Related Experiment Video
Updated: Mar 13, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.6K
Ensemble-based high-performance deep learning models for medical image retrieval in breast cancer detection
Aya E Fawzy1,2, Mohammed E Almandouh3,4, Mostafa Herajy5
1Information Technology Management Department, Faculty of Management Technology and Information Systems, Port Said University, Port Said, Egypt. aya_elsayed@himc.psu.edu.eg.
Scientific Reports
|March 12, 2026
Summary
This study introduces a deep learning model combining CNNs, RNNs, and XAI for Content-Based Medical Image Retrieval (CBMIR). The hybrid approach achieves 99.24% accuracy in classifying breast ultrasound images and enhances retrieval performance.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Science
Background:
- Digital imaging in healthcare generates vast amounts of data, posing challenges for efficient management and retrieval.
- Content-Based Medical Image Retrieval (CBMIR) systems are crucial for managing medical images but face limitations due to the semantic gap between image features and clinical meaning.
Purpose of the Study:
- To develop a novel deep learning approach for Content-Based Medical Image Retrieval (CBMIR) that addresses the semantic gap.
- To improve the accuracy and efficiency of medical image classification and retrieval in clinical settings.
Main Methods:
- A hybrid deep learning model integrating Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Explainable AI (XAI).
- Training and evaluation using the Breast Ultrasound Image (BUSI) dataset.
- The model performs image classification and retrieval based on predictive analysis.
Main Results:
- Achieved a high classification accuracy of 99.24% on the BUSI dataset.
- Demonstrated strong performance in medical image retrieval tasks, effectively bridging the semantic gap.
- The integration of XAI provides insights into the model's decision-making process.
Conclusions:
- The proposed hybrid deep learning model offers a significant advancement in Content-Based Medical Image Retrieval (CBMIR).
- This approach effectively enhances the classification and retrieval of medical images, particularly in breast ultrasound diagnostics.
- The combination of CNNs, RNNs, and XAI presents a promising direction for future medical imaging research and applications.
