Microscopic Hyperspectral Imaging Research Articles

Effects of phantom microstructure on their optical properties.

Developing stable, robust, and affordable tissue-mimicking phantoms is a prerequisite for any new clinical application within biomedical optics. To this end, a thorough understanding of the phantom structure and optical properties is paramount. We characterized the structural and optical properties of PlatSil SiliGlass phantoms using experimental and numerical approaches to examine the effects of phantom microstructure on their overall optical properties. We employed scanning electron microscope (SEM), hyperspectral imaging (HSI), and spectroscopy in combination with Mie theory modeling and inverse Monte Carlo to investigate the relationship between phantom constituent and overall phantom optical properties. SEM revealed that microspheres had a broad range of sizes with average and were also aggregated, which may affect overall optical properties and warrants careful preparation to minimize these effects. Spectroscopy was used to measure pigment and SiliGlass absorption coefficient in the VIS-NIR range. Size distribution was used to calculate scattering coefficients and observe the impact of phantom microstructure on scattering properties. The results were surmised in an inverse problem solution that enabled absolute determination of component volume fractions that agree with values obtained during preparation and explained experimentally observed spectral features. HSI microscopy revealed pronounced single-scattering effects that agree with single-scattering events. We show that knowledge of phantom microstructure enables absolute measurements of phantom constitution without prior calibration. Further, we show a connection across different length scales where knowledge of precise phantom component constitution can help understand macroscopically observable optical properties.

Squeezing the Threshold of Metal‐Halide Perovskite Micro‐Crystal Lasers Grown by Solution Epitaxy

AbstractMetal halide perovskite semiconductors have demonstrated remarkable progress not only in photovoltaics and X‐ray detection but also in laser technologies. Particularly appealing is the simplicity with which micro‐crystallites can be epitaxially grown, thereby forming micro‐resonators suitable for lasing. Here, the laser threshold is optimized by selecting excitation laser parameters and by improving material quality. The latter process is conducted for formamidinium lead tribromide, emitting in the green spectral range. Depending on the growth method and parameters, the sizes of the micro‐resonators can be tuned between ≈3 and 23 micrometers. Under laser excitation systematically lower thresholds are observed for micro‐resonators in the 4–7 micrometer size range, than for larger ones, irrespective of growth method. Among three optimized growth methods, epitaxial growth via antisolvent vapor‐assisted crystallization exhibits the smallest threshold powers, indicating the highest material quality. This conclusion is supported by hyperspectral microscopic luminescence imaging and by transient photoluminescence. The best laser structures exhibit record threshold powers for epitaxially grown perovskites, indicating that the selected antisolvent vapor epitaxial growth holds great promises also for other perovskite materials.

Classification of Microscopic Hyperspectral Images of Blood Cells Based on Lightweight Convolutional Neural Network

Hyperspectral imaging has emerged as a novel imaging modality in the medical field, offering the ability to acquire images of biological tissues while simultaneously providing biochemical insights for in-depth tissue analysis. This approach facilitates early disease diagnosis, presenting advantages over traditional medical imaging techniques. Addressing challenges such as the computational burden of existing convolutional neural networks (CNNs) and imbalances in sample data, this paper introduces a lightweight GhostMRNet for the classification of microscopic hyperspectral images of human blood cells. The proposed model employs Ghost Modules to replace conventional convolutional layers and a cascading approach with small convolutional kernels for multiscale feature extraction, aiming to enhance feature extraction capabilities while reducing computational complexity. Additionally, an SE (Squeeze-and-Excitation) module is introduced to selectively allocate weights to features in each channel, emphasizing informative features and efficiently achieving spatial–spectral feature extraction in microscopic hyperspectral imaging. We evaluated the performance of the proposed GhostMRNet and compared it with other state-of-the-art models using two real medical hyperspectral image datasets. The experimental results demonstrate that GhostMRNet exhibits a superior performance, with an overall accuracy (OA), average accuracy (AA), and Kappa coefficient reaching 99.965%, 99.565%, and 0.9925, respectively. In conclusion, the proposed GhostMRNet achieves a superior classification performance at a smaller computational cost, thereby providing a novel approach for blood cell detection.

