DETERMINATION OF MORPHOLOGICAL GROUP FROM INFRARED-SPECTRA OF CRUDE EXTRACTS OF MEDICINAL-PLANTS BY SIMCA PATTERN-RECOGNITION

This paper reports a new method of classification and identification o f medicinal plants by pattern recognition within the infrared spectra of crude extracts. The spectra of solvent extracts of 29 samples of Tr ipterygium wilfordii and T. hypoglaucum were examined for information concerning chemical classes. The Shannon information content for each wavelength channel was calculated for the absorption spectra, and the 9 wavelength channels with the highest information content were retain ed as a compressed basis set for SIMCA pattern recognition. The inhere nt class structure of the data showed two major groups, T. wilfordii r oot and T. wilfordii bark. Classification accuracy was 95.0% for the c lasses.