Quantification of plasma lipids and apolipoproteins by use of proton NMR spectroscopy, multivariate and neural network analysis

Citation
Tf. Bathen et al., Quantification of plasma lipids and apolipoproteins by use of proton NMR spectroscopy, multivariate and neural network analysis, NMR BIOMED, 13(5), 2000, pp. 271-288
Citations number
70
Language
INGLESE
art.tipo
Article
Categorie Soggetti
Medical Research Diagnosis & Treatment
Journal title
NMR IN BIOMEDICINE
ISSN journal
0952-3480 → ACNP
Volume
13
Issue
5
Year of publication
2000
Pages
271 - 288
Database
ISI
SICI code
0952-3480(200008)13:5<271:QOPLAA>2.0.ZU;2-#
Abstract
New approaches for quantification of human blood plasma lipids and apolipop roteins are presented. One method is based on multivariate analysis of prot on nuclear magnetic resonance spectra of human blood plasma. Although simil ar approaches have been developed previously, this is the first time princi pal component analysis (PCA) and partial least squares regression (PLS) hav e been applied to this particular task. Further, a large proportion of the subjects in this study were cancer patients undergoing treatment, which int roduced a new dimension to the quantification of lipoprotein distributions. Calibration models for prediction of lipids and apolipoproteins were const ructed by use of PLS, and blind samples were used to test the predictive ab ility. Comparison of the predicted vs observed data obtained by standard cl inical chemical procedures gave good agreement; the correlation coefficient for total plasma triglyceride was 0.99, for total plasma cholesterol 0.98, for LDL cholesterol 0.97, and for HDL cholesterol 0.88. These results are comparable with those obtained with other methods. The quantitative analysi s of 14 components (including total cholesterol and total triglyceride) of human blood plasma was also undertaken using various neural network (NN) an alyses of selected portions of the spectra. Conventional fully connected ba ckpropagation neural network topologies were capable of providing excellent predictions for the majority of the variables, confirming and reinforcing literature related to this approach. However HDL triglycerides were poorly predicted, while intermediate-quality results were obtained for the LDL cho lesterol, plasma apoA1 and LDL apoB variables. In these instances, applying significantly different neural network algorithms involving either general regression or polynomial neural networks in combination with genetic adapt ive components for parameter optimisation made improved predictions. Copyri ght (C) 2000 John Wiley & Sons, Ltd.