Today, unprecedented advances in machine learning applied to biomedicine have led to a variety of unsupervised techniques for remote image analysis, enabling cost-effective early detection of diseases. In this talk I will present various methods for ophthalmic image analysis, which take advantage of network theory and non-linear data analysis tools. First, I will present an unsupervised machine learning algorithm for optical coherence tomography (OCT) image analysis, which extracts features that discriminate between healthy and unhealthy subjects [1]. Then, I will show how the analysis of the tree-like structure of the network of vessels in the retina allows to extract features that discriminate between healthy subjects and those with glaucoma or diabetic retinopathy [2]. Finally, I will discuss how the network percolation transition can be exploited for outlier mining in different types of data sets, including ophthalmic images [3].
Figure 1: Example of a retinal fundus image and the extracted network.
References
[1] P. Amil, L. Gonzalez, E. Arrondo, C. Salinas, J. L. Guell, C. Masoller, and U. Parlitz, Unsupervised feature extraction of anterior chamber OCT images for ordering and classification, Sci. Rep. 9, 1157 (2019).
[2] P. Amil, F. Reyes-Manzano, L. Guzmán-Vargas, I. Sendiña-Nadal and C. Masoller, Novel network-based methods for retinal fundus image analysis and classification, PLoS ONE 14, e0220132 (2019).
[3] P. Amil, N. Almeira and C. Masoller, Outlier mining methods based on network structure analysis, Front. Phys. 7, 194 (2019).