基于小波变换的核磁共振FID信号的去噪方法研究  

A New De-noising Method of NMR FID signals Based on Wavelet Transform

作　　者：王希武[1] 董光波[1] 谢桂海[1] 

机构地区：[1]军械工程学院计算机工程系,石家庄050003

出　　处：《核电子学与探测技术》2008年第2期365-370,381,共7页Nuclear Electronics & Detection Technology

基　　金：国家自然科学基金项目(60372042)的资助

摘　　要：以医用核磁共振自由感应衰减(FID)信号的去噪处理为研究目标,基于小波变换技术,分析了有效信号和噪声的小波变换特性,针对三种去噪方法——模极大值、阈值收缩和平移不变,提出了改进方案,并以阈值收缩去噪方法为例,对临床获取的神经胶质瘤病例的大量数据样本进行了去噪、分析实验,探讨了适用于本类FID信号去噪处理的小波基函数、分解参数、分解层次和阈值策略的选取方法,通过实验建立了最优选取方案。实验结果有效地验证了新方案良好的去噪性能。本研究对于医用磁共振数据的高效处理具有很好的参考价值。Taking aim at the de-noising of clinical NMR FID signals, the paper analyzes the wavelet transform characteristics of available signals and noises separately based on wavelet transform technology. With the principles of three de-noising methods wavelet-based, wavelet transform modulus maxima, threshold shrinkage, and translation invariance, Effective solutions for selecting method of wavelet basis function, optimized thresholding function and wavelet decomposing parameters are put forward, so as the distinguishing method between signals and noises. Taking brain gliomas as an example, the paper implements effective experimental solutions based on the wavelet transform de-noising technology. About 30 clinical NMR data obtained from the NMR devices have been de-noised to provide high quality data and validated the new method.

关 键 词：去噪 小波变换 核磁共振 FID信号 

分 类 号：TN911.7[电子电信—通信与信息系统]