Abstract:

A new method for rapid detection of aerobic plate count (APC) in Parabramis pekinensis as a freshwater fish
was developed based on visualized sensor array technology. Information about the odor of P. pekinensis was extracted using
visualized sensor arrays and odor characteristics were obtained as the color difference of images of sensor arrays before
and after exposure to the headspace gas of fish samples using image processing technique. Meanwhile, the grades of fish
freshness were evaluated through measuring APC by plate counting method. A partial least squares (PLS) model and a
synergy interval partial least square (siPLS) model were set up based on the data relationship between odor characteristics
from sensor arrays and APC values. The siPLS model gave better results with correlation coefficient of 88.96% for
calibration set and 83.29% for prediction set. The results of this study indicate that visual sensor arrays can be used to predict
APC as a convenient, fast, low-cost and non-destructive method

Key words: Parabramis pekinensi, aerobic plate count (APC), visual sensor technique, partial least squares (PLS), synergy interval partial least square (siPLS)