DEVELOPMENT OF CONVOLUTIONAL NEURAL NETWORK BASED ALGORITHM FOR COMPUTED TOMOGRAPHY IMAGES OF HUMAN KIDNEY STONE(Renal Calculi)

Abstract

Computed tomography (CT) scan diagnostics procedures adopt use of image information retrieval system through expertise of radiographer; this is prone to error. Significant height of accuracy is required in healthcare decision support, as 20% of CT scan are associated with error. The application of artificial intelligence (AI) can improve performance level, mitigate human error, and enhance clinical decision support in the context of time and accuracy. This study applies machine learning algorithm to analyze a stream of anonymous CT scan kidney images. This research adopts deep learning approach for segmentation and classification of kidney stone (renal calculi) images in python (with Keras and TensorFlow) environment. A control volume of data, three hundred and thirty-six (336) kidney stone images in PNG format were used to train the deep learning network with ten (10) testing images. The training images were divided into two sets (folders) as follows; one was labeled as STONE (containing 167 images) and the other as NO-STONE (containing 169 images); twenty (20) iterations were performed for the training of model. The network layers are structured as follows: input layer with 2-D convolutional neural network machine learning (CNN-ML), ReLU activation, Maxpooling, fully connected (dense) layer (including the sigmoid activation layer). The training applied a batch size of eight (8) with 10% validation. The output result, upon testing the model, it records an accuracy of 90%, sensitivity value of 80% and effectiveness of 89%. The segmentation and classification module could be embedded in future computed tomography (CT) diagnostic procedure to enhance medical decision support and accuracy.