Non-invasive measurement of shear force in chicken meat using near infrared spectroscopy supported by neural network analysis

Abstract

The aim of the present work was to evaluate the ability of a portable near-infrared (NIR) spectroscopy integrated with machine learning methods to predict the shear force in chicken meat. Considering the benefits of dimension reduction from Principal Component Regression (PCR) and the ability to handle non-linearity from Artificial Neural Network (ANN), these two algorithms were combined. Through the augmentation, the Principal Component Neural Network (PCNN) is developed. The results show that PCNN successfully surpassed the respective versions of PCR and ANN with higher shear force prediction performances. The PCNN proved to achieve the best prediction in breast meat with root mean square error of prediction (RMSEP) of 0.0815 kg and coefficient of determination, (Rp²⁾ of 0.7977. NIRS technology integrated with machine learning yield a promising non-invasive technique in predicting the shear force of intact raw chicken meat.