Spectral fingerprinting and chemical mapping of plant-based meat analogues using portable hyperspectral imaging system and multivariate analysis

dc.contributor.authorDhanapal, Logesh
dc.contributor.examiningcommitteeBandara, Nandika (Food and Human Nutritional Sciences)
dc.contributor.examiningcommitteeMalalgoda, Maneka (Food and Human Nutritional Sciences)
dc.contributor.supervisorErkinbaev, Chyngyz
dc.date.accessioned2024-03-21T15:41:40Z
dc.date.available2024-03-21T15:41:40Z
dc.date.issued2024-03-20
dc.date.submitted2024-03-20T21:24:28Zen_US
dc.degree.disciplineBiosystems Engineeringen_US
dc.degree.levelMaster of Science (M.Sc.)
dc.description.abstractThe ultra-processed nature and multiple ingredients of plant-based meat analogues cause excessive-quality variation during storage suggesting the need for real-time quality monitoring and chemical mapping techniques. This research utilizes a non-destructive portable hyperspectral imaging technique in the visible and near-infrared ranges (400–1000 nm) to predict the quality (CIE- L*, a*, b*, moisture, texture) of plant-based meat burger patties followed by mapping of the quality distribution attributed to major ingredients and color variations over the storage period. Firstly, the study analyzed the effect of 10-30% textured vegetable protein (TVP), 5-25% pea protein (PP), and storage time (14 days) on quality using conventional quality assessment approaches, revealing the instability of overall quality even over a short storage period. Principal Components Analysis (PCA) applied to HSI data indicated that the primary variance stems from discoloration and moisture loss due to storage days, TVP, and PP. This variance is associated with the spectral variations at 500–650nm and 950–980 nm. Developed Partial Least Square Regression (PLSR) models possessed good prediction accuracies for a*, b*, moisture, and hardness. Pixel-by-pixel prediction maps highlighted the non-uniform component distribution. Secondly, the study examines the role of HSI in addressing the color instability of PBMB by visualizing color evolution during a 10-day storage period, influenced by TVP, PP, fava bean protein, and fat mimic. Simplified PLSR prediction models with improved accuracies were developed with interval-PLS (iPLS), recursive-PLS (rPLS), and genetic algorithm (GAs). Prediction maps showed the patterns of color deterioration. Findings highlighted that color evaluation of heterogenous plant-based food formulations using single-point colorimeters is insufficient to represent the whole sample. This research illustrated that spectral fingerprinting could visualize component distribution in complex food formulations, aiding in formulation optimization and understanding ingredient interactions. The findings are crucial for informed decisions on grading, labeling, and storage. Overall, VNIR-HSI in combination with multivariate analysis proves to be a valuable analytical tool for real-time quality inspection in plant-based food production, highlighting the potential of digital technologies in promoting sustainable foods.
dc.description.noteMay 2024
dc.description.sponsorshipUniversity of Manitoba Graduate Fellowship (UMGF) 2023-2024
dc.identifier.urihttp://hdl.handle.net/1993/38065
dc.language.isoeng
dc.subjectMeat alternatives
dc.subjectFood quality
dc.subjectNon-destructive food testing
dc.subjectSmart sensing
dc.subjectHyperspectral imaging
dc.subjectChemometrics
dc.subjectPlant proteins
dc.titleSpectral fingerprinting and chemical mapping of plant-based meat analogues using portable hyperspectral imaging system and multivariate analysis
local.subject.manitobano
oaire.awardNumberRGPIN-2021-03340
oaire.awardTitleDiscovery Grants Program - Individual
oaire.awardURIhttps://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=743380
project.funder.nameNatural Sciences and Engineering Research Council of Canada
Files
Original bundle
Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
Dhanapal_Logesh.pdf
Size:
4.13 MB
Format:
Adobe Portable Document Format
Description:
License bundle
Now showing 1 - 1 of 1
Loading...
Thumbnail Image
Name:
license.txt
Size:
770 B
Format:
Item-specific license agreed to upon submission
Description: