Investigating the applicability of a snapshot Computed Tomography Imaging Spectrometer for the prediction of °Brix and pH of grapes

This study compares the performance of a newly developed snapshot hyperspectral imaging (HSI) system based on Computed Tomography Imaging Spectroscopy (CTIS) and a state-of-the-art line scan HSI system for determining soluble solids content (°Brix) and pH values in four table grape cultivars (Sheegene 20, Sugrathirteen, Grapaes, and Sweet Flavor). A total of 318 grapes were imaged and predictions were validated against reference measurements obtained with a refractometer (°Brix) and a pH meter (pH). For the line scan system, Partial Least Squares Regression (PLSR) models based on the full spectral range demonstrated excellent performance with RMSEP Brix=0.63, RMSEP pH=0.09, R Brix 2=0.94, and R pH 2=0.79. Reducing the spectral range led to a decline in performance, with RMSEP Brix=1.29 and RMSEP pH=0.12, and R Brix 2=0.76 and R pH 2=0.69. The CTIS system, despite its narrower spectral range, achieved comparable performance with U-Net reconstructions (RMSEP Brix=1.35, RMSEP pH=0.012, R Brix 2=0.74, R pH 2=0.70). The CTIS system's key advantages—such as its lower cost, portability, and ability to capture snapshot hyperspectral images—make it a viable option for in-field applications in grape quality assessment. This study highlights the potential for CTIS-based systems to complement traditional line scan systems, particularly in real-world, out-of-laboratory settings.