Dataset Information

Development of a classification system based on corneal biomechanical properties using artificial intelligence predicting keratoconus severity.

ABSTRACT:

Background

To investigate machine-learning (ML) algorithms to differentiate corneal biomechanical properties between different topographical stages of keratoconus (KC) by dynamic Scheimpflug tonometry (CST, Corvis ST, Oculus, Wetzlar, Germany). In the following, ML models were used to predict the severity in a training and validation dataset.

Methods

Three hundred and eighteen keratoconic and one hundred sixteen healthy eyes were included in this monocentric and cross-sectional pilot study. Dynamic corneal response (DCR) and corneal thickness related (pachymetric) parameters from CST were chosen by appropriated selection techniques to develop a ML algorithm. The stage of KC was determined by the topographical keratoconus classification system (TKC, Pentacam, Oculus). Patients who were classified as TKC 1, TKC 2 and TKC 3 were assigned to subgroup mild, moderate, and advanced KC. If patients were classified as TKC 1-2, TKC 2-3 or TKC 3-4, they were assigned to subgroups according to the normative range of further corneal indices (index of surface variance, keratoconus index and minimum radius). Patients classified as TKC 4 were not included in this study due to the limited amount of cases. Linear discriminant analysis (LDA) and random forest (RF) algorithms were used to develop the classification models. Data were divided into training (70% of cases) and validation (30% of cases) datasets.

Results

LDA model predicted healthy, mild, moderate, and advanced KC eyes with a sensitivity (S_n)/specificity (S_p) of 82%/97%, 73%/81%, 62%/83% and 68%/95% from a validation dataset, respectively. For the RF model, a S_n/S_p of 91%/94%, 80%/90%, 63%/87%, 72%/95% could be reached for predicting healthy, mild, moderate, and advanced KC eyes, respectively. The overall accuracy of LDA and RF was 71% and 78%, respectively. The accuracy for KC detection including all subgroups of KC severity was 93% in both models.

Conclusion

The RF model showed good accuracy in predicting healthy eyes and various stages of KC. The accuracy was superior with respect to the LDA model. The clinical importance of the models is that the standalone dynamic Scheimpflug tonometry is able to predict the severity of KC without having the keratometric data.

Trial registration

NCT04251143 at Clinicaltrials.gov, registered at 12 March 2018 (Retrospectively registered).

SUBMITTER: Herber R

PROVIDER: S-EPMC8167942 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundCorneal topography is a clinically validated examination method for keratoconus. However, there is no clear guideline regarding patient selection for corneal topography. We developed and validated a novel artificial intelligence (AI) model to identify patients who would benefit from corneal topography based on basic ophthalmologic examinations, including a survey of visual impairment, best-corrected visual acuity (BCVA) measurement, intraocular pressure (IOP) measurement, and autokeratometry.MethodsA total of five AI models (three individual models with fully connected neural network including the XGBoost, and the TabNet models, and two ensemble models with hard and soft voting methods) were trained and validated. We used three datasets collected from the records of 2,613 patients' basic ophthalmologic examinations from two institutions to train and validate the AI models. We trained the AI models using a dataset from a third medical institution to determine whether corneal topography was needed to detect keratoconus. Finally, prospective intra-validation dataset (internal test dataset) and extra-validation dataset from a different medical institution (external test dataset) were used to assess the performance of the AI models.ResultsThe ensemble model with soft voting method outperformed all other AI models in sensitivity when predicting which patients needed corneal topography (90.5% in internal test dataset and 96.4% in external test dataset). In the error analysis, most of the predicting error occurred within the range of the subclinical keratoconus and the suspicious D-score in the Belin-Ambrósio enhanced ectasia display. In the feature importance analysis, out of 18 features, IOP was the highest ranked feature when comparing the average value of the relative attributions of three individual AI models, followed by the difference in the value of mean corneal power.ConclusionAn AI model using the results of basic ophthalmologic examination has the potential to recommend corneal topography for keratoconus. In this AI algorithm, IOP and the difference between the two eyes, which may be undervalued clinical information, were important factors in the success of the AI model, and may be worth further reviewing in research and clinical practice for keratoconus screening.