Quantitative prediction and visualization of matcha color physicochemical indicators using hyperspectral microscope imaging technology

Color physicochemical indicators of matcha are related to sensory quality, and the distributions of color physicochemical indicators can reflect the uniformity of matcha powder. This study aimed to explore the feasibility of using hyperspectral microscope imaging (HMI) technology to predict and visualize matcha color physicochemical indicators. The average spectra (400–998 nm) were extracted from the regions of interest (ROI). Subsequently, competitive adaptive reweighted sampling (CARS), interval random frog (iRF), successive projections algorithm (SPA), and the combination of iRF and SPA (iRF-SPA) were adopted to screen the characteristic spectra variables that were associated with specific color physicochemical indicators. Partial least squares (PLS) regression was applied to develop the calibration models for predicting the color physicochemical indicators. The results showed that the iRF-SPA-PLS models obtained the optimal prediction results, with predictive correlation coefficients (Rp) of 0.9262 for L*, 0.8826 for a*, 0.8583 for b*, 0.8243 for chlorophyll a, 0.7518 for chlorophyll b, and 0.8093 for chlorophyll total. Finally, the distribution maps of matcha color physicochemical indicators were visualized using these optimized models. The results suggested that the HMI technology coupled with chemometrics has enormous potential to predict and visualize the color physicochemical indicators of matcha, which could reflect the uniformity of matcha powder to ensure the stability of matcha quality. The approach presented in this study can also be applied to quality control of other powdery foods.

Development of a large volume line scanning, high spectral range and resolution 3D hyperspectral photoluminescence imaging microscope for diamond and other high refractive index materials

Hyperspectral photoluminescence (PL) imaging is a powerful technique that can be used to understand the spatial distribution of emitting species in many materials. Volumetric hyperspectral imaging of weakly emitting color centers often necessitates considerable data collection times when using commercial systems. We report the development of a line-scanning hyperspectral imaging microscope capable of measuring the luminescence emission spectra for diamond volumes up to 2.20 × 30.00 × 6.30 mm with a high lateral spatial resolution of 1–3 µm. In an single X-λ measurement, spectra covering a 711 nm range, in a band from 400–1100 nm, with a spectral resolution up to 0.25 nm can be acquired. Data sets can be acquired with 723 (X) × 643 (Y) × 1172 (λ) pixels at a rate of 6 minutes/planar image slice, allowing for volumetric hyperspectral imaging with high sampling. This instrument demonstrates the ability to detect emission from several different color centers in diamond both at the surface and internally, providing a non-destructive method to probe their 3D spatial distribution, and is currently not achievable with any other commonly used system or technique.

An Ensemble Learning Method for Detection of Head and Neck Squamous Cell Carcinoma Using Polarized Hyperspectral Microscopic Imaging.

Head and neck squamous cell carcinoma (HNSCC) has a high mortality rate. In this study, we developed a Stokes-vector-derived polarized hyperspectral imaging (PHSI) system for H&E-stained pathological slides with HNSCC and built a dataset to develop a deep learning classification method based on convolutional neural networks (CNN). We use our polarized hyperspectral microscope to collect the four Stokes parameter hypercubes (S0, S1, S2, and S3) from 56 patients and synthesize pseudo-RGB images using a transformation function that approximates the human eye's spectral response to visual stimuli. Each image is divided into patches. Data augmentation is applied using rotations and flipping. We create a four-branch model architecture where each branch is trained on one Stokes parameter individually, then we freeze the branches and fine-tune the top layers of our model to generate final predictions. Our results show high accuracy, sensitivity, and specificity, indicating that our model performed well on our dataset. Future works can improve upon these results by training on more varied data, classifying tumors based on their grade, and introducing more recent architectural techniques.

