Multiclass model for early Diagnosis of Renal Calculi using Deep learning techniques

The kidney crystals formed from the crystallization of certain minerals and salts, including oxalate, uric acid, and calcium compounds, can lead to the formation of kidney stones. If ignored, kidney stones can be excruciating along with other complications. Though several imaging methods have emerged for the diagnosis and classification of kidney stones, problems like reduced early-stage accuracy, increased computer complexity and longer patient styluses are still oriented in time. In this paper, three models such as DenseNet-121, Inception V3, and VGG16 are considered for the problem of kidney stones detection and their classification by ultrasound images. The Inception V3 surpassed accuracy by 99%, precision by 90%, F1 score by 89% and recall by 97%. Such conclusions indicate that advancing toward novel deep learning-based approaches can prove to be very beneficial in the early detection and management of kidney stones.