Project description:ImportanceKeratoconus is an important cause of visual loss in young adults, but little is known about its genetic causes. Understanding the genetic determinants of corneal biomechanical factors may in turn teach us about keratoconus etiology.ObjectivesTo identify genetic associations with corneal biomechanical properties and to examine whether these genetic variants are associated with keratoconus.Design, setting, and participantsA stage 1 discovery and replication genome-wide association study (GWAS) of corneal biomechanical properties was performed in 2 cross-sectional populations (6645 participants from the European Prospective Investigation into Cancer and Nutrition [EPIC]-Norfolk Eye Study and 2384 participants from the TwinsUK study). In stage 2, the association of genetic determinants identified in stage 1 with keratoconus was examined in a case-control study. A total of 752 patients with keratoconus were compared with 974 TwinsUK participants (undergoing direct sequencing) or 13 828 EPIC-Norfolk participants (undergoing genotyping and imputation) who were not part of the stage 1 analysis. Data were collected from March 1, 1993, through March 13, 2017, and analyzed from November 1, 2015, through February 1, 2018.ExposuresIn stage 1, allele dosage at genome-wide single-nucleotide polymorphisms (SNPs); in stage 2, allele dosage at SNPs with genome-wide significance (P < 5 × 10-8) in stage 1 and not previously reported as associated with corneal disease.Main outcomes and measuresIn stage 1, corneal hysteresis (CH) and corneal resistance factor (CRF), measured with the Ocular Response Analyzer (ORA); in stage 2, association with keratoconus compared with controls.ResultsAmong 6645 participants in the discovery cohort (3635 women (54.7%); mean age, 69 years [range, 48-92 years]), 7 genome-wide significant loci associated with CH or CRF were identified that were independently replicated. Two further suggestive loci were identified after meta-analysis. To date, 5 of the identified loci, at ANAPC1, ADAMTS8, ADAMTS17, ABCA6, and COL6A1, have not previously been reported as associated with corneal disease. The ABCA6 locus (rs77542162) was associated with keratoconus using the TwinsUK (odds ratio [OR], 0.50; 95% CI, 0.27-0.92; P = .03) and EPIC-Norfolk controls (OR, 0.39; 95% CI, 0.22-0.70; P = .002). The other loci were associated with keratoconus using TwinsUK (OR per effect allele for ADAMTS8, 0.51 [95% CI, 0.37-0.71; P = 7.9 × 10-5]; for COL6A1, 1.65 [95% CI, 1.05-2.59; P = .03]) or EPIC-Norfolk (OR per effect allele for ANAPC1, 0.78 [95% CI, 0.68-0.89; P = 3.7 × 10-4]; for ADAMTS17, 0.82 [95% CI, 0.68-0.99; P = .04]) controls.Conclusions and relevanceFive loci that are associated with corneal biomechanical properties and that have suggestive associations with keratoconus were reported. These findings suggest the role of type VI collagen, extracellular matrix, and connective-tissue development for corneal biomechanics and keratoconus and the role of CH and CRF as biomarkers for keratoconus.