3D-GhostNet: A novel spatial-spectral algorithm to improve foodborne bacteria classification coupled with hyperspectral microscopic imaging technology

A novel 3D-GhostNet has been developed for hyperspectral microscopic imaging (HMI) data analysis to improve classification of foodborne pathogens. The HMI, leveraging microscopic technology, surpasses traditional spectral imaging in terms of sensitivity and resolution by utilizing visible/near-infrared spectroscopy integrated with high-resolution single-cell images. The newly constructed 3D-Ghost network directly processes and recognizes single-cell hypercube, enables HMI technology to become near-real-time detection by combining with automated data processing pipeline. 3D-GhostNet incorporates Ghost modules as its backbone and adds convolutional block attention module (CBAM) for adaptively extracting high-dimensional depth spatial-spectral information. In the task of identifying four different foodborne bacteria cells, our 3D-GhostNet achieved 100% accuracy, surpassing spectral-only-based (SOB) classifiers such as linear discriminant analysis (LDA, 89.7%), support vector machine (SVM, 93.9%), 1D convolutional neural network (1D-CNN, 98%), and long short-term memory (LSTM, 98.5%). Furthermore, our independent blind test results indicate that 3D-GhostNet remains robust for identifying complex mixtures of pathogens. As a label-free detection tool, 3D-GhostNet-assisted HMI technology is fully competent in classification of different foodborne pathogens, suggesting broad applications in food safety and quality.

SSTU: Swin-Spectral Transformer U-Net for hyperspectral whole slide image reconstruction

Whole Slide Imaging and Hyperspectral Microscopic Imaging provide great quality data with high spatial and spectral resolution for histopathology. Existing Hyperspectral Whole Slide Imaging systems combine the advantages of the techniques above, thus providing rich information for pathological diagnosis. However, it cannot avoid the problems of slow acquisition speed and mass data storage demand. Inspired by the spectral reconstruction task in computer vision and remote sensing, the Swin-Spectral Transformer U-Net (SSTU) has been developed to reconstruct Hyperspectral Whole Slide images (HWSis) from multiple Hyperspectral Microscopic images (HMis) of small Field of View and Whole Slide images (WSis). The Swin-Spectral Transformer (SST) module in SSTU takes full advantage of Transformer in extracting global attention. Firstly, Swin Transformer is exploited in space domain, which overcomes the high computation cost in Vision Transformer structures, while it maintains the spatial features extracted from WSis. Furthermore, Spectral Transformer is exploited to collect the long-range spectral features in HMis. Combined with the multi-scale encoder-bottleneck-decoder structure of U-Net, SSTU network is formed by sequential and symmetric residual connections of SSTs, which reconstructs a selected area of HWSi from coarse to fine. Qualitative and quantitative experiments prove the performance of SSTU in HWSi reconstruction task superior to other state-of-the-art spectral reconstruction methods.

Polarized Hyperspectral Microscopic Imaging for Zebrafish.

Zebrafish is a well-established animal model for developmental and disease studies. Its optical transparency at early developmental stages is ideal for tissue visualization. Interaction of light with zebrafish tissues provides information on their structure and properties. In this study, we developed a microscopic imaging system for improving the visualization of unstained zebrafish tissues on tissue slides, with two different setups: polarized light imaging and polarized hyperspectral imaging. Based on the polarized light imaging setup, we collected the RGB images of Stokes vector parameters (S0, S1, S2, and S3), and calculated the Stokes vector derived parameters: the degree of polarization (DOP), the degree of linear polarization (DOLP)). We also calculated Stokes vector data based on the polarized hyperspectral imaging setup. The preliminary results demonstrate that Stokes vector data in two imaging setups (polarized light imaging and polarized hyperspectral imaging) are capable of improving the visualization of different types of zebrafish tissues (brain, muscle, skin cells, blood vessels, and yolk). Using the images collected from larval zebrafish samples by polarized light imaging, we found that DOP and DOLP could show clearer structural information of the brain and of skin cells, muscle and blood vessels in the tail. Furthermore, DOP and DOLP parameters derived from images collected by polarized hyperspectral imaging could show clearer structural information of skin cells developing around yolk as well as the surrounding blood vessel network. In addition, polarized hyperspectral imaging could provide complementary spectral information to the spatial information on Stokes vector data of zebrafish tissues. The polarized light imaging & polarized hyperspectral imaging systems provide a better insight into the microstructures of zebrafish tissues.

High-speed fluorescence excitation-scanning hyperspectral imaging microscopy using thin-film tunable filters.

Hyperspectral imaging (HSI) technologies have enabled a range of experimental techniques and studies in the fluorescence microscopy field. Unfortunately, a drawback of many HSI microscope platforms is increased acquisition time required to collect images across many spectral bands, as well as signal loss due to the need to filter or disperse emitted fluorescence into many discrete bands. We have previously demonstrated that an alternative approach of scanning the fluorescence excitation spectrum can greatly improve system efficiency by decreasing light losses associated with emission filtering. Our initial system was configured using an array of thin-film tunable filters (TFTFs, VersaChrome, Semrock) mounted in a tiltable filter wheel (VF-5, Sutter) that required ~150-200 ms to switch between wavelengths. Here, we present a new configuration for high-speed switching of TFTFs to allow rapid time-lapse HSI microscopy. A TFTF array was mounted in a custom holder that was attached to a piezoelectric rotation mount (ThorLabs), allowing high-speed rotation. Switching between adjacent filters was achieved using the internal optics of a DG-4 lightsource (Sutter Instrument), including a pair of off-axis parabolic mirrors and galvanometers. Output light was coupled to a liquid lightguide and into an inverted widefield fluorescence microscope (TI-2, Nikon Instruments). Initial tests indicate that the HSI system provides a 15-20 nm bandwidth tunable excitation band and ~10-20 ms wavelength switch time, allowing for high-speed HSI imaging of dynamic cellular events. This work was supported by NIH P01HL066299, R01HL169522, NIH TL1TR003106, and NSF MRI1725937.

Spatially resolved current density distribution in GaN-based flip-chip green mini-LEDs by microscopic hyperspectral imaging and modified two-level modeling.

A modified two-level model is proposed to study the spatially resolved current density distribution of GaN-based green miniaturized light-emitting diodes (mini-LEDs), combining with microscopic hyperspectral imaging. We found that the spatially resolved current density distribution reveals both the radiative and non-radiative recombination mappings, which can also be provided separately by this model. In addition, higher current density is not necessarily correlated with higher photon emission, especially for the regions around the electrode edges, where the high current density suggests current crowding and defect-related non-radiative recombination. The current density distribution of mini-LEDs is further verified by the laser-beam-induced current (LBIC) and the spatially resolved mappings of peak wavelength and FWHM. The modified two-level model also offers radiative/non-radiative mappings and is proved to be beneficial to determine the micro-zone current density distribution and to reveal the intrinsic radiative/non-radiative recombination mechanism of mini-LEDs.

Hyperspectral upgrade solution for biomicroscope combined with Transformer network to classify infectious bacteria.

Infectious diseases caused by bacterial pathogens pose a significant public health threat, emphasizing the need for swift and accurate bacterial species detection methods. Hyperspectral microscopic imaging (HMI) offers nondestructive, rapid, and data-rich advantages, making it a promising tool for microbial detection. In this research, we present a highly compatible and cost-effective approach to extend a standard biomicroscope system into a hyperspectral biomicroscope using a prism-grating-prism configuration. Using this prototype, we generate 600 hyperspectral data cubes for Listeria, Bacillus typhi, Bacillus pestis, and Bacillus anthracis. Additionally, we propose a Transformer-based classification network that achieves a 99.44% accuracy in classifying these infectious pathogens, outperforming traditional methods. Our results suggest that the successful combination of HMI and the optimized Transformer-based classification network highlights the potential for rapid and precise detection of infectious disease pathogens .

