DIABETES DETECTION USING CONVOLUTIONAL NEURAL NETWORKS (CNN) BASED MODEL

Abstract

Diabetes, also known as Diabetes Mellitus is a disease that happens to a person
when one’s blood glucose or blood sugar is extremely high. Insulin is a hormone secreted by
the organ pancreas and helps to convert the blood glucose into useful energy for the body. In
some cases, the body doesn’t produce enough, or any amount of insulin or doesn’t use the
produced insulin properly. Hence the Glucose remains in the blood and doesn’t reach the
body cells. Thus, having a lot of glucose in your blood can causes health problems, which is
what exactly happens in Diabetes. Long- term complications of Diabetes develop gradually.
Having Diabetes for a long time along with uncontrolled blood sugar levels can cause
dangerous complications. In the due course, diabetes complications may be disabling or even
life-threatening. Making things even worse, there is no cure for this disease yet! Even though
there’s no cure for diabetes, it can be treated and controlled, and some people may go into a
state of remission. But the very first step towards controlling and minimizing the ill effects of
Diabetes is – the early detection of the disease! Thus, we need comfortable, reliable, and
quick methods of detection. Hence, we are proposing an efficient, reliable, comfortable, and
time-saving Diabetes detection system for Diabetes detection using diabetic Retinopathy and
Implementation of Convolutional Neural Networks (CNN). The level of Diabetic retinopathy
present will also give a direct indication about the level of Diabetes. The implementation of
this method of Diabetes detection will increase accuracy, efficiency, and ease of Diabetes
detection, and further the prognosis and treatment. Also, it will prove to be a better alternative
to conventional testing for the disease