Uncovering Language Disparity of ChatGPT on Retinal Vascular Disease Classification: Cross-Sectional Study.

BACKGROUND: Benefiting from rich knowledge and the exceptional ability to understand text, large language models like ChatGPT have shown great potential in English clinical environments. However, the performance of ChatGPT in non-English clinical settings, as well as its reasoning, have not been explored in depth. OBJECTIVE: This study aimed to evaluate ChatGPT's diagnostic performance and inference abilities for retinal vascular diseases in a non-English clinical environment. METHODS: In this cross-sectional study, we collected 1226 fundus fluorescein angiography reports and corresponding diagnoses written in Chinese and tested ChatGPT with 4 prompting strategies (direct diagnosis or diagnosis with a step-by-step reasoning process and in Chinese or English). RESULTS: Compared with ChatGPT using Chinese prompts for direct diagnosis that achieved an F1-score of 70.47%, ChatGPT using English prompts for direct diagnosis achieved the best diagnostic performance (80.05%), which was inferior to ophthalmologists (89.35%) but close to ophthalmologist interns (82.69%). As for its inference abilities, although ChatGPT can derive a reasoning process with a low error rate (0.4 per report) for both Chinese and English prompts, ophthalmologists identified that the latter brought more reasoning steps with less incompleteness (44.31%), misinformation (1.96%), and hallucinations (0.59%) (all P<.001). Also, analysis of the robustness of ChatGPT with different language prompts indicated significant differences in the recall (P=.03) and F1-score (P=.04) between Chinese and English prompts. In short, when prompted in English, ChatGPT exhibited enhanced diagnostic and inference capabilities for retinal vascular disease classification based on Chinese fundus fluorescein angiography reports. CONCLUSIONS: ChatGPT can serve as a helpful medical assistant to provide diagnosis in non-English clinical environments, but there are still performance gaps, language disparities, and errors compared to professionals, which demonstrate the potential limitations and the need to continually explore more robust large language models in ophthalmology practice.

ANALYSIS OF TRANSFER LEARNING FOR SELECT RETINAL DISEASE CLASSIFICATION.

PURPOSE: To analyze the effect of transfer learning for classification of diabetic retinopathy (DR) by fundus photography and select retinal diseases by spectral domain optical coherence tomography (SD-OCT). METHODS: Five widely used open-source deep neural networks and four customized simpler and smaller networks, termed the CBR family, were trained and evaluated on two tasks: 1) classification of DR using fundus photography and 2) classification of drusen, choroidal neovascularization, and diabetic macular edema using SD-OCT. For DR classification, the quadratic weighted Kappa coefficient was used to measure the level of agreement between each network and ground truth-labeled test cases. For SD-OCT-based classification, accuracy was calculated for each network. Kappa and accuracy were compared between iterations with and without use of transfer learning for each network to assess for its effect. RESULTS: For DR classification, Kappa increased with transfer learning for all networks (range of increase 0.152-0.556). For SD-OCT-based classification, accuracy increased for four of five open-source deep neural networks (range of increase 1.8%-3.5%), slightly decreased for the remaining deep neural network (-0.6%), decreased slightly for three of four CBR networks (range of decrease 0.9%-1.8%), and decreased by 9.6% for the remaining CBR network. CONCLUSION: Transfer learning improved performance, as measured by Kappa, for DR classification for all networks, although the effect ranged from small to substantial. Transfer learning had minimal effect on accuracy for SD-OCT-based classification for eight of the nine networks analyzed. These results imply that transfer learning may substantially increase performance for DR classification but may have minimal effect for SD-OCT-based classification.

MHANet: A hybrid attention mechanism for retinal diseases classification.

With the increase of patients with retinopathy, retinopathy recognition has become a research hotspot. In this article, we describe the etiology and symptoms of three kinds of retinal diseases, including drusen(DRUSEN), choroidal neovascularization(CNV) and diabetic macular edema(DME). In addition, we also propose a hybrid attention mechanism to classify and recognize different types of retinopathy images. In particular, the hybrid attention mechanism proposed in this paper includes parallel spatial attention mechanism and channel attention mechanism. It can extract the key features in the channel dimension and spatial dimension of retinopathy images, and reduce the negative impact of background information on classification results. The experimental results show that the hybrid attention mechanism proposed in this paper can better assist the network to focus on extracting thr fetures of the retinopathy area and enhance the adaptability to the differences of different data sets. Finally, the hybrid attention mechanism achieved 96.5% and 99.76% classification accuracy on two public OCT data sets of retinopathy, respectively.

