The use of optical coherence tomography (OCT) in medical diagnostics is now common. The growing amount of data leads us to propose an automated support system for medical staff. The key part of the system is a classification algorithm developed with modern machine learning techniques. The main contribution is to present a new approach for the classification of eye diseases using the convolutional neural network model. The research concerns the classification of patients on the basis of OCT B-scans into one of four categories: Diabetic Macular Edema (DME), Choroidal Neovascularization (CNV), Drusen, and Normal. Those categories are available in a publicly available dataset of above 84,000 images utilized for the research. After several tested architectures, our 5-layer neural network gives us a promising result. We compared them to the other available solutions which proves the high quality of our algorithm. Equally important for the application of the algorithm is the computational time, which is reduced by the limited size of the model. In addition, the article presents a detailed method of image data augmentation and its impact on the classification results. The results of the experiments were also presented for several derived models of convolutional network architectures that were tested during the research. Improving processes in medical treatment is important. The algorithm cannot replace a doctor but, for example, can be a valuable tool for speeding up the process of diagnosis during screening tests.
Background: Plus disease is a treatment-relevant sign of retinopathy of prematurity, but fundus images from the same infant may differ in image quality, posterior-pole coverage, and visible vascular severity. We evaluated deep-learning classification using a patient-disjoint design, within-infant variability, and an ophthalmologist reader study. Methods: The primary analysis included 732 Phoenix ICON fundus images from 38 infants, assigned exclusively to a 634-image development pool or a 98-image hold-out set from eight previously unseen infants. Five pretrained convolutional neural networks were evaluated at the image level. Hold-out predictions were grouped by infant and laterality to characterize descriptive within-infant variation. A historical secondary analysis used a random image-level partition without enforced patient separation and was considered exploratory. Two ophthalmologists independently graded 30 unique isolated images for inter-observer analysis; 10 were presented twice, resulting in 40 presentations for intra-observer repeatability. Results: DenseNet121 achieved an image-level ROC-AUC of 0.929; however, infant-equal-weighted image-level ROC-AUC was 0.905 and leave-one-infant-out ROC-AUC ranged from 0.838 to 0.989, demonstrating sensitivity to the small clustered hold-out cohort. At a development-defined threshold of 0.0184, sensitivity increased to 98.0%, while specificity decreased to 46.8%. Within-infant probability variability was observed across all five CNN architectures. Ophthalmologists 1 and 2 assessed 14/18 and 17/18 plus-reference images, respectively, as Healthy. Exact three-category inter-observer agreement was 26/30 (86.7%), and both reproduced the same severity grade for all 10 repeated images. Conclusions: Patient-disjoint evaluation revealed substantial uncertainty in image-level performance and variability among images from the same infant. Cluster-aware analyses showed dependence on the small hold-out cohort, while the development-defined operating point demonstrated a marked sensitivity–specificity trade-off. The reader study showed that an examination-derived reference label and the diagnostic evidence visible in an isolated image are not equivalent, despite high inter-observer agreement and complete categorical repeatability. Because examination identifiers were unavailable, within-infant analyses were interpreted descriptively and not as infant or examination-level diagnostic performance. Clinically useful systems should assess image quality, select informative views, and integrate evidence across multiple images and both eyes. External prospective multicentre validation is required before clinical implementation.