de Certaines J D, Nadal L, Leray G, Serrai H, Lewa C J
Laboratoire de Résonance Magnétique en Biologie et Médecine, Faculté de Médecine, Rennes, France.
Anticancer Res. 1996 May-Jun;16(3B):1451-60.
This paper reviews several methods presently available for analysing lipoprotein NMR spectra. Two main steps can be distinguished: NMR signal processing and data analysis. Time domain (wavelet transform) and frequency domain (curve fitting) signal processing methods are compared. Statistical methods of data analysis (Ascending Hierarchical Classification, Correspondence Analysis and Principal Component Analysis) have been tested on simulated NMR data of plasma lipoprotein with different numbers of sampling points and different noise levels. These few examples clearly attest that the NMR approach to complex "mixture" (such as body fluids) analysis is emerging from its infancy. New interest in plasma lipoprotein analysis in cancer biology is finally discussed in the light of previous clinical and experimental results and of understanding of lipid metabolism in cancer.
本文综述了目前可用于分析脂蛋白核磁共振光谱的几种方法。可区分出两个主要步骤：核磁共振信号处理和数据分析。对时域（小波变换）和频域（曲线拟合）信号处理方法进行了比较。已对具有不同采样点数和不同噪声水平的血浆脂蛋白模拟核磁共振数据测试了数据分析的统计方法（升序层次分类、对应分析和主成分分析）。这几个例子清楚地证明，用于复杂“混合物”（如体液）分析的核磁共振方法正从起步阶段发展起来。最后，根据先前的临床和实验结果以及对癌症脂质代谢的理解，讨论了癌症生物学中对血浆脂蛋白分析的新兴趣。