Intravitreal injections during COVID-19 outbreak: Real-world experience from an Italian tertiary referral center.

We report our experience during COVID-19 outbreak for intravitreal injections in patients with maculopathy. We proposed a treatment priority levels and timings; the "High" priority level includes all monocular patients; the "Moderate" is assigned to all patients with an active macular neovascularization; the patients affected by diabetic macular edema or retinal vein occlusion belong to the "Low" class. This organization allowed us to treat the most urgent patients although the injections performed had a 91.7% drop compared to the same period of 2019.

Multiclass retinal disease classification and lesion segmentation in OCT B-scan images using cascaded convolutional networks.

Disease classification and lesion segmentation of retinal optical coherence tomography images play important roles in ophthalmic computer-aided diagnosis. However, existing methods achieve the two tasks separately, which is insufficient for clinical application and ignores the internal relation of disease and lesion features. In this paper, a framework of cascaded convolutional networks is proposed to jointly classify retinal diseases and segment lesions. First, we adopt an auxiliary binary classification network to identify normal and abnormal images. Then a novel, to the best of our knowledge, U-shaped multi-task network, BDA-Net, combined with a bidirectional decoder and self-attention mechanism, is used to further analyze abnormal images. Experimental results show that the proposed method reaches an accuracy of 0.9913 in classification and achieves an improvement of around 3% in Dice compared to the baseline U-shaped model in segmentation.

Implementation of a registry and open access genetic testing program for inherited retinal diseases within a non-profit foundation.

The Foundation Fighting Blindness is a 50-year old 501c(3) non-profit organization dedicated to supporting the development of treatments and cures for people affected by the inherited retinal diseases (IRD), a group of clinical diagnoses that include orphan diseases such as retinitis pigmentosa, Usher syndrome, and Stargardt disease, among others. Over $760 M has been raised and invested in preclinical and clinical research and resources. Key resources include a multi-national clinical consortium, an international patient registry with over 15,700 members that is expanding rapidly, and an open access genetic testing program that provides no cost comprehensive genetic testing to people clinically diagnosed with an IRD living in the United States. These programs are described with particular focus on the challenges and outcomes of establishing the registry and genetic testing program.

Next generation sequencing using phenotype-based panels for genetic testing in inherited retinal diseases.

INTRODUCTION: Diagnostic next generation sequencing (NGS) services for patients with inherited retinal diseases (IRD) traditionally use gene panel based approaches, which have cost and resource implications. Phenotype-based gene panels use a targeted strategy with further testing protocols, if initial results are negative. We present the molecular findings of the Oxford phenotype-based NGS panels for genetic testing in IRD. METHODS: Results of 655 consecutive patients referred for phenotype-based panel testing over 54 months were analysed to assess diagnostic yield. RESULTS: Variants were identified in 450 patients (68.7%). The overall diagnostic yield from phenotype-based panels was 42.8%. The diagnostic yield was highest from panels representing distinct clinical phenotypes: Usher panel 90.9% and congenital stationary night blindness panel 75.0%. Retinitis pigmentosa/rod-cone dystrophy was the commonest presenting phenotype (n = 243) and Usher syndrome was the commonest presenting syndromic disease (n = 39). Patients presenting with late-onset (≥50 years) macular disease had a lower diagnostic yield (18.0%) compared with patients <50 years (24.2%). Additionally, a diagnostic yield of 1.8% was attributable to copy number variants. CONCLUSIONS: Phenotype-based genetic testing panels provide a targeted testing approach and reduce bioinformatics demand. The overall diagnostic yield achieved in this study reflects the wide range of phenotypes that were referred. This pragmatic approach provides a high yield for early-onset and clearly defined genetically determined disorders but clinical utility is not as clear for late-onset macular disorders. This phenotype-based panel approach is clinician-referrer orientated, and can be used as a front-end virtual panel, when whole genome sequencing is introduced into diagnostic services for IRD.

An enhanced OCT image captioning system to assist ophthalmologists in detecting and classifying eye diseases.

