Cross-sectional study: Does combining optical coherence tomography measurements using the 'Random Forest' decision tree classifier improve the prediction of the presence of perimetric deterioration in glaucoma suspects?

Identifying important microbial biomarkers for the diagnosis of colon cancer using a random forest approach

Colon cancer is one of the most common cancers, with 30-50 % of patients returning or metastasizing within 5 years of treatment. Increasingly, researchers have highlighted the influence of microbes on cancer malignant activity, while no studies have explored the relationship between colon cancer and the microbes in tumors. Here, we used tissue and blood samples from 67 colon cancer patients to identify pathogenic microorganisms associated with the diagnosis and prediction of colon cancer and evaluate the predictive performance of each pathogenic marker and its combination based on the next-generation sequencing data by using random forest algorithms. The results showed that we constructed a database of 13,187 pathogenic microorganisms associated with human disease and identified 2 pathogenic microorganisms (Synthetic.construct_32630 and Dicrocoelium.dendriticum_57078) associated with colon cancer diagnosis, and the constructed diagnostic prediction model performed well for tumor tissue samples and blood samples. In summary, for the first time, we provide new molecular markers for the diagnosis of colon cancer based on the expression of pathogenic microorganisms in order to provide a reference for improving the effective screening rate of colon cancer in clinical practice and ameliorating the personalized treatment of colon cancer patients.

Integrated multi-omics and machine learning approach reveals lipid metabolic biomarkers and signaling in age-related meibomian gland dysfunction

Meibomian gland dysfunction (MGD) is a prevalent inflammatory disorder of the ocular surface that significantly impacts patients' vision and quality of life. The underlying mechanism of aging and MGD remains largely uncharacterized. The aim of this work is to investigate lipid metabolic alterations in age-related MGD (ARMGD) through integrated proteomics, lipidomics and machine learning (ML) approach. For this purpose, we collected samples of female mouse meibomian glands (MGs) dissected from eyelids at age two months (n = 9) and two years (n = 9) for proteomic and lipidomic profilings using the liquid chromatography with tandem mass spectrometry (LC-MS/MS) method. To further identify ARMGD-related lipid biomarkers, ML model was established using the least absolute shrinkage and selection operator (LASSO) algorithm. For proteomic profiling, 375 differentially expressed proteins were detected. Functional analyses indicated the leading role of cholesterol biosynthesis in the aging process of MGs. Several proteins were proposed as potential biomarkers, including lanosterol synthase (Lss), 24-dehydrocholesterol reductase (Dhcr24), and farnesyl diphosphate farnesyl transferase 1 (Fdft1). Concomitantly, lipidomic analysis unveiled 47 lipid species that were differentially expressed and clustered into four classes. The most notable age-related alterations involved a decline in cholesteryl esters (ChE) levels and an increase in triradylglycerols (TG) levels, accompanied by significant differences in their lipid unsaturation patterns. Through ML construction, it was confirmed that ChE(26:0), ChE(26:1), and ChE(30:1) represent the most promising diagnostic molecules. The present study identified essential proteins, lipids, and signaling pathways in age-related MGD (ARMGD), providing a reference landscape to facilitate novel strategies for the disease transformation.

Uncovering the role of transient receptor potential channels in pterygium: a machine learning approach

OBJECTIVES

We aimed at identifying the role of transient receptor potential (TRP) channels in pterygium.

METHODS

Based on microarray data GSE83627 and GSE2513, differentially expressed genes (DEGs) were screened and 20 hub genes were selected. After gene correlation analysis, 5 TRP-related genes were obtained and functional analyses of gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) were performed. Multifactor regulatory network including mRNA, microRNAs (miRNAs) and transcription factors (TFs) was constructed. The 5 gene TRP signature for pterygium was validated by multiple machine learning (ML) programs including support vector classifiers (SVC), random forest (RF), and k-nearest neighbors (KNN). Additionally, we outlined the immune microenvironment and analyzed the candidate drugs. Finally, in vitro experiments were performed using human conjunctival epithelial cells (CjECs) to confirm the bioinformatics results.

