Abstract:
As the years pass by, more and more diseases are being diagnosed and there has been a gradual
increase in the number of patients suffering from diabetic. With the advancing technology,
medical sciences have been fast-forwarding towards machines that help in the diagnosis of
the disease in every field. In our thesis, we worked with diabetic retinopathy images to classify
between retinopathy eye and normal eye using a convolutional neural network. The
convolutional neural network is very efficient for image classification problems as it extracts
the features from the images that help us differentiate between the normal eye and retinopathy
eye. The implementation of such technology has enabled the doctors to provide faster and
better treatments to patients suffering from diabetic. Many diseases can be diagnosed through
such systems in hospitals. Some of the hospitals have already started using these systems and
the results are quite impressive. This will benefit both the people and the doctors as the
treatments will be provided within a short period and the patients don’t have to suffer through
diabetic retinopathy. We developed a convolutional neural network model to classify diabetic
retinopathy images in [1] dataset. The dataset is an open-source diabetic retinopathy dataset
that has about 35126 images of various patients appropriate for our research.
Description:
This thesis submitted in partial fulfillment of the requirements for the degree of Bachelor of Science in Computer Science and Engineering of East West University, Dhaka, Bangladesh.