Social Group Optimization Assisted Image Processing Practice to Extract Retinal Optic Disc/Cup

摘要

Examination of retina and its associated parts, such as optic-nerve, cup, and disc are widely preferred in Ophthalmology to identify the retinal abnormalities and to plan/implement appropriate treatment procedure to cure the disease. This paper implements a Machine-Learning (ML) technique to extract the Optic Disc/Cup from the Benchmark-Fundus-Image-Dataset (BFID). The proposed ML technique is implemented by integrating the Social-Group-Optimization (SGO) monitored Kapur's thresholding and Level-Set-Segmentation (LSS). In this work, the BFID, such as RIM-ONE and IDRiD are chosen for the examination and the essential retinal parts such as Optic Disc/Cup are extracted using the proposed ML technique. After mining the essential part from the chosen BFID, the advantage of the proposed approach is then appraised by employing a relative analysis among mined retinal section and its associated ground-truth. Further, the performance of SMOA is validated against the Bacterial-Foraging-Optimization (BFO), Teaching-Learning-Based-Optimization (TLBO) and Firefly-Algorithm (FA) existing in the literature. The result of this work validates that, SGO based practice is capable in investigative the BFID pictures with significant image similarity measures.