Polarized hyperspectral microscopic imaging system for enhancing the visualization of collagen fibers and head and neck squamous cell carcinoma.

Polarized hyperspectral microscopes with the capability of full Stokes vector imaging have potential for many biological and medical applications. The aim of this study is to investigate polarized hyperspectral imaging (PHSI) for improving the visualization of collagen fibers, which is an important biomarker related to tumor development, and improving the differentiation of normal and tumor cells on pathologic slides. We customized a polarized hyperspectral microscopic imaging system comprising an upright microscope with a motorized stage, two linear polarizers, two liquid crystal variable retarders (LCVRs), and a compact SnapScan hyperspectral camera. The polarizers and LCVRs worked in tandem with the hyperspectral camera to acquire polarized hyperspectral images, which were further used to calculate four Stokes vectors: , , , and . Synthetic RGB images of the Stokes vectors were generated for the visualization of cellular components in PHSI images. Regions of interest of collagen, normal cells, and tumor cells in the synthetic RGB images were selected, and spectral signatures of the selected components were extracted for comparison. Specifically, we qualitatively and quantitatively investigated the enhanced visualization and spectral characteristics of dense fibers and sparse fibers in normal stroma tissue, fibers accumulated within tumors, and fibers accumulated around tumors. By employing our customized polarized hyperspectral microscope, we extract the spectral signatures of Stokes vector parameters of collagen as well as of tumor and normal cells. The measurement of Stokes vector parameters increased the image contrast of collagen fibers and cells in the slides. With the spatial and spectral information from the Stokes vector data cubes (, , , and ), our PHSI microscope system enhances the visualization of tumor cells and tumor microenvironment components, thus being beneficial for pathology and oncology.

Rapid and high accurate identification of Escherichia coli active and inactivated state by hyperspectral microscope imaging combing with machine learning algorithm

Rapid identification of the active state of foodborne bacteria is crucial for ensuring the safety and quality control of food or pharmaceutical products. In this study, a combination of hyperspectral microscope imaging (HMI) and machine learning algorithm is employed for the identification of active state of Escherichia coli (E. coli). Hyperspectral microscope images of live, 100 ℃ heat inactivation and 121 ℃ high-pressure inactivation of E. coli are collected in wavelength range of 370–1060 nm. Savitzky-Golay (SG) smoothing combing with normalization is used for spectra preprocessing. And principal component analysis (PCA) is employed for spectral dimension reduction. Four different regions of interest (ROIs), including the entire bacterial cell ROI (cell), the outer cell wall ROI (cell_r), the membrane structure ROI (cell_w) formed by the cell wall and cell membrane, and the central of the cell ROI (cell_cy), are extracted and used as model input variables to investigate the influence on the modeling results. Five model algorithms, support vector machines (SVM), random forests (RF), k-nearest neighbors (KNN) algorithms, discriminant analysis (DA) classifiers, and long short-term memory (LSTM) neural networks are used and compared. Modeling results with spectral data of cell_r perform better than those with other ROIs. Accuracy of the models with data of the cell_r ROI are as follows: 79.78% for SVM, 95.11% for RF, 91.33% for KNN, 98.22% for DA, and 93.78% for LSTM. DA achieves the highest classification accuracy. The results show that high-temperature inactivation induces changes in bacterial tissue and morphology, resulting in certain spectral differences among bacteria in three different states. The combination of hyperspectral microscope imaging and machine learning algorithm can provide an effective method for identification of active and inactive states of E. coli. Furthermore, the model, constructed with the data of cell_r ROI, exhibits the best performance in identification.

