Classification of Optical Coherence Tomography using Convolutional Neural Networks A. A. Saraiva 2,6 a , D. B. S. Santos 1 b , Pimentel Pedro 1 c , Jose Vigno Moura Sousa 1 d , N. M. Fonseca Ferreira 3,4 e , J. E. S. Batista Neto 6 , Salviano Soares 3 f and Antonio Valente 2,5 g 1 UESPI - University of State Piaui, Piripiri, Brazil 2 University of Tr´ as-os-Montes and Alto Douro,Vila Real, Portugal 3 Coimbra Polytechnic, ISEC, Coimbra, Portugal 4 Knowledge Engineering and Decision-Support Research Center (GECAD) of the Institute of Engineering, Polytechnic Institute of Porto, Portugal 5 INESC-TEC Technology and Science, Porto, Portugal 6 University of S˜ ao Paulo, S˜ ao Carlos, Brazil Keywords: OCT, CNN, Classification, K-fol, Labeled Optical Coherence Tomography. Abstract: This article describes a classification model of optical coherence tomography images using convolution neural network. The dataset used was the Labeled Optical Coherence Tomography provided by (Kermany et al., 2018) with a total of 84495 images, with 4 classes: normal, drusen, diabetic macular edema and choroidal neovascularization. To evaluate the generalization capacity of the models k-fold cross-validation was used. The classification models were shown to be efficient, and as a result an average accuracy of 94.35% was obtained. 1 INTRODUCTION An examination known as optical coherence tomog- raphy (OCT) has gained ground in the latest comple- mentary clinical tests for the diagnosis of retinal and vitreous disease (Preti et al., 2018). This technology was developed by Fujimoto at the Massachusetts Institute of Technology, applied in the ophthalmological diagnosis by Puliafito. The use of this examination has become fundamental in the diag- nosis, on evolution and postoperative control of mul- tiple macular conditions (Dimitrova et al., 2017). According to (Swanson and Fujimoto, 2017) ap- proximately 30 million procedures of optical coher- ence tomography (OCT) images are performed per year, the analysis and interpretation of these images, a https://orcid.org/0000-0002-3960-697X b https://orcid.org/0000-0003-4018-242X c https://orcid.org/0000-0002-5291-0810 d https://orcid.org/0000-0002-5164-360X e https://orcid.org/0000-0002-2204-6339 f https://orcid.org/0000-0001-5862-5706 g https://orcid.org/0000-0002-5798-1298 consumes a significant amount of time. OCT helped patients prevent or minimize vision loss by detecting retinal diseases in the early stages of treatment (Swan- son and Fujimoto, 2017). According to (Sivaprasad and Moore, 2008), the growth of new choroidal blood vessels is known as choroidal neovascularization (CNV). These new ves- sels come from a rupture in the Bruch membrane that is located in the subretinal pigment epithelium. Ac- cording to (Baxter et al., 2013) CNV occurs in about 2 to 3% of cases of posterior uveitis. Diabetic macular edema (DME) is a complication of diabetes caused by fluid accumulation in the mac- ula, or central portion of the eye, that causes the mac- ula to swell (Wells et al., 2016). The macula is filled with cells responsible for direct vision that aid in read- ing and directing (Wells et al., 2016). When the macula begins to fill with fluid and swell, the capacity of these cells is impaired, caus- ing blurred vision (Bressler et al., 2016). The DME is diabetic retinopathy, in which the blood vessels of the eye are damaged, allowing the fluid to escape, this type of disease can also be diagnosed through the 168 Saraiva, A., Santos, D., Pedro, P., Sousa, J., Ferreira, N., Neto, J., Soares, S. and Valente, A. Classification of Optical Coherence Tomography using Convolutional Neural Networks. DOI: 10.5220/0009091001680175 In Proceedings of the 13th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2020) - Volume 3: BIOINFORMATICS, pages 168-175 ISBN: 978-989-758-398-8; ISSN: 2184-4305 Copyright c  2022 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved