Classification of dry age‐related macular degeneration and diabetic macular oedema from optical coherence tomography images using dictionary learning. Issue 8 (14th May 2020)

• Record Type: Journal Article
• Authors: Mousavi, Elahe; Kafieh, Rahele; Rabbani, Hossein
• Is Part Of: IET image processing. Volume 14:Issue 8(2020)
• Page Start: 1571
• Page End: 1579
• Publication Date: 2020-05-14
• Subjects: feature extraction -- vision defects -- image classification -- diseases -- image segmentation -- eye -- biomedical optical imaging -- medical image processing -- optical tomography; dry age‐related macular degeneration -- diabetic macular oedema -- optical coherence tomography images -- AMD -- DME -- vision loss -- developed countries -- retinal layer structure -- diseases -- automatic classification -- classification algorithm -- retinal layer segmentation -- normal region identifications -- abnormal region identifications -- local intensity gradients -- dictionary learning‐based classifiers -- normal subjects -- normal OCT images -- edge directions -- extracted features; Image processing -- Periodicals; 621.36705
• DOI: 10.1049/iet-ipr.2018.6186
• Languages: English
• ISSNs: 1751-9659

Abstract

Abstract : Age‐related Macular Degeneration (AMD) and Diabetic Macular Edema (DME) are the major causes of vision loss in developed countries. Alteration of retinal layer structure and appearance of exudates are the most significant signs of these diseases. In this paper, with the aim of automatic classification of DME, AMD, and normal subjects using Optical Coherence Tomography (OCT) images, a dictionary‐learning based classification is proposed. The two important issues intended in this approach are avoiding retinal layer segmentation and attempting to mimic the authors' understanding based on normal and abnormal region identifications, considering that the signs of diseases appear in a small fraction of B‐Scans. The histogram of oriented gradients feature descriptor was utilized to characterize the distribution of local intensity gradients and edge directions. To capture the structure of extracted features, different dictionary learning‐based classifiers are employed. The dataset consists of 45 subjects: 15 patients with AMD, 15 patients with DME, and 15 normal subjects. The proposed classifier leads to an accuracy of 95.13, 100.00, and 100.00% for DME, AMD, and normal OCT images, respectively, only by considering 4% of all B‐Scans of a volume, which outperforms the state‐of‐the‐art methods.