RESULTS

Five TRP-related genes (MCOLN1, MCOLN3, TRPM3, TRPM6, and TRPM8) were validated by ML algorithms. Functional analyses revealed the participation of lysosome and TRP-regulated inflammatory pathways. A comprehensive immune infiltration landscape and TFs-miRNAs-mRNAs network was studied, which indicated several therapeutic targets (LEF1 and hsa-miR-455-3p). Through correlation analysis, MCOLN3 was proposed as the most promising immune-related biomarker. In vitro experiments further verified the reliability of our in silico results and demonstrated that the 5 TRP-related genes could influence the proliferation and proinflammatory signaling in conjunctival tissue contributing to the pathogenesis of pterygium.

CONCLUSIONS

Our study suggested that TRP channels played an essential role in the pathogenesis of pterygium. The identified pivotal biomarkers (especially MCOLN3) and pathways provide novel directions for future mechanistic and therapeutic studies for pterygium.

The Comparison of Age-related Change in Retinal Nerve Fiber Layer and Ganglion Cell Complex Thicknesses between Glaucoma Suspects and Healthy Individuals

Aim

The purpose of this study is to investigate the difference of change in the retinal nerve fiber layer (RNFL) and the ganglion cell complex (GCC) thickness according to age in glaucoma suspect individuals and healthy subjects.

Materials and methods

Thicknesses of RNFL and GCC were measured with spectral domain optical coherence tomography (SD-OCT) in glaucoma-suspected individuals and healthy subjects. The differences in age, overall mean, four quadrants, and 12 clock-hour sectors of RNFL and overall mean, superior half, and inferior half GCC thicknesses between glaucoma suspects and healthy participants were analyzed and compared using linear regression analyses.

Results

There were 201 glaucoma-suspect individuals and 121 healthy subjects with a mean age of 38.89 and 40.26 years, respectively (p = 0.27). The mean overall RNFL thickness was found to be 97.76 and 97.43 µm in healthy individuals and glaucoma suspects (p = 0.72). The mean overall GCC thickness was found to be 111.30 and 104.67 µm in healthy individuals and glaucoma suspects, respectively (p < 0.001). There was a 0.11 µm decrease per year found in overall GCC thickness in glaucoma suspects and 0.23 µm decreases per year in overall GCC thickness in healthy individuals (p < 0.001). There was a 0.02 µm decrease per year found in overall RNFL thickness in glaucoma suspects and a 0.29 µm decrease per year in overall RNFL thickness in healthy individuals (p < 0.001). However, these per-year decreases in GCC thickness glaucoma suspects and healthy individuals were not found to be statistically significant (p = 0.21); on the other hand, this difference for RNFL thickness was significant (p < 0.001).

Conclusion

It was found that the thicknesses of RNFL and GCC were different between glaucoma suspects and healthy individuals. However, age-related decay in the RNFL and GCC thicknesses was not uniform in healthy individuals and glaucoma suspects.

Clinical significance

It was found that the RNFL thickness and GCC thickness were lower in glaucoma suspects than in healthy controls eyes. However, an age-related decrease of RNFL and GCC thicknesses were found to be less in glaucoma suspects compared with healthy controls.

How to cite this article

Firatli G, Elibol A, Altinbas E, et al. The Comparison of Age-related Change in Retinal Nerve Fiber Layer and Ganglion Cell Complex Thicknesses between Glaucoma Suspects and Healthy Individuals. J Curr Glaucoma Pract 2023;17(1):22-29.

