Training of signal denoising model and signal denoising method, training of CT signal denoising model and CT signal denoising method
By training the input signal data at multiple scales, the signal denoising model is trained on a scale-by-scale encoding and decoding path, which solves the problem of insufficient signal preservation capability and improves the signal preservation capability of the signal denoising model.
CN120419982BActive Publication Date: 2026-07-24SHANGHAI UNITED IMAGING HEALTHCARE
View PDF 2 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHANGHAI UNITED IMAGING HEALTHCARE
- Filing Date
- 2025-04-03
- Publication Date
- 2026-07-24
AI Technical Summary
Technical Problem
Existing signal denoising models have poor signal preservation capabilities during training due to the single scale of input and output data.
Method used
Multi-scale training input signal data is used to train the signal denoising model through scale-wise encoding and decoding paths, including the connection of scale-wise encoding and decoding nodes. The network is trained using multi-scale training input signal data.
Benefits of technology
This improves the signal denoising model's ability to preserve the original signal information and enhances the signal denoising effect.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN120419982B_ABST
Abstract
The application relates to the technical field of signal processing, and provides a signal denoising model training method and a signal denoising method, and a CT signal denoising model training method and a CT signal denoising method, which can improve the signal denoising model's ability to retain signals. In the application, multi-scale training input signal data is acquired; a first network is acquired; the first network comprises a scale-by-scale encoding path and a scale-by-scale decoding path; each scale of the training input signal data is input into a corresponding scale of encoding nodes in the encoding path, and / or each scale of the training input signal data is input into a corresponding scale of decoding nodes in the decoding path, so as to train the first network and obtain a signal denoising model.
Need to check novelty before this filing date? Find Prior Art