ISSN 2063-5346
For urgent queries please contact : +918130348310

Multi-Tier and Pool Residual Convolutional Neural Network Architecture for Glaucoma Grade Classification

Main Article Content

A.Padma, Dr.M.Sivajothi, Dr.M.Mohamed Sathik
» doi: 10.48047/ecb/2023.12.9.257


Technology development has its major footsteps in medical applications. One of the major health care applications is glaucoma detection. There are several researches based on cup and disc segmentation of retinal images. As neural networks has its development in research field, several researchers focus on Convolutional Neural Network (CNN) for their research. In this research, a novel Multi-Tier and Pool Residual CNN (MTPRCNN) is designed for glaucoma detection. Due to reduction in computational cost, training time of network and overfitting, pool residual CNN is used. On account of better feature extraction and improved network training, multitier CNN is implemented as a combination with pool residual CNN. It consists of three tiers with three convolutional layers. This research uses MIAG RIM-one r1 dataset for its experiments. The dataset is divided into five different grades. The experimental results obtained 96.7% accuracy with 90.9% sensitivity. It is also tested with Random Forest (RF) Classifier which gives 97.5% accuracy with 93.6% sensitivity.

Article Details