Hyperspectral image recovery using a color camera for detecting colonies of foodborne pathogens on agar plate

Authors: Yoon, Seung-Chul; Shin, Tae-Sung; Heitschmidt, Gerald - Jerry; Lawrence, Kurt; Park, Bosoon; Gamble, Gary

Submitted to: Journal of Biosystems Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 8/1/2019
Publication Date: 9/27/2019
Citation: Yoon, S.C., Shin, T., Heitschmidt, G.W., Lawrence, K.C., Park, B., Gamble, G.R. 2019. Hyperspectral image recovery using a color camera for detecting colonies of foodborne pathogens on agar plate. Journal of Biosystems Engineering. https://doi.org/10.1007/s42853-019-00024-y.
DOI: https://doi.org/10.1007/s42853-019-00024-y

Interpretive Summary: Screening of foodborne pathogens using conventional agar plates is a laborious and time-consuming task. Previously, engineers at the Agricultural Research Service (ARS) of the U.S. Department of Agriculture (USDA) developed a hyperspectral imaging technique to rapidly detect and identify pathogens in agar plates. In reality, however, high cost and uncertain demand for new technology such as hyperspectral imaging have significantly slowed down the fast deployment of hyperspectral imaging as a daily routine screening tool in microbiology laboratories in regulatory agencies and companies worldwide. Therefore, there was an immediate research need to find a compact and cost-effective alternative to the bulky and expensive hyperspectral imaging technology. In this study, the ARS engineers developed a hyperspectral image recovery technique using a consumer-grade digital color camera for detecting colonies of non-O157 E. coli pathogens on agar media. Hyperspectral images were mathematically recovered in the visible and near-infrared spectral range between 400 and 1,000 nm only with color images and spectral recovery models. The classification accuracy of the developed color imaging technology (spectral recovery methods) was up to about 90percent, compared to 95percent-99percent accuracy of hyperspectral imaging technology. These results suggest that a color camera is potentially feasible as a cost-effective and rapid imaging tool to predict hyperspectral images and classify the test pathogen samples. However, there are still challenges ahead, such as improving robustness and accuracy of spectral recovery, before this technology becomes commercially viable.

Technical Abstract: In this paper, a hyperspectral image recovery technique using a cost-effective RGB color camera and least-squares regression is presented to reconstruct hyperspectral images from color images and detect colonies of foodborne pathogens on agar plates. The target bacteria were the six representative non-O157 Shiga-toxin producing Escherichia coli (STEC) serogroups (O26, O45, O103, O111, O121, and O145). The spectral range was from 400 to 1,000 nm. Unlike many other studies using color charts with known and noise-free spectra for training their spectral recovery models, we directly measured the hyperspectral and color images of real scenes for training the spectral recovery models. Both hyperspectral and color images were calibrated to relative reflectance and spatially registered. Two spectral recovery models (polynomial multivariate linear regression, PMLR and partial least squares regression, PLSR) were evaluated by cross-validation. The results showed that PLSR was more effective with higher order polynomial regressions than PMLR. The classification accuracy measured with an independent test set was about 90%. The results suggested the potential of a cost-effective color imaging system using hyperspectral image classification algorithms for rapidly differentiating pathogens in agar plates.