Utilizing X-Ray Imaging Techniques to Determine the Density and Composition of Biological Materials

Two factors to assess the quality of food are composition and density, but both require destructive experimental techniques that are typically not performed during manufacturing processes. Porosity and the true density of a material are directly correlated, and can be difficult to estimate with current testing methodologies. Non-destructive x-ray imaging techniques have been the focus of recent research for the development of in-line quality experiments. X-ray powder diffraction (XRPD) and x-ray µ-computed tomography (µCT) were explored potential experimental techniques to determine the mass composition of a food mixture. Powdered materials that represent carbohydrates, salts and proteins were analyzed by the two imaging techniques to investigate proposed models for both methods. The objectives of this study were to (a) explore the possibility of the use of x-ray powder diffraction to determine the mass composition of an amorphous powder food mixture and (b) investigate current density and porosity estimation models for µCT imaging. XRPD was ineffective in distinguishing between broad diffraction peaks for unique amorphous materials through Rietveld refinement analysis, due to overlapping peak regions and machine noise. Powders that exhibited a crystallinity less than 30% could not be differentiated from the background noise of the XRPD, which renders the Rietveld analysis unusable for amorphous powder mixtures. X-ray µCT was utilized to scan both powdered and aqueous samples of the following materials: dextrose, α-lactose monohydrate, fructose, sucrose, KCl, NaCl, albumin, starch, gluten and casein. Standard mass attenuation coefficient values for water over the energy range 10 keV – 100 keV were utilized to define the Compton scattering coefficient and absorbed radiation energies at three possible energy levels within the µCT. Dual-energy density models and a composition model were investigated based on the mass attenuation coefficient. Two possible outcomes resulted from the models: density and volume fraction of pure components decreased as the mass fraction of salts increased, whereas the density and volume fraction estimations increased as the mass fraction increased in solution. When compared to literature value for the true density of salts, the average difference between the literature values and the extrapolated model values for pure components was Δp= 3.968 g/cm3. When compared to literature value for the true density of carbohydrates, the average difference between the literature values and the extrapolated model values for pure components was Δp=0.242 +/- 0.030 g/cm3. Changes in the energy range, and thus the constant representing the Compton scattering coefficient, rectified the negative density estimation of salt solutions, but did not improve the estimation of the true density of each pure component overall. Further investigation into the variable representing the Compton scattering coefficient is required to understand how it affects the density and volume fraction estimation of porous and aqueous samples. Future research should focus on image reconstruction and explore options to rectify the amplified beam hardening effects for aqueous solutions, and include the detection of coherent scattering effects. The established models show promise to allow for in-line quality analysis; however more research is required to understand the limitations of the µCT technology