Human eye is affected by the different eye diseases including choroidal neovascularization (CNV), diabetic macular edema (DME) and age-related macular degeneration (AMD). This work aims to design an artificial intelligence (AI) based clinical decision support system for eye disease detection and classification to assist the ophthalmologists more effectively detecting and classifying CNV, DME and drusen by using the Optical Coherence Tomography (OCT) images depicting different tissues. The methodology used for designing this system involves different deep learning convolutional neural network (CNN) models and long short-term memory networks (LSTM). The best image captioning model is selected after performance analysis by comparing nine different image captioning systems for assisting ophthalmologists to detect and classify eye diseases. The quantitative data analysis results obtained for the image captioning models designed using DenseNet201 with LSTM have superior performance in terms of overall accuracy of 0.969, positive predictive value of 0.972 and true-positive rate of 0.969using OCT images enhanced by the generative adversarial network (GAN). The corresponding performance values for the Xception with LSTM image captioning models are 0.969, 0.969 and 0.938, respectively. Thus, these two models yield superior performance and have potential to assist ophthalmologists in making optimal diagnostic decision.

VALIDATION OF THE RECENTLY DEVELOPED ATN CLASSIFICATION AND GRADING SYSTEM FOR MYOPIC MACULOPATHY.

PURPOSE: To validate the recently developed ATN grading system for myopic maculopathy to classify eyes with pathologic myopia. METHODS: Cross-sectional study. A series of consecutive eyes diagnosed with pathologic myopia and signs of myopic maculopathy (grade ≥1 for atrophic, tractional, or neovascular components of the ATN), with a refractive error > -6.0 diopters (D), were included. All patients underwent complete ophthalmological examination including fundus photography and swept-source optical coherence tomography. Six observers graded each eye twice using the ATN system (≥15 days between assessments) based only on the aforementioned data. RESULTS: Sixty eyes from 47 patients (61.7% female) were graded. Mean patient age was 63.2 ± 11.7 years. The mean spherical equivalent was -13.8 ± 6.5 D. Mean axial length was 28.6 ± 2.16 mm. Overall, the mean intraobserver agreement (%) for the same image was 92.0%, and the mean interobserver agreement for the second image was 77.5%. The weighted Fleiss k showed excellent correlation (k > 0.8) for the traction and neovascularization components and good correlation (0.75) for atrophy. Interobserver agreement for each of these three components was 95.2%, 98.4%, 95.0%, respectively. CONCLUSION: Application of the ATN resulted in high intraobserver and interobserver correlation, underscoring the reproducibility of the system.

EVALUATION OF ACCURACY AND UNIFORMITY OF THE NOMENCLATURE OF VITREORETINAL INTERFACE DISORDERS.

PURPOSE: To evaluate the accuracy and uniformity of the definitions used to diagnose vitreoretinal (VR) interface disorders and to assess it after review of its definitions. METHODS: A case-series study, consisting of a questionnaire of 46 optical coherence tomography images of six VR interface disorders: vitreomacular adhesion, vitreomacular traction, epiretinal membrane, full-thickness macular hole, lamellar macular hole, and pseudohole. Images were presented to 41 practicing ophthalmologists (13 residents, 11 VR specialists, and 17 non-VR specialists), and a diagnosis was recorded for each image. The questionnaire was repeated after review of the International Vitreomacular Traction Study (IVTS) group classification. Rates of accuracy and uniformity for each condition were analyzed. RESULTS: Overall correct identification rates according to the IVTS classification were achieved in 67.4% of cases and were highest for epiretinal membrane and full-thickness macular hole, followed by vitreomacular adhesion, vitreomacular traction, and lamellar macular hole, and were significantly lower for pseudohole (P < 0.001). Accuracy was higher among VR specialists and was associated with previous familiarity with the IVTS classification (P = 0.043) but not with length of experience in ophthalmology (P = 0.74). After review of the IVTS classification, overall correct identification rates improved to 71.7% (P = 0.004), with the significant improvement in pseudohole identification (P = 0.002). CONCLUSION: The IVTS classification is effective in standardizing the diagnosis of VR interface disorders. It is expected to become increasingly assimilated among ophthalmologists over time, leading to higher rates of accuracy and uniformity in diagnosing VR interface disorders.

AOCT-NET: a convolutional network automated classification of multiclass retinal diseases using spectral-domain optical coherence tomography images.

