Deep Learning Based Kidney Stone Detection Using CT Scan Images

Kidney stones have emerged as a significant health concern, and their early detection is crucial to avoid complications, often requiring surgical intervention for stone removal. Our primary objective is to develop a model that aids doctors in diagnosing kidney stones. CT scans have proven to be the most accurate diagnostic test for identifying kidney stones, as small stones can be missed through ultrasonography. Consequently, we have undertaken an in-depth analysis of kidney stone detection using image processing techniques on CT images. The global significance of kidney stone detection underscores the importance of identifying its presence for surgical planning. The diagnosis of kidney stones is conducted using a CT scan without contrast dye, providing detailed images spanning from the upper part of the kidneys to the base of the bladder, surpassing the capabilities of traditional X-rays. Additionally, our dataset encompasses classifications of cysts and tumors, along with distinguishing stones from normal tissue. To enhance the accuracy of kidney stone detection in CT images, we employ image pre-processing techniques to remove noise from the input images. Subsequently, the pre-processed images are fed into a hybrid model, combining a Convolutional Neural Network with SVM, for stone detection in CT scan images.