Related Experiment Video
Updated: May 9, 2026
08:48
Optical Recording of Suprathreshold Neural Activity with Single-cell and Single-spike Resolution
Published on: September 5, 2012
11.9K
Improved noise reduction in photon-counting detector CT using prior knowledge-aware iterative denoising neural
Shaojie Chang1, Jeffrey F Marsh1, Emily K Koons1
1Mayo Clinic, Department of Radiology, Rochester, Minnesota, United States.
Journal of Medical Imaging (Bellingham, Wash.)
|May 27, 2024
Summary
A novel deep learning network, PKAID-Net, significantly reduces noise in photon-counting detector CT virtual monoenergetic images. This method preserves image quality and detail, enhancing diagnostic accuracy.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiology
Background:
- Photon-counting detector (PCD) CT enables high-resolution (HR) virtual monoenergetic images (VMIs).
- Image noise in VMIs can degrade diagnostic quality and accuracy.
- Developing effective denoising techniques is crucial for PCD-CT applications.
Purpose of the Study:
- To develop and evaluate a prior knowledge-aware iterative denoising neural network (PKAID-Net).
- To reduce image noise in HR VMIs from PCD-CT scans.
- To exploit unique noise characteristics of VMIs at different energy levels.
Main Methods:
- PKAID-Net utilizes a lower-noise VMI as a prior input.
- It iteratively refines the training dataset for improved denoising performance.
- The study included 10 coronary CT angiography exams reconstructed at various keV levels.
Main Results:
- PKAID-Net achieved 96% noise reduction compared to filtered back projection.
- A 65% relative noise reduction was observed compared to iterative reconstruction.
- Spatial and spectral fidelity, along with natural noise texture, were preserved.
Conclusions:
- PKAID-Net effectively reduces noise in HR VMIs from PCD-CT.
- The method maintains crucial spatial and spectral fidelity.
- Iterative data refinement enhances denoising robustness and minimizes spatial detail loss.
