Evaluation of growth characteristics of Aspergillus parasiticus by shortwave infrared (SWIR) hyperspectral imaging

Abstract. This work focuses on the evaluation of growth characteristics of Aspergillus parasiticus inoculated on rose bengal agar by shortwave infrared (SWIR) hyperspectral imaging. The Aspergillus parasiticus colonies were grown for 6 days and their SWIR hyperspectral images (1000-2500 nm) were acquired at regular intervals (24h) after inoculation. Firstly, the average spectra of each colony were extracted and divided into calibration samples and validation samples according to the ratio of 3:1 randomly. A classification model was built to identify the colonies with different growth period. Results showed the classification accuracy of calibration and validation set were 97.00% and 94.0%. Secondly, principal component analysis (PCA) was applied on the hyperspectral image data. A score image was obtained based on the first principal component (PC). Based on the hyperspectral image and the score image, distinctive radial growth rings can be identified. The optical characteristics of the colonies within different growth stages including the lag phase, exponential and stationary phase can be indicated. Finally, three characteristic wavelengths i.e. 1442 nm and 1929 nm used to classify younger or older mycelia were selected. To sum up, SWIR hyperspectral imaging is a powerful tool to evaluate the growth characteristics of Aspergillus parasiticus.