Compressive hyperspectral microscopic imaging using spectral-coded illumination

A hyperspectral microscopic imaging system based on compressive sensing theory using spectral-coded illumination is presented in this paper. The spectral modulator, which consists of a liquid crystal variable retarder and two polarizers, is employed to encode the spectral transmittance of the incident light by adjusting the voltage of the liquid crystal variable retarder. The paper also provides the imaging model and reconstruction method, and the feasibility is confirmed through computer simulations and laboratory experiments. The imaging system combines traditional microscopy with spectrally encoded illumination and offers an alternative approach for designing a low-cost and compact hyperspectral microscope.

Microscopic hyperspectral imaging and an improved detection model based detection of Mycogone perniciosa chlamydospore in soil

The Mycogone perniciosa disease of Agaricus bisporus is highly contagious, with insignificant early symptoms and a long infestation period. Currently, there is a lack of a convenient and rapid means to detect the disease for early control. By using the transmission route of the disease, this paper focuses on M. perniciosa chlamydospore detection in contaminated soil. Microscopic hyperspectral images of M. perniciosa chlamydospore in contaminated soil were obtained to establish a detection model. Given a small target and a complex background, we proposed an improved spore detection model based on a faster regional convolutional neural network (Faster R-CNN). Furthermore, we combined the residual network Resnet50 and feature pyramid network (FPN) to extract thick spore target features at multiple scales. Meanwhile, we optimized the region proposal network (RPN) region proposal generation by adding two small scales to improve the performance of the detection model. The first three principal components (with 95% or more information) and RGB images were selected as model inputs, respectively, and the final average precision (AP) was 94.68% and 92.35%, respectively. This PC-based model also was compared to the VGG16 and Resnet50-based feature extraction networks of Faster R-CNN and Darknet53-based feature extraction network of the YOLOv3 model, and the AP was found to improve by 5.41%, 4.78%, and 6.34%, respectively. The results showed that micro-hyperspectral imaging combined with deep learning methods could accurately detect the chlamydospore in the soil, providing new methods and ideas for the early prevention and detection of M. perniciosa disease of A. bisporus.

Characterization of Rice-Magnaporthe oryzae Interactions by Hyperspectral Imaging.

Hyperspectral imaging has the potential to detect, characterize, and quantify plant diseases objectively and nondestructively to improve phenotyping in breeding for disease resistance. In this study, leaf spectral reflectance characteristics of five rice genotypes diseased with blast caused by three Magnaporthe oryzae isolates differing in virulence were compared with visual disease ratings under greenhouse conditions. Spectral information (140 wavebands, range 450 to 850 nm) of infected leaves was recorded with a hyperspectral imaging microscope at 3, 5, and 7 days postinoculation to examine differences in symptom phenotypes and to characterize the compatibility of host-pathogen interactions. Depending on the rice genotype × M. oryzae genotype interaction, blast symptoms varied from tiny necrosis to enlarged lesions with symptom subareas differing in tissue coloration and indicated gene-for-gene-specific interactions. The blast symptom types were differentiated based on their spectral characteristics in the visible/near-infrared range. Symptom-specific spectral signatures and differences in the composition of leaf blast symptom type(s) resulted in unique spectral and spatial patterns of the rice × M. oryzae interactions based on the size, shape, and color of the symptom subareas. Spectral angle mapper classification of spectra enabled (i) discrimination between healthy (green) and diseased tissue of rice genotypes, (ii) classification and quantification of different blast symptom subareas, and (iii) grading of the host-pathogen compatibility (low - intermediate - high). Hyperspectral imaging was more sensitive to small changes in disease resistance than visual disease assessments and enabled the characterization of various types of resistance/susceptibility reactions of tissue subjected to M. oryzae infection.

