Inspired by human olfactory system: Deep-learning-assisted portable chemo-responsive dye-based odor sensor array for the rapid sensing of shrimp and fish freshness

Abstract

Colorimetric sensor arrays (CSAs) can detect the volatile organic compounds (VOCs) produced during food storage, enabling odor visualization for freshness monitoring. Herein, sixteen chemo-responsive dyes that undergo observable color changes to VOCs are absorbed on ammonium quaternized cellulose nanofibres (C-CNFs) to design a paper-based CSA. The strong ionic interactions between the dyes and C-CNFs immobilize the dyes onto the sensor without leakage and provide stable and consistent colorimetric change even at a high relative humidity (RH = 60 %). Owing to the three-dimensional porous structure of C-CNFs, the CSA demonstrates high sensitivity for detecting VOCs produced during shrimp and fish spoilage. The limits of detection for ammonia, dimethylamine, trimethylamine, cadaverine, and putrescine are 2, 21, 3, 1, and 1 ppm, respectively. By selecting dyes covering a wide range of VOCs, the color changes in the CSA can accurately differentiate fresh, less fresh, slightly spoilt, and spoilt shrimp and fish. These color changes are further digitized and used to train a convolutional neural network (CNN). The CNN method has high accuracy (99 %) for determining abovementioned freshness levels. The combination of CSA and CNN is a promising approach for fabricating a portable sensing platform for the advanced non-destructive determination of food freshness levels.