| Nuclear Magnetic Resonance(NMR),owing to its rapid,non-destructive,accurate,repeatable and other superiorities,has been successfully and widely applied as an analytical means for quantitative and qualitative analysis to multiple research fields from pharmacy,bromatology,biology to chemistry,as well as metabonomics.Baseline distortion and peak overlapping in NMR data complicates the distinction and quantification of metabolites,which makes analysis difficult.The commonly used baseline correction methods usually require considerable time for manual parameter and filter tuning by experienced users,meanwhile,the effectiveness needs to be improved in regions containing wide peaks and overlapping peaks.Optimizing baseline correction method will be of great help to NMR analysis.As for peak overlapping,which is another frequent phenomenon affecting the analysis,global deconvolution algorithms have been proved to be a good solution.We worked on the automatic baseline correction method and global deconvolution algorithms of NMR spectral and present the achievements as follows:We proposed a baseline correction method by mathematical modeling on radius tangent.By calculating the tangent point between artificially constructed sliding arcs and spectrum with baseline distortion,the baseline points are found and used to establish the baseline model.Simulated 1D as well as experimental 1D and 2D data sets are used to verify the new method respectively.Compared with current methods,the new method has the following improvements:better correction results in regions containing wide peaks and overlapping peaks;fewer parameters and more automated setup;faster calculation process owing to no generation required;and effective for 2D spectra.We present here a global deconvolution algorithm based on simulated annealing(SA).The sample spectra decomposes into standard spectra belongs to metabolites in the sample by calculating the optimal fitting solution.The calculation is conducted by a algorithm based on SA.The new algorithm comprehensively utilizes the full spectral information and takes peak drift into consideration.Experiments show that this algorithm performs well in both accuracy and speed. |
