chapter  17
39 Pages

Deep Learning for Retinal Analysis

WithHenry A. Leopold, John S. Zelek, Vasudevan Lakshminarayanan

This chapter provides a introduction to deep learning with applications focused on retinal image analysis in the realm of health informatics. The chapter begins by describing the similarities of computational and biological neural networks. The next section provides an overview of the human vision system and retinal disease as a basis for the subsequent sections. Following is a brief overview of computer-aided diagnostics as it relates to diabetic retinopathy. The last part of this chapter frames a set of biomedical cases before discussing the performances of traditional and deep methods within that context. These cases represent the three main types of classification tasks as well as three critical areas of retinal health assessments: The first case delves into pixe-level classification with a focus on algorithms for retinal vessel segmentation. The second case investigates microaneurysm detection and the use of region-based classification. The third case discusses diabetic retinopathic screening methods and image-level classification.