Automatic Identification of Glaucoma from Circumpapillary OCT Images Through the Use of Convolutional Neural Networks

• Published in: 2023 International Conference on Data Science, Agents & Artificial Intelligence (ICDSAAI) pp. 1 - 7
• Main Authors: Nishanth, J., Janarthanan, R.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 21-12-2023
• Subjects: Circumpapillary OCT images; Convolutional Neural Networks (CNN); Glaucoma; Machine learning; Memory management; Neural networks; Optical Coherence Tomography (OCT); Pathology; Tomography; Training; Transfer learning
• DOI: 10.1109/ICDSAAI59313.2023.10452491

The main objective of this work is the development of automatic classification models based on deep learning that, using circumpapillary OCT images, allow us to differentiate between healthy patients and patients suffering from glaucoma. This work focuses on design, development and implementation of its own convolutional neural architecture through the use of different convolutional blocks. After that, various modifications will be proposed with the aim of obtaining better classification results. This work further make use of complex pre-trained architectures through the knowledge transfer technique, so that it allows a comparison to be made with the model designed from scratch, assessing the different advantages and disadvantages that this entails. The choice of the classification model that allows for better discernment between healthy and glaucomatous patients based on various metrics and criteria.