Classification of Glaucoma in Fundus Images Using Convolutional Neural Network with MobileNet Architecture

Abstract

Glaucoma is a damaged optic nerve due to increased pressure on the eyeball. The cause is a mismatch between eye fluid (aqueous humor) produced and the amount of eye fluid secreted. Ophthalmologists usually detect glaucoma using Cup to Disc Ratio or CDR parameter. However, the calculation of CDR parameters is still done manually, usually done by trained doctors and relatively expensive and limited equipment. This study proposes a system that can classify glaucoma using the Convolutional Neural Network method with MobileNet architecture. MobileNet has two convolution parts: depthwise convolution and pointwise convolution. The function of the Depthwise Convolution is to apply a single convolution filter per input channel, while the function of the pointwise convolution is to build new features by calculating the linear combination of the input channels by applying the 1x1 convolution. The data used comes from rimone-r1 database. Result accuracy of the proposed method reaches 99%. Automated glaucoma classification can assist medical staff in identifying the best treatment for their patients.