Enhancing Strabismus Diagnosis from Detection to Classification with Deep Learning

An accurate and early diagnosis of strabismus, a disorder characterized by misalignment of the ocular system, is essential to prevent long-term visual impairment. Traditional diagnostic approaches heavily rely on clinical expertise, often introducing subjectivity and inconsistency. In this study, we have forwarded a deep learning-based framework for automated stra-bismus classification using four state-of-the-art convolutional neu-ral networks: EfficientNet-B7, ResNet-50, AlexNet, and VGGNet-19. The models are trained on a diverse open-source dataset to classify strabismus into five categories: Exotropia, Esotropia, Hypertropia, Hypotropia, and Normal eye alignment. Our exten-sive experiments demonstrate that EfficientNet-B7 outperforms other models, achieving superior accuracy and generalization across performance evaluation metrics, including accuracy, pre-cision, recall, and F1-score. The proposed AI -driven approach offers an objective, scalable, consistent alternative to manual diagnosis, significantly enhancing clinical decision-making. This study underscores the transformative potential of deep learning in ophthalmology, paving the way for automated, high-accuracy strabismus screening in real-world clinical settings. Future work will explore real-time implementation and integration of additional clinical parameters to optimize performance and usability further.