An Outperforming Artificial Intelligence Model to Identify Referable Blepharoptosis for General Practitioners.

The aim of this study is to develop an AI model that accurately identifies referable blepharoptosis automatically and to compare the AI model's performance to a group of non-ophthalmic physicians. In total, 1000 retrospective single-eye images from tertiary oculoplastic clinics were labeled by three oculoplastic surgeons as having either ptosis, including true and pseudoptosis, or a healthy eyelid. A convolutional neural network (CNN) was trained for binary classification. The same dataset was used in testing three non-ophthalmic physicians. The CNN model achieved a sensitivity of 92% and a specificity of 88%, compared with the non-ophthalmic physician group, which achieved a mean sensitivity of 72% and a mean specificity of 82.67%. The AI model showed better performance than the non-ophthalmic physician group in identifying referable blepharoptosis, including true and pseudoptosis, correctly. Therefore, artificial intelligence-aided tools have the potential to assist in the diagnosis and referral of blepharoptosis for general practitioners.

A deep learning approach to identify blepharoptosis by convolutional neural networks.

PURPOSE: Blepharoptosis is a known cause of reversible vision loss. Accurate assessment can be difficult, especially amongst non-specialists. Existing automated techniques disrupt clinical workflow by requiring user input, or placement of reference markers. Neural networks are known to be effective in image classification tasks. We aim to develop an algorithm that can accurately identify blepharoptosis from a clinical photo. METHODS: A total of 500 clinical photographs from patients with and without blepharoptosis were sourced from a tertiary ophthalmic center in Taiwan. Images were labeled by two oculoplastic surgeons, with an independent third oculoplastic surgeon to adjudicate disagreements. These images were used to train a series of convolutional neural networks (CNNs) to ascertain the best CNN architecture for this particular task. RESULTS: Of the models that trained on the dataset, most were able to identify ptosis images with reasonable accuracy. We found the best performing model to use the DenseNet121 architecture without pre-training which achieved a sensitivity of 90.1 % with a specificity of 82.4 %, compared to the worst performing model which was used a Resnet34 architecture with pre-training, achieving a sensitivity of 74.1 %, and specificity of 63.6 %. Models with and without pre-training performed similarly (mean accuracy 82.6 % vs. 85.8 % respectively, p = 0.06), though models with pre-training took less time to train (1-minute vs. 16 min, p < 0.01). CONCLUSIONS: We report the use of AI to accurately diagnose blepharoptosis from a clinical photograph with no external reference markers or user input requirement. Most current-generation CNN architectures performed reasonably on this task, with the DenseNet121, and Resnet18 architectures without pre-training performing best in our dataset.