Glaucoma classification based on scanning laser ophthalmoscopic images using a deep learning ensemble method

This study aimed to assess the utility of optic nerve head (onh) en-face images, captured with scanning laser ophthalmoscopy (slo) during standard optical coherence tomography (oct) imaging of the posterior segment, and demonstrate the potential of deep learning (dl) ensemble method that operates in a low data regime to differentiate glaucoma patients from healthy controls. The two groups of subjects were initially categorized based on a range of clinical tests including measurements of intraocular pressure, visual fields, oct derived retinal nerve fiber layer (rnfl) thickness and dilated stereoscopic examination of onh. 227 slo images of 227 subjects (105 glaucoma patients and 122 controls) were used. A new task-specific convolutional neural network architecture was developed for slo image-based classification. To benchmark the results of the proposed method, a range of classifiers were tested including five machine learning methods to classify glaucoma based on rnfl thickness—a well-known biomarker in glaucoma diagnostics, ensemble classifier based on inception v3 architecture, and classifiers based on features extracted from the image. The study shows that cross-validation dl ensemble based on slo images achieved a good discrimination performance with up to 0.962 of balanced accuracy, outperforming all of the other tested classifiers.

Prediction model of random forest for the risk of hyperuricemia in a Chinese basic health checkup test

Objectives: The present study aimed to develop a random forest (RF) based prediction model for hyperuricemia (HUA) and compare its performance with the conventional logistic regression (LR) model. Methods: This cross-sectional study recruited 91,690 participants (14,032 with HUA, 77,658 without HUA). We constructed a RF-based prediction model in the training sets and evaluated it in the validation sets. Performance of the RF model was compared with the LR model by receiver operating characteristic (ROC) curve analysis. Results: The sensitivity and specificity of the RF models were 0.702 and 0.650 in males, 0.767 and 0.721 in females. The positive predictive value (PPV) and negative predictive value (NPV) were 0.372 and 0.881 in males, 0.159 and 0.978 in females. AUC of the RF models was 0.739 (0.728–0.750) in males and 0.818 (0.799–0.837) in females. AUC of the LR models were 0.730 (0.718–0.741) for males and 0.815 (0.795–0.835) for females. The predictive power of RF was slightly higher than that of LR, but was not statistically significant in females (Delong tests, P=0.0015 for males, P=0.5415 for females). Conclusion: Compared with LR, the good performance in HUA status prediction and the tolerance of features associations or interactions showed great potential of RF in further application. A prospective cohort is necessary for HUA developing prediction. People with high risk factors should be encouraged to actively control to reduce the probability of developing HUA.

High interpretable machine learning classifier for early glaucoma diagnosis

AIM

To develop a classifier for differentiating between healthy and early stage glaucoma eyes based on peripapillary retinal nerve fiber layer (RNFL) thicknesses measured with optical coherence tomography (OCT), using machine learning algorithms with a high interpretability.

METHODS

Ninety patients with early glaucoma and 85 healthy eyes were included. Early glaucoma eyes showed a visual field (VF) defect with mean deviation >-6.00 dB and characteristic glaucomatous morphology. RNFL thickness in every quadrant, clock-hour and average thickness were used to feed machine learning algorithms. Cluster analysis was conducted to detect and exclude outliers. Tree gradient boosting algorithms were used to calculate the importance of parameters on the classifier and to check the relation between their values and its impact on the classifier. Parameters with the lowest importance were excluded and a weighted decision tree analysis was applied to obtain an interpretable classifier. Area under the ROC curve (AUC), accuracy and generalization ability of the model were estimated using cross validation techniques.

RESULTS

Average and 7 clock-hour RNFL thicknesses were the parameters with the highest importance. Correlation between parameter values and impact on classification displayed a stepped pattern for average thickness. Decision tree model revealed that average thickness lower than 82 µm was a high predictor for early glaucoma. Model scores had AUC of 0.953 (95%CI: 0.903-0998), with an accuracy of 89%.

CONCLUSION

Gradient boosting methods provide accurate and highly interpretable classifiers to discriminate between early glaucoma and healthy eyes. Average and 7-hour RNFL thicknesses have the best discriminant power.

Predicting intraocular pressure using systemic variables or fundus photography with deep learning in a health examination cohort

The purpose of the current study was to predict intraocular pressure (IOP) using color fundus photography with a deep learning (DL) model, or, systemic variables with a multivariate linear regression model (MLM), along with least absolute shrinkage and selection operator regression (LASSO), support vector machine (SVM), and Random Forest: (RF). Training dataset included 3883 examinations from 3883 eyes of 1945 subjects and testing dataset 289 examinations from 289 eyes from 146 subjects. With the training dataset, MLM was constructed to predict IOP using 35 systemic variables and 25 blood measurements. A DL model was developed to predict IOP from color fundus photographs. The prediction accuracy of each model was evaluated through the absolute error and the marginal R-squared (mR2), using the testing dataset. The mean absolute error with MLM was 2.29 mmHg, which was significantly smaller than that with DL (2.70 dB). The mR2 with MLM was 0.15, whereas that with DL was 0.0066. The mean absolute error (between 2.24 and 2.30 mmHg) and mR2 (between 0.11 and 0.15) with LASSO, SVM and RF were similar to or poorer than MLM. A DL model to predict IOP using color fundus photography proved far less accurate than MLM using systemic variables.

Artificial Intelligence Algorithms to Diagnose Glaucoma and Detect Glaucoma Progression: Translation to Clinical Practice

Purpose

This concise review aims to explore the potential for the clinical implementation of artificial intelligence (AI) strategies for detecting glaucoma and monitoring glaucoma progression.

Methods

Nonsystematic literature review using the search combinations “Artificial Intelligence,” “Deep Learning,” “Machine Learning,” “Neural Networks,” “Bayesian Networks,” “Glaucoma Diagnosis,” and “Glaucoma Progression.” Information on sensitivity and specificity regarding glaucoma diagnosis and progression analysis as well as methodological details were extracted.

Results

Numerous AI strategies provide promising levels of specificity and sensitivity for structural (e.g. optical coherence tomography [OCT] imaging, fundus photography) and functional (visual field [VF] testing) test modalities used for the detection of glaucoma. Area under receiver operating curve (AROC) values of > 0.90 were achieved with every modality. Combining structural and functional inputs has been shown to even more improve the diagnostic ability. Regarding glaucoma progression, AI strategies can detect progression earlier than conventional methods or potentially from one single VF test.

Conclusions

AI algorithms applied to fundus photographs for screening purposes may provide good results using a simple and widely accessible test. However, for patients who are likely to have glaucoma more sophisticated methods should be used including data from OCT and perimetry. Outputs may serve as an adjunct to assist clinical decision making, whereas also enhancing the efficiency, productivity, and quality of the delivery of glaucoma care. Patients with diagnosed glaucoma may benefit from future algorithms to evaluate their risk of progression. Challenges are yet to be overcome, including the external validity of AI strategies, a move from a “black box” toward “explainable AI,” and likely regulatory hurdles. However, it is clear that AI can enhance the role of specialist clinicians and will inevitably shape the future of the delivery of glaucoma care to the next generation.

Translational Relevance

The promising levels of diagnostic accuracy reported by AI strategies across the modalities used in clinical practice for glaucoma detection can pave the way for the development of reliable models appropriate for their translation into clinical practice. Future incorporation of AI into healthcare models may help address the current limitations of access and timely management of patients with glaucoma across the world.

Applying Machine Learning Techniques in Nomogram Prediction and Analysis for SMILE Treatment

PURPOSE

To analyze the outcome of machine learning technique for prediction of small incision lenticule extraction (SMILE) nomogram.

DESIGN

Prospective, comparative clinical study.

METHODS

A comparative study was conducted on the outcomes of SMILE surgery between surgeon group (nomogram set by surgeon) and machine learning group (nomogram predicted by machine learning model). The machine learning model was trained by 865 ideal cases (spherical equivalent [SE] within ±0.5 diopter [D] 3 months postoperatively) from an experienced surgeon. The visual outcomes of both groups were compared for safety, efficacy, predictability, and SE correction.

RESULTS

There was no statistically significant difference between the baseline data in both groups. The efficacy index in the machine learning group (1.48 ± 1.08) was significantly higher than in the surgeon group (1.3 ± 0.27) (t = -2.17, P < .05). Eighty-three percent of eyes in the surgeon group and 93% of eyes in the machine learning group were within ±0.50 D, while 98% of eyes in the surgeon group and 96% of eyes in the machine learning group were within ±1.00 D. The error of SE correction was -0.09 ± 0.024 and -0.23 ± 0.021 for machine learning and surgeon groups, respectively.

CONCLUSIONS

The machine learning technique performed as well as surgeon in safety, but significantly better than surgeon in efficacy. As for predictability, the machine learning technique was comparable to surgeon, although less predictable for high myopia and astigmatism.

Early Glaucoma Discrimination Index

PURPOSE

To develop a new structural algorithm derived from optical coherence tomography (OCT) retinal nerve fiber layer (RNFL) thickness and asymmetry and validate it as a discriminate among normal, suspect, and early primary open-angle glaucoma (POAG).

STUDY DESIGN

A case-controlled observational clinical study.

MATERIALS AND METHODS

In total, 150 subjects (299 eyes) were selected, 61 normal, 46 suspect, and 43 early glaucoma, from Al-Azhar University Hospitals. They were in fifth decade and free from any ocular or systemic diseases affecting the retinal nerve fiber layer. They were investigated by two consecutive perimetry (1 month apart), and three scans of circumpapillary retinal nerve fiber layer (cpRNFL) by using Nidek spectral domain (SD)-OCT 3000 Lite. The cpRNFL thickness (cpRNFLT) and inter-eye asymmetry parameters were analyzed among the three groups. Then some selected parameters were selected and analyzed using a binary logistic regression analysis for developing the new algorithm. The new algorithm was tested for the best fitting, accuracy, and diagnostic ability among the three groups and was validated in the suspect group.

RESULTS

The new algorithm model [early glaucoma discrimination index (EGDI)] works well with only four variables; whole cpRNFLT, inferior quadrant cpRNFLT, inferotemporal clock hour (CH) cpRNFLT, and absolute inter-eye inferior quadrants asymmetry. The highest area under the curve (AUC) obtained from the EGDI among the three groups was 0.854. The validation analysis in the suspect group revealed a higher diagnostic ability in discrimination of early glaucoma with AUC of 0.989 (0.976-1.003).

CONCLUSION

The EGDI showed better diagnostic ability for diagnosis of glaucoma in the pre-perimetric stage. The new OCT algorithm is simple and can be run in any SD-OCT device without dependence on normative data.

HOW TO CITE THIS ARTICLE

Safwat H, Nassar E, Rashwan A. Early Glaucoma Discrimination Index. J Curr Glaucoma Pract 2020;14(1):16-24.

Validation of the UNC OCT Index for the Diagnosis of Early Glaucoma

Purpose

To independently validate the performance of the University of North Carolina Optical Coherence Tomography (UNC OCT) Index in diagnosing and predicting early glaucoma.

Methods

Data of 118 normal subjects (118 eyes) and 96 subjects (96 eyes) with early glaucoma defined as visual field mean deviation (MD) greater than -4 decibels (dB), aged 40 to 80 years, and who were enrolled in the Full-Threshold Testing Size III, V, VI comparison study were used in this study. CIRRUS OCT average and quadrants' retinal nerve fiber layer (RNFL); optic disc vertical cup-to-disc ratio (VCDR), cup-to-disc area ratio, and rim area; and average, minimum, and six sectoral ganglion cell-inner plexiform layer (GCIPL) measurements were run through the UNC OCT Index algorithm. Area under the receiver operating characteristic curve (AUC) and sensitivities at 95% and 99% specificity were calculated and compared between single parameters and the UNC OCT Index.

