Corneal Features Obtained through LBP, LDP, LOOP and CSO

Keratoconus, by its name, is the condition of the eye wherein the cornea assumes a conical shape due to the thinning and protrusion of the cornea. Keratoconus, though bilateral, can be asymmetric in that it can progress differently in the eyes of the patient. Keratoconus can start from early adulthood and progress till the age of 40. Early detection of keratoconus is vital in preventing vision loss or costly repairs. The diagnostic tools available range from keratoscope to videokeratoscope but involves human efforts and thereby human errors. Automatic detection of keratoconus is required for large screening camps. With the advances in artificial intelligence techniques for medical diagnosis, new algorithms and techniques have been developed for the early and rapid screening of keratoconus, which aids clinicians in fast diagnosis. Artificial Neural Networks, Support Vector Machines, Radial Basis Function Neural Networks, Decision Trees, Computational Neural Networks, and various optimisation techniques have been used in different studies. The progression of keratoconus is identified by analyzing the shape of the cornea with Local Binary Pattern (LBP), Local Directional Pattern (LDP), Local Optimal Oriented Pattern (LOOP), and Cat Swarm Optimization (CSO) to detect the changes in cornea edges. The image processing with the CSO algorithm optimizes the result for the changes in the cornea and keratoconus detection. A new automated solution for detecting keratoconus is presented that employs texture analysis techniques such as LBP, LDP, LOOP, and CSO. The CSO extracts morphological and granular features from images of the cornea. The proposed method can be used to detect keratoconus by identifying the cornea shape change and improving clinical decisions. Further research can be in the way of grading the level of keratoconus. HIGHLIGHTS Extraction of corneal features for Diagnosis of Keratoconus from corneal topography images Feature vector extraction from corneal images using Local Binary Pattern, Local Direction Pattern and Local Optimal Oriented Pattern Optimisation using Cat Swarm Optimisation GRAPHICAL ABSTRACT

Diagnosis of Subclinical Keratoconus Based on Machine Learning Techniques.

(1) Background: Keratoconus is a non-inflammatory corneal disease characterized by gradual thinning of the stroma, resulting in irreversible visual quality and quantity decline. Early detection of keratoconus and subsequent prevention of possible risks are crucial factors in its progression. Random forest is a machine learning technique for classification based on the construction of thousands of decision trees. The aim of this study was to use the random forest technique in the classification and prediction of subclinical keratoconus, considering the metrics proposed by Pentacam and Corvis. (2) Methods: The design was a retrospective cross-sectional study. A total of 81 eyes of 81 patients were enrolled: sixty-one eyes with healthy corneas and twenty patients with subclinical keratoconus (SCKC): This initial stage includes patients with the following conditions: (1) minor topographic signs of keratoconus and suspicious topographic findings (mild asymmetric bow tie, with or without deviation; (2) average K (mean corneal curvature) < 46, 5 D; (3) minimum corneal thickness (ECM) > 490 μm; (4) no slit lamp found; and (5) contralateral clinical keratoconus of the eye. Pentacam topographic and Corvis biomechanical variables were collected. Decision tree and random forest were used as machine learning techniques for classifications. Random forest performed a ranking of the most critical variables in classification. (3) Results: The essential variable was SP A1 (stiffness parameter A1), followed by A2 time, posterior coma 0°, A2 velocity and peak distance. The model efficiently predicted all patients with subclinical keratoconus (Sp = 93%) and was also a good model for classifying healthy cases (Sen = 86%). The overall accuracy rate of the model was 89%. (4) Conclusions: The random forest model was a good model for classifying subclinical keratoconus. The SP A1 variable was the most critical determinant in classifying and identifying subclinical keratoconus, followed by A2 time.

Analysis of OPD-Scan and Pentacam Parameters for Early Keratoconus Detection

PURPOSE: To evaluate Pentacam and OPD-Scan parameters in the early detection of keratoconus. Retrospective case-control study. Case group included 50 clinically unaffected fellow eyes diagnosed with asymmetric keratoconus showing subtle qualitative changes at the 0.5-D sensitivity OPD-Scan scale, as well as normal anterior and back elevation difference map at Belin/Ambrósio enhanced ectasia display (BAD) at the Pentacam. Control group included 172 normal eyes that underwent Lasik surgery and presented no complications throughout the 2-year follow-up period. OPD-Scan and Pentacam parameters were compared, calculating sensitivity, specificity, and area under the receiver operating characteristic curve (AUC). A multivariate analysis was performed using Pentacam or OPD-Scan variables, and a model using variables of both devices. Pentacam variables with AUC ≥0.8 were keratoconus index (0.85), index of height decentration (0.81), and overall deviation at BAD (0.8). OPD-Scan variables with AUC ≥0.8 were keratoconus prediction index (0.83), surface asymmetry index (0.83), and total of higher-order trefoil aberration (0.8). In the multivariate analysis, the AUC was 0.85 in the case of OPD-Scan whereas it was 0.89 in the case of Pentacam. When combining all variables from the 2 devices, the AUC was 0.93, with a sensitivity of 82% and a specificity of 94%. Several parameters of OPD-Scan and Pentacam can be useful to differentiate cases from normal control eyes, demonstrating even better results when combining parameters of both devices. Anterior corneal indexes were the most important parameters to discriminate both groups.

