Research Article: Feed-forward neural network assisted by discriminant analysis for the spectroscopic discriminantion of cracked spores Ganoderma lucidum: A prospective biotechnology production tool

Date Published: November 15, 2011

Publisher: Springer

Author(s): Chee Wei Lim, Sheot Harn Chan, Angelo Visconti.


A major problem for manufacturers of cracked spores Ganoderma lucidum, a traditional functional food/Chinese medicine (TCM), is to ensure that raw materials are consistent as received from the producer. To address this, a feed-forward artificial neural network (ANN) method assisted by linear discriminant analysis (LDA) and principal component analysis (PCA) was developed for the spectroscopic discrimination of cracked spores of Ganoderma lucidum from uncracked spores. 120 samples comprising cracked spores, uncracked spores and concentrate of Ganoderma lucidum were analyzed. Differences in the absorption spectra located at ν1 (1143 – 1037 cm-1), ν2 (1660 – 1560 cm-1), ν3 (1745 – 1716 cm-1) and ν4 (2845 – 2798 cm-1) were identified by applying fourier transform infra-red (FTIR) spectroscopy and used as variables for discriminant analysis. The utilization of spectra frequencies offered maximum chemical information provided by the absorption spectra. Uncracked spores gave rise to characteristic spectrum that permitted discrimination from its cracked physical state. Parallel application of variables derived from unsupervised LDA/PCA provided useful (feed-forward) information to achieve 100% classification integrity objective in ANN. 100% model validation was obtained by utilizing 30 independent samples. ν1 was used to construct the matrix-matched calibration curve (n = 10) based on 4 levels of concentration (20%, 40%, 60% and 80% uncracked spores in cracked spores). A coefficient of correlation (r) of 0.97 was obtained. Relative standard deviation (RSD) of 11% was achieved using 100% uncracked spores (n = 30). These results demonstrate the feasibility of utilizing a combination of spectroscopy and prospective statistical tools to perform non destructive food quality assessment in a high throughput environment.

Partial Text

Ganoderma lucidum, a fungus commonly known as Lingzhi, has been used as a traditional functional food/medicine for centuries by rulers of the Chinese and Japanese dynasties to achieve enhanced vitality and longevity. These formulae take on exotic forms of special tea and mushroom concoction suitable for daily intake as supplements. Owing to the perceived benefits of these highly desirable medicinal properties localized within the spores (Lin 2001) and further amplified by profit focused producers, commercial demand for Ganoderma lucidum outstripped its natural occurrence in nature. Consequently, cultivation techniques were developed to cater for mass production. Some channels used to perform cultivation include horizontal stirred tank reactor and solid state fermentation (Habijanic and Berovic, 2000; Yang et al. 2003; Hsieh and Yang 2004). Both types of cultivation strategies have been reported to yield reasonable fruit bodies suitable for general use. A major problem for manufacturers of cracked spores Ganoderma lucidum therefore, is to ensure the raw materials supplied are consistent (Li et al. 2011). According to Recital 11 of the European Union Regulation on the hygience of foodstuffs No 852/2004, the application of hazard analysis and critical control point (HACCP) principles to primary produce is not yet generally feasible (Cerf and Donnat, 2011). By this same principle, rapid methods (practicable in a factory environment) are therefore required to test materials prior to its conversion into the finished product.

While it is possible to improve the first PC score further by reducing the variables (frequency bands), such approach raised some concerns within the framework of spectroscopy. Indeed, while PCA and LDA are useful tools suitably used to extract features that are focused on discriminating between classes via dimension reduction strategy, the error increment due to dimension reduction has to be without sacrificing the discriminative power of classifiers (Benediktsson and Sveinsson 1997). In this work, we did not observe such limitation. Rather, by shrinking the variables pool further, the advantage of utilizing the maximum chemical information provided by the absorption spectra will not be fully tapped. For this purpose, the values obtained for the PCs and canonical functions (LDA) were fed into an ANN model using 4 hidden nodes (33% random data holdback) to ascertain the classification outcome obtained when PCA and LDA were applied.

The authors declare that they have no competing interests.




Leave a Reply

Your email address will not be published.