A Comparative Study Between Clinical Optical Coherence Tomography (OCT) Analysis and Artificial Intelligence-Based Quantitative Evaluation in the Diagnosis of Diabetic Macular Edema

Abstract

Recent advances in artificial intelligence (AI) have transformed ophthalmic diagnostics, particularly for retinal diseases. In this prospective, non-randomized study, we evaluated the performance of an AI-based software system against conventional clinical assessment-both quantitative and qualitative-of optical coherence tomography (OCT) images for diagnosing diabetic macular edema (DME). A total of 700 OCT exams were analyzed across 26 features, including demographic data (age, sex), eye laterality, visual acuity, and 21 quantitative OCT parameters (Macula Map A X-Y). We tested two classification scenarios: binary (DME presence vs. absence) and multiclass (six distinct DME phenotypes). To streamline feature selection, we applied paraconsistent feature engineering (PFE), isolating the most diagnostically relevant variables. We then compared the diagnostic accuracies of logistic regression, support vector machines (SVM), K-nearest neighbors (KNN), and decision tree models. In the binary classification using all features, SVM and KNN achieved 92% accuracy, while logistic regression reached 91%. When restricted to the four PFE-selected features, accuracy modestly declined to 84% for both logistic regression and SVM. These findings underscore the potential of AI-and particularly PFE-as an efficient, accurate aid for DME screening and diagnosis.

Keywords: artificial intelligence; diabetic macular edema; machine learning; optical coherence tomography; paraconsistent feature engineering; retinal diseases; support vector machine.

Figure 1

Paraconsistent plane: distribution of features by certainty and contradiction. Adapted from [27]. CV: Coefficient of Variation.

Figure 2

Confusion matrix for SVM/KNN with 24 features (binary scenario). This matrix shows that of the 26 actual edema cases (Y), 17 were correctly predicted, while nine were classified as negative (false negatives). Of the 114 negative cases (N), 112 were correctly predicted and only two were incorrectly classified as positive (false positives) [18].

Figure 3

ROC curve for SVM with 24 features (multiclass scenario). Displays micro-average and per-class AUC scores, illustrating the model’s ability to differentiate multiple phenotypes.

Figure 4

ROC curve for SVM model with 4 PFE features (multiclass scenario). Performance curve of the paraconsistent model, highlighting improvement in Y class detection and overall reduction in accuracy for rare classes.