Results

Mean age was 60.1 ± 11.0 years for normal subjects and 66.5 ± 8.1 years for glaucoma patients (P < 0.001). MD was 0.29 ± 1.04 dB and -1.30 ± 1.35 dB in normal and glaucomatous eyes (P < 0.001), respectively. The AUC of the UNC OCT Index was 0.96. The best single metrics when compared to the UNC OCT Index were VCDR (0.93, P = 0.054), average RNFL (0.92, P = 0.014), and minimum GCIPL (0.91, P = 0.009). The sensitivities at 95% and 99% specificity were 85.4% and 76.0% (UNC OCT Index), 71.9% and 62.5% (VCDR, all P < 0.001), 64.6% and 53.1% (average RNFL, all P < 0.001), and 66.7% and 58.3% (minimum GCIPL, all P < 0.001), respectively.

Conclusions

The findings confirm that the UNC OCT Index may provide improved diagnostic perforce over that of single OCT parameters and may be a good tool for detection of early glaucoma.

Translational Relevance

The UNC OCT Index algorithm may be incorporated easily into routine clinical practice and be useful for detecting early glaucoma.

The Relationship between the Waveform Parameters from the Ocular Response Analyzer and the Progression of Glaucoma

PURPOSE

To investigate the usefulness of waveform parameters measured with the Ocular Response Analyzer (Reichert Ophthalmic Instruments, Depew, NY) in assessing the progression of glaucomatous visual field (VF).

DESIGN

Observational cross-sectional study.

PARTICIPANTS

One hundred and one eyes with primary open-angle glaucoma in 68 patients with 8 reliable VFs using the Humphrey Field Analyzer (Carl Zeiss Meditec, Inc., Dublin, CA).

METHODS

The mean of total deviation (mTD) value of the 52 test points in the 24-2 Humphrey Field Analyzer VF test pattern was calculated, and the progression rate of mTD was determined using 8 VFs. Ocular Response Analyzer measurement was performed 3 times in the same day, and the average values of the 3 measurements were used in the analysis. Then, the optimal linear mixed model was selected using 7 parameters: age, mean and standard deviation of intraocular pressure with the Goldmann applanation tonometry during the observation period, central corneal thickness, axial length, mTD in the initial VF, and corneal hysteresis (CH) other than waveform parameters, henceforth known as the basic model. In addition, using the 37 waveform parameters, the optimal model for the mTD progression rate was identified, according to the second-order bias-corrected Akaike information criterion (AICc) index, using 15 preselected waveform parameters with the least absolute shrinkage and selection operator regression (henceforth known as the waveform model).

MAIN OUTCOME MEASURES

Optimal linear mixed models for the mTD progression rate, as determined by AICc index.

RESULTS

The mean ± standard deviation mTD progression rate was -0.25±0.31 dB/year. The basic model was mTD progression rate = -0.94 + 0.075 × CH (AICc = 46.71). The waveform model was mTD progression rate = 1.25 - 0.066 × path2 - 0.000099 × p2area + 0.0021 × mslew2 (AICc = 44.95). The relative likelihood of the latter model being the optimal model was 6.23 times greater than that of the former model.

CONCLUSIONS

Ocular Response Analyzer waveform parameters were correlated significantly with glaucomatous VF progression and showed a stronger than correlation with VF progression than CH.

Validating the Usefulness of the "Random Forests" Classifier to Diagnose Early Glaucoma With Optical Coherence Tomography

PURPOSE

To validate the usefulness of the "Random Forests" classifier to diagnose early glaucoma with spectral-domain optical coherence tomography (SDOCT).

METHODS

design: Comparison of diagnostic algorithms.

SETTING

Multiple institutional practices.

STUDY PARTICIPANTS

Training dataset included 94 eyes of 94 open-angle glaucoma (OAG) patients and 84 eyes of 84 normal subjects and testing dataset included 114 eyes of 114 OAG patients and 82 eyes of 82 normal subjects. In both groups, OAG eyes with mean deviation (MD) values better than -5.0 dB were included.

OBSERVATION PROCEDURE

Using the training dataset, classifiers were built to discriminate between glaucoma and normal eyes using 84 OCT measurements using the Random Forests method, multiple logistic regression models based on backward or bidirectional stepwise model selection, a least absolute shrinkage and selection operator regression (LASSO) model, and a Ridge regression model.

