Related Experiment Video
Updated: May 4, 2026
07:15
Transient Optical Clearing Using Absorbing Molecules for Ex Vivo and In Vivo Imaging
Published on: July 11, 2025
3.8K
Adaptive nonlocal means filtering based on local noise level for CT denoising
Zhoubo Li1, Lifeng Yu2, Joshua D Trzasko1
1Department of Physiology and Biomedical Engineering, Mayo Clinic, Rochester, Minnesota 55905.
Medical Physics
|January 7, 2014
Summary
This study introduces an adaptive nonlocal means (NLM) algorithm for computed tomography (CT) image denoising that adjusts to local noise levels. This method effectively reduces noise while preserving image quality and can be integrated into clinical workflows, potentially lowering radiation doses.
Area of Science:
- Medical Imaging
- Image Processing
- Computational Science
Background:
- Computed tomography (CT) imaging is crucial for medical diagnosis.
- Image noise is an inherent challenge in CT, potentially affecting diagnostic accuracy.
- Existing denoising methods may compromise image quality or are not clinically practical.
Purpose of the Study:
- To develop and evaluate an adaptive image-domain noise reduction method for CT images.
- The method is based on a modified nonlocal means (NLM) algorithm.
- The goal is to adapt denoising to local noise levels and ensure clinical workflow compatibility.
Main Methods:
- Developed a computationally efficient technique for local noise estimation directly from CT images.
- Modified a 3D NLM algorithm to adapt denoising strength based on an analytically derived noise map.
- Utilized graphical processing units (GPU) for efficient implementation and validated performance using phantom and clinical data.
Main Results:
- The adaptive NLM filter effectively reduced noise without degrading high-contrast spatial resolution.
- Noise power spectrum analysis showed better preservation of noise characteristics compared to commercial filters.
- Both subjective and objective evaluations demonstrated improved low-contrast performance.
- GPU implementation processed 300 slice images within 5 minutes.
- Clinical pilot studies led to permanent implementation in radiology practices.
Conclusions:
- A computationally efficient adaptive NLM filtering technique for CT image denoising was developed and validated.
- The method allows for noise map calculation and filtering within clinically acceptable timeframes.
- The adaptive NLM algorithm effectively denoises CT volumes and may enable significant radiation dose reduction.
