Related Experiment Video
Updated: Jan 17, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.3K
Enhancing Diagnostic Precision: A Distribution-Based Compressed Denoising Scheme using Transfer Learning for Noise
Shtwai Alsubai1, Waseem Ahmad2, Mohd Anjum3
1Department of Computer Science, College of Computer Engineering and Sciences in Al-Kharj, Prince Sattam bin Abdulaziz University, Al-Kharj 11942, Saudi Arabia.
Current Medical Imaging
|September 19, 2025
Summary
This study introduces a Distribution-based Compressed Denoising Scheme (DCDS) using transfer learning for medical image denoising. DCDS enhances diagnostic accuracy and efficiency by effectively reducing noise in CT scans.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Image Processing
Background:
- Medical image noise impedes accurate diagnosis and treatment.
- Traditional denoising methods often require extensive data or sacrifice image detail.
- Developing effective denoising techniques is crucial for reliable medical image interpretation.
Purpose of the Study:
- To introduce a novel Distribution-based Compressed Denoising Scheme (DCDS) for enhanced medical image denoising.
- To leverage transfer learning for improved diagnostic precision and efficiency.
- To analyze pixel distributions for distinguishing normal and noisy pixels.
Main Methods:
- Proposed a Distribution-based Compressed Denoising Scheme (DCDS) utilizing transfer learning.
- Employed incremental and decremental distribution verification for pixel analysis.
- Trained the model on a chest CT scan dataset with Gaussian-like noise, using correlation mapping for variance estimation.
Main Results:
- DCDS demonstrated significant improvements in noise reduction, variance detection, and diagnostic precision.
- Achieved a peak signal-to-noise ratio improvement across evaluated ranges (24-32 dB) and over 82% noise reduction.
- Reduced mean error and analysis time, showcasing enhanced system efficiency and effective use of resources via transfer learning.
Conclusions:
- DCDS offers superior generalization compared to existing methods through statistical modeling and knowledge sharing.
- The framework is a promising, lightweight, adaptive, and accurate solution for medical image denoising.
- Further validation across diverse imaging modalities and noise types, alongside clinical integration, is recommended.