一种基于图像域和张量域的双域深度学习扩散张量成像去噪方法
By designing the DuTD network and combining DDNet and TDNet subnetworks, dual-domain deep learning diffusion tensor imaging denoising in both the image and tensor domains was achieved, solving the problem of high-definition diffusion image reconstruction, improving the denoising effect, and validating its clinical applicability.
CN118570088BActive Publication Date: 2026-07-17XIAMEN UNIV
View PDF 2 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XIAMEN UNIV
- Filing Date
- 2024-05-17
- Publication Date
- 2026-07-17
AI Technical Summary
Technical Problem
Existing diffusion reconstruction and quantification methods struggle to reconstruct high-resolution diffusion images down to the submillimeter level, and lack methods for joint reconstruction and quantification using images from multiple diffusion directions.
Method used
The DuTD network is designed, which includes two sub-networks, DDNet and TDNet, for denoising in the image domain and tensor domain, respectively. The final denoised DWI image is obtained by weighted averaging.
Benefits of technology
It improved the denoising effect, especially in clinically acceptable 6-direction DWI data, demonstrating excellent denoising performance and validating its clinical applicability.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN118570088B_ABST
Abstract
一种基于图像域和张量域的双域深度学习扩散张量成像去噪方法。本方法包含对DWI图像和张量元素图像分别去噪的网络DDNet和TDNet。DuTD去噪方法包括以下步骤：1)设计DDNet网络与TDNet网络分别完成对带噪声的DWI图像和扩散张量的去噪；2)去噪后的扩散张量通过生物物理模型还原到图像域；3)从去噪后的张量中还原的DWI图像与直接去噪的DWI图像通过加权平均得到最终去噪的DWI图像。针对采集时间临床可接受的6方向DWI数据，提出图像域与参数域双域深度学习去噪方法，并将该方法应用于自采的帕金森病人数据，去噪效果优秀，并且验证其临床适用性。
Need to check novelty before this filing date? Find Prior Art