Three-Dimensional Morphogeometric and Volumetric Characterization of Cornea in Pediatric Patients With Early Keratoconus

To present morphogeometric and volumetric characteristics of the cornea and its diagnostic value in pediatric patients with keratoconus (KC) using 3-dimensional (3-D) corneal modeling. Cross-sectional study. This single-center (VISSUM Innovation, Alicante, Spain) study comprised 49 eyes of 49 pediatric patients (age ≤16 years) with KC and 31 eyes of 31 healthy pediatric controls. Eyes were graded as early (n= 21) and mild KC (n= 28) based on the RETICS (Thematic Network for Co-Operative Research in Health) classification system. The 3-D corneal model was generated using raw topographic data. Deviation of anterior (Dapexant) and posterior (Dapexpost) apex and minimum thickness points (Dmctant, Dmctpost), Dapexant-Dapexpost difference, total corneal volume (Vtotal), volumetric distribution (VOLAAP, VOLPAP, and VOLMCT), and percentage of relative volume increase (VOLAAPrel, VOLPAPrel, and VOLMCTrel) between 2 consecutive radii centered to anterior/posterior apex and thinnest point were evaluated. Dapexpost and Dapexant-Dapexpost difference were higher in the early and mild KC groups compared to the control group (P < .05). Eyes with early and mild KC had decreased Vtotal compared with the control group (P < .05). Dapexpost, Dapexant-Dapexpost difference, and VOLMCTrel between 1.0 and 1.4mm diameters had area under receiver operating characteristics curve (AUROC) values over 0.93 in discrimination of early KC from normal. This is the first study presenting morphogeometric and volumetric characterization of cornea in pediatric patients with early and mild KC using a 3-D corneal model. Integration of the morphogeometric and volumetric parameters to topography software can add value in early detection of KC in pediatric patients.

Artificial intelligence in cornea, refractive, and cataract surgery

The subject of artificial intelligence has recently been responsible for the advancement of many industries including aspects of medicine and many of its subspecialties. Within ophthalmology, artificial intelligence technology has found ways of improving the diagnostic and therapeutic processes in cornea, glaucoma, retina, and cataract surgery. As demands on the modern ophthalmologist grow, artificial intelligence can be utilized to help address increased demands of modern medicine and ophthalmology by adding to the physician's clinical and surgical acumen. The purpose of this review is to highlight the integration of artificial intelligence into ophthalmology in recent years in the areas of cornea, refractive, and cataract surgery. Within the realms of cornea, refractive, and cataract surgery, artificial intelligence has played a major role in identifying ways of improving diagnostic detection. In keratoconus, artificial intelligence algorithms may help with the early detection of keratoconus and other ectatic disorders. In cataract surgery, artificial intelligence may help improve the performance of intraocular lens (IOL) calculation formulas. Further, with its potential integration into automated refraction devices, artificial intelligence can help provide an improved framework for IOL formula optimization that is more accurate and customized to a specific cataract surgeon. The future of artificial intelligence in ophthalmology is a promising prospect. With continued advancement of mathematical and computational algorithms, corneal disease processes can be diagnosed sooner and IOL calculations can be made more accurate.

Topometric and Tomographic Evaluation of Subclinical Keratoconus

ABSTRACT Purpose To investigate the corneal topometric and tomographic findings that can be used in the diagnosis of subclinical keratoconus. Methods A retrospective cohort study. The study group was selected from patients with clinically evident keratoconus in one eye and subclinical keratoconus without evident topographic findings in fellow eye. The age-matched control group was selected from patients who were candidates for laser in situ keratomileusis (LASIK) and did not develop ectasia after LASIK surgery at least 1-year follow-up. All subjects underwent topographic, topometric and tomographic (Belin-Ambrósio Enhanced Ectasia Display III) analyses via a Pentacam HR rotating Scheimpflug camera (Oculus, Germany, version 1.20r.98) before LASIK surgery. Results The study group consisted of 151 patients (69 male and 82 female, mean age of 24.8 ± 7.2 years) and the control group also consisted of 150 patients (70 male and 80 female, mean age of 26.0 ± 6.3 years). There were statistically significant differences in all measured topometric (p˂.05) and tomographic (p˂.001) parameters between the eyes with subclinical keratoconus and those of the control group. In discriminating eyes with subclinical keratoconus from normal eyes, final D showed the highest area under curve value (0.858, sensitivity 85.2%, specificity 66.7%), followed by maximum pachymetric progression index (0.809, sensitivity 81.9%, specificity 69.4%) and average pachymetric progression index (0.796, sensitivity 81.9%, specificity 68.1%) in receiver operating characteristic analysis. Conclusion Topometric and tomographic parameters might be useful for early detection of keratoconus, but the sensitivity and specificity of any parameter are not high enough to be used alone.

