Research Article
AI-Driven CKD Diagnosis: A Deep Learning Framework for Accurate and Efficient Kidney Disease Detection
@INPROCEEDINGS{10.4108/eai.28-4-2025.2357754, author={C. Ravindra Murthy and Manchineela Madhuri and Mayana Saif Ali Khan and Ragala Venkata Narasimha Prasad and Vama Narendra}, title={AI-Driven CKD Diagnosis: A Deep Learning Framework for Accurate and Efficient Kidney Disease Detection}, proceedings={Proceedings of the 4th International Conference on Information Technology, Civil Innovation, Science, and Management, ICITSM 2025, 28-29 April 2025, Tiruchengode, Tamil Nadu, India, Part I}, publisher={EAI}, proceedings_a={ICITSM PART I}, year={2025}, month={10}, keywords={chronic kidney disease (ckd) image-based detection convolutional neural network (cnn) matlab kaggle dataset median filter fuzzy c-means (fcm) clustering etc}, doi={10.4108/eai.28-4-2025.2357754} }- C. Ravindra Murthy
Manchineela Madhuri
Mayana Saif Ali Khan
Ragala Venkata Narasimha Prasad
Vama Narendra
Year: 2025
AI-Driven CKD Diagnosis: A Deep Learning Framework for Accurate and Efficient Kidney Disease Detection
ICITSM PART I
EAI
DOI: 10.4108/eai.28-4-2025.2357754
Abstract
This paper presents a deep learning-based approach for the detection of Chronic Kidney Disease (CKD) using Convolutional Neural Networks (CNNs). The proposed method involves a comprehensive pipeline that includes image pre-processing using median filtering, segmentation through Fuzzy C-Means (FCM) clustering, and feature extraction based on statistical and texture-based measures. The CNN classifier is trained on a dataset of kidney images and evaluated based on key performance metrics such as accuracy, sensitivity, and time consumption. Experimental results demonstrate that the proposed CNN model achieves an accuracy of 97% and a sensitivity of 88%, significantly outperforming traditional machine learning algorithms like Naïve Bayes, Decision Tree, and Support Vector Machine (SVM), which achieved accuracies of 91%, 93.2%, and 94.1%, respectively. Additionally, the CNN model shows the lowest time consumption at 21%, highlighting its efficiency in real- time applications. These results confirm the effectiveness of the CNN-based approach for early-stage CKD detection, offering a promising solution for improving diagnostic accuracy in clinical settings.
