Classification of kidney abnormalities using deep learning with explainable AI

Abstract

Kidney is a critical organ in the human body, and any damage to it causes substantial injury to the body. Early detection is essential. As a result, Artificial Intelligence, particularly deep learning, is being used to detect kidney disorders at an early stage. This paper focuses on the classification of kidney diseases using a deep learning model notably XResNet50. The dataset includes 9527 CT pictures in both the axial and coronal planes, and it is split into three categories: normal, stone, and tumor. The model was used to distinguish between these three classes using training and validation data, and it was applied to predict unseen testing data. The model achieved a testing accuracy of 97%, a precision of 0.97, a sensitivity of 0.97 and a F1-score of 0.97. Furthermore, the model was able to recognize the kidney region in the CT-scan of the Stone, Normal, and most tumor classes based on the SHAP images.

Ähnliche Arbeiten