Since introducing optical coherence tomography (OCT) technology for 2D eye imaging, it has become one of the most important and widely used imaging modalities for the noninvasive assessment of retinal eye diseases. Age-related macular degeneration (AMD) and diabetic macular edema eye disease are the leading causes of blindness being diagnosed using OCT. Recently, by developing machine learning and deep learning techniques, the classification of eye retina diseases using OCT images has become quite a challenge. In this paper, a novel automated convolutional neural network (CNN) architecture for a multiclass classification system based on spectral-domain optical coherence tomography (SD-OCT) has been proposed. The system used to classify five types of retinal diseases (age-related macular degeneration (AMD), choroidal neovascularization (CNV), diabetic macular edema (DME), and drusen) in addition to normal cases. The proposed CNN architecture with a softmax classifier overall correctly identified 100% of cases with AMD, 98.86% of cases with CNV, 99.17% cases with DME, 98.97% cases with drusen, and 99.15% cases of normal with an overall accuracy of 95.30%. This architecture is a potentially impactful tool for the diagnosis of retinal diseases using SD-OCT images.

Morphological Characteristics and Risk Factors of Myopic Maculopathy in an Older High Myopia Population-Based on the New Classification System (ATN).

PURPOSE: To investigate the characteristics, mergers, and risk factors of different types of myopic maculopathy (MM) in a highly myopic population. DESIGN: Population-based, cross-sectional study. METHODS: A total of 1086 eyes (762 patients) were enrolled. Each participant underwent detailed ocular examinations. Combining the fundus photographs and optical coherence tomography images, types of MM were assessed as myopic atrophy maculopathy (MAM), myopic tractional maculopathy (MTM), or myopic neovascular maculopathy (MNM) according to the ATN classification system. Peripapillary atrophy (PPA) area, tilt ratio, and macular choroidal thickness (mChT) were measured individually. RESULTS: Eyes with larger PPA area were more likely to have MAM (odds ratio [OR], 1.220; P = .037 per 1-mm2 increase) and MNM (OR, 1.723; P < .001 per 1-mm2 increase), and eyes with thicker mChT were less likely to have MAM (OR, 0.740; P < .001 per 10-µm increase) and MNM (OR, 0.784; P < .001 per 10-µm increase), whereas eyes with higher tilt ratio were less likely to have MTM (OR, 0.020; P < .001 per 1 increase). The severity of MTM and MNM was not precisely consistent with that of MAM. CONCLUSIONS: Different types of MM have different risk factors; larger PPA area and thinner mChT are risk factors for MAM and MNM, whereas lower tilt ratio is a risk factor for MTM. Our results indicate that the pathogenesis of MTM is different from that of MAM and MNM, and a tractional component should be considered as a possible component to the myopic macular classification.

Myopic maculopathy: Current status and proposal for a new classification and grading system (ATN).

Myopia is a highly frequent ocular disorder worldwide and pathologic myopia is the 4th most common cause of irreversible blindness in developed countries. Pathologic myopia is especially common in East Asian countries. Ocular alterations associated with pathologic myopia, especially those involving the macular area-defined as myopic maculopathy-are the leading causes of vision loss in patients with pathologic myopia. High myopia is defined as the presence of a highly negative refractive error (>-6 to -8 diopters) in the context of eye elongation (26-26.5 mm). Although the terms high myopia and pathologic myopia are often used interchangeably, they do not refer to the same eye disease. The two key factors driving the development of pathologic myopia are: 1) elongation of the axial length and 2) posterior staphyloma. The presence of posterior staphyloma, which is the most common finding in patients with pathologic myopia, is the key differentiating factor between high and pathologic myopia. The occurrence of staphyloma will, in most cases, eventually lead to other conditions such as atrophic, traction, or neovascular maculopathy. Posterior staphyloma is for instance, responsible for the differences between a myopic macular hole (MH)-with and without retinal detachment-and idiopathic MH. Posterior staphyloma typically induces retinal layer splitting, leading to foveoschisis in myopic MH, an important differentiating factor between myopic and emmetropic MH. Myopic maculopathy is a highly complex disease and current classification systems do not fully account for the numerous changes that occur in the macula of these patients. Therefore, a more comprehensive classification system is needed, for several important reasons. First, to more precisely define the disease stage to improve follow-up by enabling clinicians to more accurately monitor changes over time, which is essential given the progressive nature of this condition. Second, unification of the currently-available classification systems would establish standardized classification criteria that could be used to compare the findings from international multicentric studies. Finally, a more comprehensive classification system could help to improve our understanding of the genetic origins of this disease, which is clearly relevant given the interchangeable-but erroneous-use of the terms high and pathologic myopia in genetic research.

The European Eye Epidemiology spectral-domain optical coherence tomography classification of macular diseases for epidemiological studies.

PURPOSE: The aim of the European Eye Epidemiology (E3) consortium was to develop a spectral-domain optical coherence tomography (SD-OCT)-based classification for macular diseases to standardize epidemiological studies. METHODS: A European panel of vitreoretinal disease experts and epidemiologists belonging to the E3 consortium was assembled to define a classification for SD-OCT imaging of the macula. A series of meeting was organized, to develop, test and finalize the classification. First, grading methods used by the different research groups were presented and discussed, and a first version of classification was proposed. This first version was then tested on a set of 50 SD-OCT images in the Bordeaux and Rotterdam centres. Agreements were analysed and discussed with the panel of experts and a final version of the classification was produced. RESULTS: Definitions and classifications are proposed for the structure assessment of the vitreomacular interface (visibility of vitreous interface, vitreomacular adhesion, vitreomacular traction, epiretinal membrane, full-thickness macular hole, lamellar macular hole, macular pseudo-hole) and of the retina (retinoschisis, drusen, pigment epithelium detachment, hyper-reflective clumps, retinal pigment epithelium atrophy, intraretinal cystoid spaces, intraretinal tubular changes, subretinal fluid, subretinal material). Classifications according to size and location are defined. Illustrations of each item are provided, as well as the grading form. CONCLUSION: The E3 SD-OCT classification has been developed to harmonize epidemiological studies. This homogenization will allow comparing and sharing data collection between European and international studies.

Peripheral retinal findings in populations with macular disease are similar to healthy eyes.

PURPOSE: Recent evidence suggests several macular diseases are associated with peripheral retinal changes. This study investigated the number, type and management consequences of peripheral retinal findings detected in patients attending a referral only, eye-care clinic, the Centre for Eye Health(CFEH) with macular disease. METHODS: Records of 537 patients attending CFEH for a macular assessment were included in the study. Subjects were classified as having age-related macular degeneration (AMD), epiretinal membrane (ERM), central serous chorioretinopathy (CSCR), inherited macular dystrophy or no macular disease. Data extracted included reason for referral, macular findings, peripheral findings (based on examination by ultra-widefield scanning laser ophthalmoscopy), diagnosis and management. RESULTS: After age-matching, the number of peripheral findings in subjects with AMD, ERM or CSCR was not significant different to normal subjects. The most common finding for all cohorts were non-specific, degenerative changes such as drusen or pigmentation (61-72%) except inherited macular dystrophy subjects who had mostly vascular findings (30%; p < 0.05). Subjects with AMD and ERM with peripheral findings were significantly more likely to be reviewed or referred to an ophthalmologist than discharged back to their community eye care provider compared to subjects without findings. However only 8% of subjects had altered management based specifically on peripheral findings suggesting the macular findings in most subjects dictated their management. For those with a change, it was significant (upgrade to referral to an ophthalmologist). Peripheral findings also flagged 5% of subjects with vascular findings for referral to their general practitioner (GP). CONCLUSIONS: Overall, the percentage and distribution of peripheral retinal findings in some macular diseases was similar to normal subjects. However, subjects with peripheral findings appeared to have significant differences in management. Considering some common findings, such as peripheral drusen may be relevant to AMD pathogenesis and therefore affect management of this disease, assessment of the peripheral retina should not be overlooked when the clinical focus is on the posterior pole.

Distribution and Severity of Myopic Maculopathy Among Highly Myopic Eyes.

