Optic disc analysis in retinal fundus using L2 norm of contourlet subbands, superimposed edges, and morphological filling

Autor: G. Sangeethapriya, S. Gayathri, S. Anand
Rok vydání: 2021
Předmět:
Zdroj: Multimedia Tools and Applications. 81:36129-36152
ISSN: 1573-7721
1380-7501
DOI: 10.1007/s11042-021-11569-6
Popis: Optic disc (OD) analysis is an important stage in detecting retinal diseases and existing approaches are not suitable for analyzing multiple examinations in a single process. This paper proposes a new unified method to detect (i) OD center, (ii) OD boundary, and (iii) enhancement of vasculature within the OD region in a single algorithm. This paper presents a unique strategy, which involves L2 norm of contourlet subbands of retinal images to locate the OD center. A novel OD boundary tracing approach that integrates morphological operations, and segmentation region growing techniques, with the aid of the OD center as a seed point, is presented. Among the two novel OD vasculature enhancement techniques, one of which involves image sharpening to improve the local contrast followed by histogram equalization to increase the overall contrast of the fundus image. Edge superimposing with the flood fill technique is another approach used for vasculature enhancement. The algorithm is tested on DRIVE (40), STARE (81), MESSIDOR (1200), E-ophtha (87), Diaretdb1 (89) images, out of which 40, 77, 1182, 87, and 87 images are detected correctly. The Figure of Merit (FOM) is used to confirm the correctness of the boundary tracking performance (97.15%). EME, SSIM, and other parameters are used to assess the efficiency of the vasculature enhancement technique.
Databáze: OpenAIRE
Externí odkaz: