Brown T R, Stoyanova R
Department of NMR and Medical Spectroscopy, Fox Chase Cancer Center, Philadelphia, Pennsylvania 19111, USA.
J Magn Reson B. 1996 Jul;112(1):32-43. doi: 10.1006/jmrb.1996.0106.
This paper extends the use of principal-component analysis in spectral quantification to the estimation of frequency and phase shifts in a single resonant peak across a series of spectra. The estimated parameters can be used to correct the spectra accordingly, resulting in more accurate peak-area estimation. Further, the removal of the variations in phase and frequency cause by instrumental and experimental fluctuations makes it possible to determine more accurately the remaining variations, which bear biological significance. The procedure is demonstrated on simulated data, a 3D chemical-shift-imaging dataset acquired from a cylinder of inorganic phosphate (Pi), and a set of 736 31P NMR in vivo spectra taken from a kinetic study of rate muscle energetics. In all cases, the procedure rapidly and automatically identifies the frequency and phase shifts present in the individual spectra. In the kinetic study, the procedure is used twice, first to adjust the phase and frequency of a reference peak (phosphocreatine) and then to determine the individual frequencies of the Pi peak in each of the spectra which further can be used for estimation of pH changes during the experiment.
本文将主成分分析在光谱定量中的应用扩展到对一系列光谱中单个共振峰的频率和相移进行估计。估计出的参数可据此对光谱进行校正,从而得到更准确的峰面积估计值。此外,消除由仪器和实验波动引起的相位和频率变化,使得更准确地确定具有生物学意义的剩余变化成为可能。该程序在模拟数据、从无机磷酸盐(Pi)圆柱体获取的三维化学位移成像数据集以及一组来自骨骼肌能量代谢动力学研究的736个体内31P NMR光谱上得到了验证。在所有情况下,该程序都能快速自动识别各个光谱中存在的频率和相移。在动力学研究中,该程序使用了两次,首先调整参考峰(磷酸肌酸)的相位和频率,然后确定每个光谱中Pi峰的各个频率,这些频率可进一步用于估计实验过程中的pH变化。