Diagnosis of Kidney Disease Using Pre-Trained Convolutional Neural Network and Multilayer Perceptron

Abstract

Chronic kidney disease has become an increasingly significant global health concern, highlighting the need for accurate and early diagnostic approaches. In ultrasound-based kidney disease assessment, the extraction of informative and discriminative features plays a crucial role in achieving reliable classification. In this study, a hybrid deep learning framework is proposed for the diagnosis of chronic kidney disease from renal ultrasound images. A pre-trained ResNet-50 convolutional neural network is employed to extract high-level image features, generating a 1000-dimensional feature vector for each ultrasound image. Principal component analysis is subsequently applied to reduce feature dimensionality, eliminate redundant information, and minimize correlations among the extracted features. The resulting optimized feature set is then classified using a multilayer perceptron neural network. The proposed ResNet-50-PCA-MLP framework achieved an overall classification accuracy of 98.1\%, demonstrating its effectiveness and potential as a computer-aided diagnostic tool for the early detection of chronic kidney disease.