MAIN OUTCOME MEASURES

Diagnostic accuracy.

RESULTS

With the testing data, the area under the receiver operating characteristic curve (AROC) with the Random Forests method (93.0%) was significantly (P < .05) larger than those with other models of the stepwise model selections (71.9%), LASSO model (89.6%), and Ridge model (89.2%).

CONCLUSION

It is useful to analyze multiple SDOCT parameters concurrently using the Random Forests method to diagnose glaucoma in early stages.

Machine Learning Techniques in Clinical Vision Sciences

This review presents and discusses the contribution of machine learning techniques for diagnosis and disease monitoring in the context of clinical vision science. Many ocular diseases leading to blindness can be halted or delayed when detected and treated at its earliest stages. With the recent developments in diagnostic devices, imaging and genomics, new sources of data for early disease detection and patients' management are now available. Machine learning techniques emerged in the biomedical sciences as clinical decision-support techniques to improve sensitivity and specificity of disease detection and monitoring, increasing objectively the clinical decision-making process. This manuscript presents a review in multimodal ocular disease diagnosis and monitoring based on machine learning approaches. In the first section, the technical issues related to the different machine learning approaches will be present. Machine learning techniques are used to automatically recognize complex patterns in a given dataset. These techniques allows creating homogeneous groups (unsupervised learning), or creating a classifier predicting group membership of new cases (supervised learning), when a group label is available for each case. To ensure a good performance of the machine learning techniques in a given dataset, all possible sources of bias should be removed or minimized. For that, the representativeness of the input dataset for the true population should be confirmed, the noise should be removed, the missing data should be treated and the data dimensionally (i.e., the number of parameters/features and the number of cases in the dataset) should be adjusted. The application of machine learning techniques in ocular disease diagnosis and monitoring will be presented and discussed in the second section of this manuscript. To show the clinical benefits of machine learning in clinical vision sciences, several examples will be presented in glaucoma, age-related macular degeneration, and diabetic retinopathy, these ocular pathologies being the major causes of irreversible visual impairment.

Discriminating between glaucoma and normal eyes using optical coherence tomography and the 'Random Forests' classifier

PURPOSE

To diagnose glaucoma based on spectral domain optical coherence tomography (SD-OCT) measurements using the 'Random Forests' method.

METHODS

SD-OCT was conducted in 126 eyes of 126 open angle glaucoma (OAG) patients and 84 eyes of 84 normal subjects. The Random Forests method was then applied to discriminate between glaucoma and normal eyes using 151 OCT parameters including thickness measurements of circumpapillary retinal nerve fiber layer (cpRNFL), the macular RNFL (mRNFL) and the ganglion cell layer-inner plexiform layer combined (GCIPL). The area under the receiver operating characteristic curve (AROC) was calculated using the Random Forests method adopting leave-one-out cross validation. For comparison, AROCs were calculated based on each one of the 151 OCT parameters.

RESULTS

The AROC obtained with the Random Forests method was 98.5% [95% Confidence interval (CI): 97.1-99.9%], which was significantly larger than the AROCs derived from any single OCT parameter (maxima were: 92.8 [CI: 89.4-96.2] %, 94.3 [CI: 91.1-97.6] % and 91.8 [CI: 88.2-95.4] % for cpRNFL-, mRNFL- and GCIPL-related parameters, respectively; P<0.05, DeLong's method with Holm's correction for multiple comparisons). The partial AROC above specificity of 80%, for the Random Forests method was equal to 18.5 [CI: 16.8-19.6] %, which was also significantly larger than the AROCs of any single OCT parameter (P<0.05, Bootstrap method with Holm's correction for multiple comparisons).

CONCLUSIONS

The Random Forests method, analyzing multiple SD-OCT parameters concurrently, significantly improves the diagnosis of glaucoma compared with using any single SD-OCT measurement.