Simplifying and understanding various topographic indices for keratoconus using Scheimpflug based topographers.

Keratoconus (KC) is a progressive ectatic corneal disorder. There are multiple topographic devices and their varied indices used for diagnosis, detecting progression, and deciding management. It is important to understand the repeatablility, intra- test variabililty, and comparability amongst various topographic devices. The Scheimpflug camera-based devices, such as the Pentacam (Oculus, Wetzlar, Germany), Galilei (Ziemer, Biel, Switzerland), and Sirius (Costruzione Strumenti Oftalmici, Florence, Italy) are known to assist in the detection of early keratoconus and subclinical keratoconus. This article reviews the various Scheimpflug camera-based devices in depth, addressing their different indices, diagnostic accuracy, repeatability, and agreement and identifying the strongest parameter of each device. It will guide the practicing clinician by giving practical tips for decision making in the diagnosis and management of keratoconus.

Computer aided diagnosis for suspect keratoconus detection

PurposeTo develop a stable and low-cost computer aided diagnosis (CAD) system for early keratoconus detection for clinical use. MethodsThe CAD combines a custom-made mathematical model, a feedforward neural network (FFN) and a Grossberg-Runge Kutta architecture to detect clinical and suspect keratoconus. It was applied to retrospective data of 851 subjects for whom corneal elevation and thickness data was available. These data were divided into four groups: a control group (312 eyes) with bilateral normal tomography, keratoconus suspect (77 eyes) with a clinically diagnosed keratoconus in one eye and a normal fellow eye, mild keratoconus (220 eyes), and moderate keratoconus (229 eyes). The proposed framework is validated using 10-cross-validation, holdout validation and ROC curves. ResultsThe CAD detects suspect keratoconus with an accuracy of 96.56% (sensitivity 97.78%, specificity 95.56%) versus an accuracy of 89.00% (sensitivity 83.00%, specificity 95.00%) for Belin/Ambrosio Deviation (BADD), and an accuracy of 79.00% (sensitivity 58.00%, specificity 99.70%) for Topographical Keratoconus Classification (TKC). For the detection of mild to moderate keratoconus CAD shows nearly similar accuracies as previously described methods, with an average accuracy of 99.50% for CAD, versus 99.46% for BADD and 96.50% for TKC. The proposed algorithm also provides a 70% reduction in computation time, while increasing stability and convergence with respect to traditional machine learning techniques. ConclusionThe proposed algorithm is highly accurate and provides a stable screening platform to assist ophthalmologists with the early detection of keratoconus. This framework could potentially be set up for any Scheimpflug tomography system.

Innate Immunity Biomarkers for Early Detection of Keratoconus

ABSTRACTPurpose: To compare toll-like receptors 2 (TLR2) and 4 (TLR4) expression in cells of corneal and conjunctival epithelium between unilateral KC patients and control subjects.Methods: Prospective cross-sectional case-control study, including 50 unilateral KC patients and 20 control subjects. TLR2 and TLR4 expression was measured with flow cytometry.Results: Mean expression of TLR2 and TLR4, in corneal and conjunctival cells, was significantly higher in KC than in subclinical and control groups (all p < 0.0001). TLR2 expression in corneal epithelial cells can predict with the highest sensitivity and specificity the probability of KC existence compared to a control (area under the curve 0.995, 95% CI: 0.987–1.000; p < 0.0001). Corneal TLR2 expression also has a predictive capacity to discriminate between subclinical KC and controls (area under the curve 0.989, 95% CI: 0.975–1.000; p < 0.0001).Conclusions: TLR2 and TLR4 expression in corneal and conjunctival epithelial cells may constitute predictive biomarkers for the early detection of KC.

Sensitivity and specificity of corneal elevation data in the early detection of keratoconus

Purpose: To use the topography and aberrometry data provided by a Scheimpflug imagining system to determine which parameters offer the best sensitivity and specificity in the detection of early stages of keratoconus, as compared to healthy eyes.

Comparison of Discriminant Analysis and Decision Trees for the Detection of Subclinical Keratoconus

Iatrogenic keratectasia is one of the most dreaded complications of refractive surgery. In most cases, keratectasia develops after refractive surgery of eyes suffering from subclinical stages of keratoconus with few or no signs. Unfortunately, there has been no reliable procedure for the early detection of keratoconus. In this study, we used binary decision trees (recursive partitioning) to assess their suitability for discrimination between normal eyes and eyes with subclinical keratoconus. The method of decision tree analysis was compared with discriminant analysis which has shown good results in previous studies. Input data were 32 eyes of 32 patients with newly diagnosed keratoconus in the contralateral eye and preoperative data of 10 eyes of 5 patients with keratectasia after laser in-situ keratomileusis (LASIK). The control group was made up of 245 normal eyes after LASIK and 12-month follow-up without any signs of iatrogenic keratectasia. Decision trees gave better accuracy and specificity than did discriminant analysis. The sensitivity of decision trees was lower than the sensitivity of discriminant analysis. On the basis of the patient population of this study, decision trees did not prove to be superior to linear discriminant analysis for the detection of subclinical keratoconus.