Determination of Some Chemical Components of Fermented Cordyceps Mycelia Using Near-Infrared Spectroscopy

The study evaluated the potential of using near-infrared spectroscopy (NIRS) to predict adenosine, moisture and total amino acid content of fermented mycelia of Paecilomyces hepialid, a derivative of Cordyceps sinensis. A total of 200 samples were collected and randomly divided for calibration (n=140) and prediction (n=60). Spectra were generated by a multi-purpose analyser (MPA (TM)) Fourier transform NIRS system with recorded spectral range of 12500-4000 cm(-1). Several spectral pretreatment techniques were adopted for spectral transformation. The transformed spectra in calibration set were subjected to a partial least squares (PLS) algorithm with leave-one-out cross-validation. Prediction results showed that the PLS models developed for the orthogonal signal correction (OSC)-transformed spectra achieved best performance with coefficient of determination (R-2) of 0.996, 0.996 and 0.990 for adenosine, moisture and total amino acid contents, respectively. This experiment suggests that the NIR spectroscopy, if coupled with appropriate spectrum transformation, is a promising tool for rapid, nondestructive and accurate measurement of chemical components of natural and fermented mycelia of Cordyceps sinensis.