Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Hybrid Deep Learning Framework for Classification of Kidney CT Images: Diagnosis of Stones, Cysts, and Tumors

Kiran Sharma, Ziya Uddin, Adarsh Wadal, Dhruv Gupta

TL;DR

This work tackles four-class kidney CT image classification (normal, stone, cyst, tumor) and introduces a hybrid CNN that fuses a pre-trained ResNet101 with a custom CNN via a dual-branch architecture. The ResNet101 branches specialize in normal-vs-stone and cyst-vs-tumor tasks, with their features merged with the custom CNN to form a robust four-class classifier. On a Kaggle-derived dataset of 12,446 images (augmented to 27,035), the hybrid model achieves 100% testing accuracy, surpassing ResNet101 (≈99.81%) and a base CNN (≈77.39%), with distinct improvements in precision, recall, and F1-score. The results demonstrate the effectiveness of feature fusion and dual-branch design for accurate kidney disease diagnosis, offering potential for faster and more reliable clinical decision support.

Abstract

Medical image classification is a vital research area that utilizes advanced computational techniques to improve disease diagnosis and treatment planning. Deep learning models, especially Convolutional Neural Networks (CNNs), have transformed this field by providing automated and precise analysis of complex medical images. This study introduces a hybrid deep learning model that integrates a pre-trained ResNet101 with a custom CNN to classify kidney CT images into four categories: normal, stone, cyst, and tumor. The proposed model leverages feature fusion to enhance classification accuracy, achieving 99.73% training accuracy and 100% testing accuracy. Using a dataset of 12,446 CT images and advanced feature mapping techniques, the hybrid CNN model outperforms standalone ResNet101. This architecture delivers a robust and efficient solution for automated kidney disease diagnosis, providing improved precision, recall, and reduced testing time, making it highly suitable for clinical applications.

Hybrid Deep Learning Framework for Classification of Kidney CT Images: Diagnosis of Stones, Cysts, and Tumors

TL;DR

This work tackles four-class kidney CT image classification (normal, stone, cyst, tumor) and introduces a hybrid CNN that fuses a pre-trained ResNet101 with a custom CNN via a dual-branch architecture. The ResNet101 branches specialize in normal-vs-stone and cyst-vs-tumor tasks, with their features merged with the custom CNN to form a robust four-class classifier. On a Kaggle-derived dataset of 12,446 images (augmented to 27,035), the hybrid model achieves 100% testing accuracy, surpassing ResNet101 (≈99.81%) and a base CNN (≈77.39%), with distinct improvements in precision, recall, and F1-score. The results demonstrate the effectiveness of feature fusion and dual-branch design for accurate kidney disease diagnosis, offering potential for faster and more reliable clinical decision support.

Abstract

Medical image classification is a vital research area that utilizes advanced computational techniques to improve disease diagnosis and treatment planning. Deep learning models, especially Convolutional Neural Networks (CNNs), have transformed this field by providing automated and precise analysis of complex medical images. This study introduces a hybrid deep learning model that integrates a pre-trained ResNet101 with a custom CNN to classify kidney CT images into four categories: normal, stone, cyst, and tumor. The proposed model leverages feature fusion to enhance classification accuracy, achieving 99.73% training accuracy and 100% testing accuracy. Using a dataset of 12,446 CT images and advanced feature mapping techniques, the hybrid CNN model outperforms standalone ResNet101. This architecture delivers a robust and efficient solution for automated kidney disease diagnosis, providing improved precision, recall, and reduced testing time, making it highly suitable for clinical applications.

Paper Structure

This paper contains 21 sections, 6 figures, 6 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related work
  3. Motivation of study
  4. Research objectives
  5. Methodology
  6. Data description and pre-processing
  7. Data augmentation
  8. Architecture of proposed model: Hybrid CNN
  9. Architecture description
  10. Testing of the model
  11. Experimental setup
  12. Configuration
  13. Computational setup
  14. Evaluation metrics
  15. Results and discussion
  16. ...and 6 more sections

Figures (6)

  • Figure 1: An overview of the augmented data after rotation, flipping, and median filter.
  • Figure 2: Architecture of the proposed hybrid CNN model.
  • Figure 3: Schematic outline of hybrid CNN testing.
  • Figure 4: (left) Model accuracy trend. (right) Model loss trend.
  • Figure 5: Comparative analysis of confusion matrix. (left) ResNet101 and (right) Hybrid CNN.
  • ...and 1 more figures