Classification of Kidney Ailments using CNN in CT Images

CT-SCAN (computer tomography) is a non-invasive diagnostic imaging procedure which helps us to have clear and detailed images of any specific part of the body. It produces horizontal or axial images of the body. In order to produce these slices of the body parts, it uses a combination of x-rays along with computer technology. In this research work, the authors have used the CT scan images of kidney. The kidneys are basically a fist sized and beam shaped organ which serves certain important purposes in a human body such as detoxification of body, balancing the fluids in the body, maintaining the levels of electrolytes such as sodium, potassium, calcium etc, removing wastes via urine etc,this it is very important to ensure the proper health of the kidneys. The dataset used in the research has total 12510 images labelled into 4 different classes - cyst, tumor, stone and normal. The dataset was collected from various hospitals in Dhaka. The authors tried to build a model that would classify the input images into the above 4 classes. The authors built a convolutional neural network using Sigmoid and ReLu as activation functions to classify these images. The model was then evaluated based on performance metrics such as accuracy, precision, recall and specificity. The network was able to have almost 100% performance for all the evaluation metrics.