Microscopic FT-IR imaging-based meat and bone meal species discrimination using the screened matrix fragments-related spectral pixels and tableting pretreatment

This study proposes a novel microscopic FT-IR imaging-based analytical strategy for the discrimination of meat and bone meal (MBM) species. The feasibility of correlation coefficient and PLS-DA methods for the screening of bone and meat fragment-related spectral pixels in MBM microscopic FT-IR imaging was validated. Subsequently, PLS-DA models were established for MBM species-specific analysis based on the screened bone and meat fragment-related spectral pixels. Comparative analysis with the PLS-DA model using all FT-IR pixels demonstrated a significant improvement in the performance of the PLS-DA model based on the fused matrix fragment-related spectral pixels, achieving an accuracy of approximately 0.980. Furthermore, the proposed analytical strategy exhibited a time efficiency advantage with a consumption period of approximately 4.5 min, thereby ensuring high spectral technical efficiency. The findings of this study offer a theoretical basis for the spectral discrimination analysis of agricultural materials with complex matrix compositions.