Nondestructive prediction of textural properties of model plant-based meat analogs prepared with sesame seed meal protein isolate using hyperspectral imaging

Abstract Texture, a key quality attribute in plant-based meat analogs (PMAs), is conventionally evaluated using destructive methods. This study evaluated the use of hyperspectral imaging (HSI) coupled with partial least squares regression (PLSR) for the nondestructive prediction of six textural attributes in model plant-based meat analogs formulated with sesame seed meal protein isolate (SMPI). Samples (n = 80) covering a broad textural range were imaged at 900–1700 nm, and their spectra were used to develop PLSR models. Models for hardness (R 2 CV = 0.882, RPD = 2.92), resilience (0.865, 2.74), and chewiness (0.829, 2.44) showed adequate predictive performance. Cohesiveness and adhesiveness exhibited limited performance, while no valid model was obtained for springiness. Variable importance in projection analysis identified informative wavelength regions below 1200 nm. Collectively, HSI–PLSR showed potential for the nondestructive evaluation of selected textural attributes in PMAs.