Purpose: The purpose of this study was to document the distribution of the severity of myopic maculopathy in a cohort of highly myopic patients and to explore the associated risk factors. Methods: A total of 890 Chinese highly myopes aged between 7 and 70 years (median age 19 years) and with spherical refraction -6.00 diopter (D) or worse in both eyes were investigated. All participants underwent detailed ophthalmic examination. Myopic maculopathy was graded into 5 categories according to the International Photographic Classification and Grading System using color fundus photographs: category 0, no myopic retinal lesions, category 1, tessellated fundus only; category 2, diffuse chorioretinal atrophy; category 3, patchy chorioretinal atrophy; category 4, macular atrophy. Category 2 or greater were further classified as clinically significant myopic maculopathy (CSMM). Results: Data from 884 of 890 right eyes were available for analysis. The proportions of category 1, category 2, category 3, and category 4 were 20.0% (177 eyes), 20.2% (178 eyes), 2.6% (23 eyes), and 0.2% (2 eyes), respectively. The proportion of CSMM increased with more myopic refraction (odds ratio 1.57; 95% confidence interval: 1.46-1.68), longer axial length (odds ratio 2.97; 95% confidence interval: 2.50-3.53), and older age (40-70 years compared to 12-18 years, odds ratio 6.77; 95% confidence interval: 3.61-12.70). However, there was a higher proportion of CSMM in children aged 7 to 11 years than those aged 12 to 18 years (20.9% vs. 11.0%, P = 0.008). Conclusions: Older age, more myopic refraction, and longer axial length were associated with more severe myopic maculopathy. Although CSMM was uncommon among younger participants, children with early-onset high myopia have a disproportionately increased risk.

Evaluation of light- and dark-adapted ERGs using a mydriasis-free, portable system: clinical classifications and normative data.

PURPOSE: The full-field electroretinogram (ff-ERG) is a widely used clinical tool to evaluate generalized retinal function by recording electrical potentials generated by the cells in the retina in response to flash stimuli and requires mydriasis. The purpose of this study was to determine the intra-visit reliability and diagnostic capability of a handheld, mydriasis-free ERG, RETeval (LKC Technologies, Gaithersburg, MD, USA), in comparison with the standard clinical ff-ERG by measuring responses recommended by the International Society for Clinical Electrophysiology of Vision (ISCEV). METHODS: This prospective, cross-sectional study included 35 patients recruited at the Hospital for Sick Children (median age = 17, range 11 months-69 years) who had undergone a clinical ff-ERG according to ISCEV standards. For RETeval (n = 35), pupils were undilated in most (n = 29) and sensor strip electrodes were placed under the inferior orbital rim. Stimulus settings on RETeval were equivalent to those used in the clinical ERG. Fifty-seven control participants (median age = 22, range 8-65 years) underwent undilated RETeval ERG to establish standard values for comparison. Patient waveform components with amplitudes < 5th percentile, or implicit times > 95th percentile of normal relative to control data were classified as abnormal for the RETeval system. RESULTS: The RETeval system demonstrated a high degree of within-visit reliability for amplitudes (ICC = 0.82) and moderate reliability for implicit times (ICC = 0.53). Cohen's Kappa analysis revealed a substantial level of agreement between the diagnostic capability of RETeval in comparison with clinical ff-ERG (k = 0.82), with a sensitivity and specificity of 1.00 and 0.82, respectively. Pearson's correlations for clinical ERG versus RETeval demonstrated a positive correlation for amplitudes across the rod (r = 0.65) and cone (r = 0.74) ERG waveforms. Bland-Altman plots showed no bias between the mean differences across all amplitude and implicit time parameters of the two systems. CONCLUSIONS: The present study demonstrated that RETeval is a reliable tool with reasonable accuracy in comparison with the clinical ERG. The portable nature of RETeval system enables its incorporation at resource-limited centers where the ff-ERG is not readily available. The avoidance of sedation and pupillary dilation are added advantages of RETeval ERG.

Classification of healthy and diseased retina using SD-OCT imaging and Random Forest algorithm.

In this paper, we propose a novel classification model for automatically identifying individuals with age-related macular degeneration (AMD) or Diabetic Macular Edema (DME) using retinal features from Spectral Domain Optical Coherence Tomography (SD-OCT) images. Our classification method uses retinal features such as the thickness of the retina and the thickness of the individual retinal layers, and the volume of the pathologies such as drusen and hyper-reflective intra-retinal spots. We extract automatically, ten clinically important retinal features by segmenting individual SD-OCT images for classification purposes. The effectiveness of the extracted features is evaluated using several classification methods such as Random Forrest on 251 (59 normal, 177 AMD and 15 DME) subjects. We have performed 15-fold cross-validation tests for three phenotypes; DME, AMD and normal cases using these data sets and achieved accuracy of more than 95% on each data set with the classification method using Random Forrest. When we trained the system as a two-class problem of normal and eye with pathology, using the Random Forrest classifier, we obtained an accuracy of more than 96%. The area under the receiver operating characteristic curve (AUC) finds a value of 0.99 for each dataset. We have also shown the performance of four state-of-the-methods for classification the eye participants and found that our proposed method showed the best accuracy.