摘要
针对现有二维反演算法对噪声敏感或过度依赖于对噪声的估计的问题,提出了一种基于噪声拟合的核磁共振二维谱反演新方法。首先,使用小波滤波进行噪声提取;然后将噪声分布按照高斯白噪声的数学模型进行拟合,获取噪声方差;最后依据制定的新的停机准则和进化算法进行拟合,拟合残差与估计的噪声水平相当或最接近时,迭代停止并得到二维谱。通过使用该算法对不同信噪比的仿真数据和多组实验采集数据进行处理后发现,该算法具有很高的鲁棒性和准确性。这种新的反演算法能够满足不同应用领域、不同信噪比数据的反演要求,具有很高的实际应用价值。
Two-dimensional NMR inversion algorithms have been intensively studied all over the world. Aiming at the problem that existing two-dimensional NMR inversion algorithms are either too sensitive to noise or too dependent to the noise estimation, a new noise-fitting based 2D NMR inversion algorithm is proposed. Firstly, wavelet-based filtering is used to extract noise;then the noise distribution is fitted with a white Gaussian noise mathematic model to get the noise variance;and lastly the original data are fitted according to the new defined stop criteria and evolution algo- rithm. When the fitted residual is comparable or close to the estimated noise level,the iterations stop and the 2D spectrum is obtained. The proposed algorithm was used on the simulation data with different SNRs and multiple groups of experiment acquired data, and data processing was conducted. The results demonstrate that the proposed al- gorithm has high robustness and accuracy, can meet the inversion requirements of the data with different SNRs in different application fields, and possesses good actual application values.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2013年第7期1640-1649,共10页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(60972122)
上海市重点学科建设项目(P0502)资助
关键词
核磁共振
二维反演
扩散-横向弛豫
纵向-横向弛豫
nuclear magnetic resonance
2D inversion
diffusion coefficient and transverse relaxation time （ D-T2 ）
longitudinal and transverse relaxation time （T1-T2 ）