Project description:ObjectiveThe study aimed to evaluate the character of corneal stiffness parameter at the first applanation (SP-A1) in normal and keratoconus eyes and explore the association between SP-A1 and keratoconus severity indicators.MethodsA total of 351 normal and 351 keratoconus eyes were included in the current study. Keratoconus was diagnosed according to the corneal topography map and slit-lamp examination. The severity of keratoconus was classified to mild (steep keratometry (Ks) < 48D), moderate (48 ≤ Ks < 55D), and severe (Ks ≥ 55D). The SP-A1 was measured using the Corvis ST software. The correlation analyses and receiver operating characteristic (ROC) curve were performed in the current analysis.ResultsThe SP-A1 values of keratoconus were lower than that of normal eyes (72.11 (57.02, 83.08) mmHg/mm vs 110.89 (100.45, 122.47) mmHg/mm, P < 0.001). With the severity of keratoconus increasing, the SP-A1 decreased and the value of SP-A1 was 79.54 (70.30, 90.93) mmHg/mm, 65.11 (53.14, 77.46) mmHg/mm, and 47.59 (37.50, 62.14) mmHg/mm in mild, moderate, and severe keratoconus eyes, respectively (P < 0.001). The negative association between SP-A1 and Ks was found in mild, moderate, and severe keratoconus eyes (r mild = -0.171, r moderate = -0.317, r severe = -0.288, all P < 0.05). A positive association between SP-A1 and the thinnest corneal thickness (TCT) was found in all eyes (rnormal = 0.687, r mild = 0.519, r moderate = 0.488, r severe = 0.382, all P < 0.05). SP-A1 was found to be statistically positively associated with intraocular pressure (IOP), biomechanical corrected IOP (bIOP), time from the initiation of air puff until the first applanation (A1T), corneal velocity at the second applanation (A2V), and negatively associated with deformation amplitude (DA), peak distance (PD), corneal velocity at the first applanation (A1V), time from the initiation of air puff until the second applanation (A2T), and DA Ratio Max [2 mm] both in normal and keratoconus eyes (all P < 0.05). The ROC analysis indicated that the AUC (95% CI) of SP-A1 was 0.952 (0.934-0.967) and 0.930 (0.904-0.951) in detecting keratoconus eyes and mild keratoconus eyes from normal eyes, respectively.ConclusionsThe SP-A1 value decreased while the keratoconus severity increased. It was lower in keratoconus than that in normal eyes and could be helpful in identifying keratoconus eyes from normal eyes. Further researches would be warranted to expand the clinical utility of SP-A1.

Project description:ImportanceColorectal polyps are common, and their histopathologic classification is used in the planning of follow-up surveillance. Substantial variation has been observed in pathologists' classification of colorectal polyps, and improved assessment by pathologists may be associated with reduced subsequent underuse and overuse of colonoscopy.ObjectiveTo compare standard microscopic assessment with an artificial intelligence (AI)-augmented digital system that annotates regions of interest within digitized polyp tissue and predicts polyp type using a deep learning model to assist pathologists in colorectal polyp classification.Design, setting, and participantsIn this diagnostic study conducted at a tertiary academic medical center and a community hospital in New Hampshire, 100 slides with colorectal polyp samples were read by 15 pathologists using a microscope and an AI-augmented digital system, with a washout period of at least 12 weeks between use of each modality. The study was conducted from February 10 to July 10, 2020.Main outcomes and measuresAccuracy and time of evaluation were used to compare pathologists' performance when a microscope was used with their performance when the AI-augmented digital system was used. Outcomes were compared using paired t tests and mixed-effects models.ResultsIn assessments of 100 slides with colorectal polyp specimens, use of the AI-augmented digital system significantly improved pathologists' classification accuracy compared with microscopic assessment from 73.9% (95% CI, 71.7%-76.2%) to 80.8% (95% CI, 78.8%-82.8%) (P < .001). The overall difference in the evaluation time per slide between the digital system (mean, 21.7 seconds; 95% CI, 20.8-22.7 seconds) and microscopic examination (mean, 13.0 seconds; 95% CI, 12.4-13.5 seconds) was -8.8 seconds (95% CI, -9.8 to -7.7 seconds), but this difference decreased as pathologists became more familiar and experienced with the digital system; the difference between the time of evaluation on the last set of 20 slides for all pathologists when using the microscope and the digital system was 4.8 seconds (95% CI, 3.0-6.5 seconds).Conclusions and relevanceIn this diagnostic study, an AI-augmented digital system significantly improved the accuracy of pathologic interpretation of colorectal polyps compared with microscopic assessment. If applied broadly to clinical practice, this tool may be associated with decreases in subsequent overuse and underuse of colonoscopy and thus with improved patient outcomes and reduced health care costs.