Biomolecular condensates modulate membrane lipid packing and hydration

Membrane wetting by biomolecular condensates recently emerged as a key phenomenon in cell biology, playing an important role in a diverse range of processes across different organisms. However, an understanding of the molecular mechanisms behind condensate formation and interaction with lipid membranes is still missing. To study this, we exploited the properties of the dyes ACDAN and LAURDAN as nano-environmental sensors in combination with phasor analysis of hyperspectral and lifetime imaging microscopy. Using glycinin as a model condensate-forming protein and giant vesicles as model membranes, we obtained vital information on the process of condensate formation and membrane wetting. Our results reveal that glycinin condensates display differences in water dynamics when changing the salinity of the medium as a consequence of rearrangements in the secondary structure of the protein. Remarkably, analysis of membrane-condensates interaction with protein as well as polymer condensates indicated a correlation between increased wetting affinity and enhanced lipid packing. This is demonstrated by a decrease in the dipolar relaxation of water across all membrane-condensate systems, suggesting a general mechanism to tune membrane packing by condensate wetting.

Study on CAT activity of tomato leaf cells under salt stress based on microhyperspectral imaging and transfer learning algorithm

Salt stress easily leads to oxidative stress and promotes the catalase (CAT) response in tomato leaves. For the changes in catalase activity in leaf subcells, there is a need for a visual in situ detection method and mechanism analysis. This paper, taking catalase in leaf subcells under salt stress as the starting point, describes the use of microscopic hyperspectral imaging technology to dynamically detect and study catalase activity from a microscopic perspective, and lay the theoretical foundation for exploring the detection limit of catalase activity under salt stress. In this study, a total of 298 microscopic images were obtained under different concentrations of salt stress (0 g/L, 1 g/L, 2 g/L, 3 g/L) in the spectral range of 400–1000 nm. With the increase in salt solution concentration and the advancement of the growth period, the CAT activity value increased. Regions of interest were extracted according to the reflectance of the samples, and the model was established by combining CAT activity. The characteristic wavelength was extracted by five methods (SPA, IVISSA, IRFJ, GAPLSR and CARS), and four models (PLSR, PCR, CNN and LSSVM) were established according to the characteristic wavelengths. The results show that the random sampling (RS) method was better for the selection samples of the correction set and prediction set. Raw wavelengths are optimized as the pretreatment method. The partial least-squares regression model based on the IRFJ method is the best, and the coefficient of correlation (Rp) and root mean square error of the prediction set (RMSEP) are 0.81 and 58.03 U/g, respectively. According to the ratio of microarea area to the area of the macroscopic tomato leaf slice, the Rp and RMSEP of the prediction model for the detection of microarea cells are 0.71 and 23.00 U/g, respectively. Finally, the optimal model was used for quantitative visualization analysis of CAT activity in tomato leaves, and the distribution of CAT activity was consistent with its color trend. The results show that it is feasible to detect the CAT activity in tomato leaves by microhyperspectral imaging combined with stoichiometry.

Classification between live and dead foodborne bacteria with hyperspectral microscope imagery and machine learning

Identifying live foodborne bacteria is essential for ensuring food safety and preventing foodborne illnesses. This study investigated the use of hyperspectral microscope imaging and deep learning methods to accurately distinguish between live and dead foodborne bacteria based on their spectral and morphological features. Three deep learning models, Fusion-Net I, II, and III, were developed and evaluated for their ability to classify live and dead bacterial cells of six pathogenic strains, including Escherichia coli (EC), Listeria innocua (LI), Staphylococcus aureus (SA), Salmonella Enteritidis (SE), Salmonella Heidelberg (SH), and Salmonella Typhimurium (ST). The models utilized both morphological and spectral characteristics of the bacterial cells, with inputs of average spectra and 546 nm band images. Fusion-Net I achieved high accuracy in identifying live bacterial cells, with a classification accuracy of 100% for LI, SE, ST strains and over 92.9% for EC, SA, SH. Fusion-Net II and III models were even more robust, achieving 100% accuracy consistently in classifying dead cells in all six strains. Fusion-Net III also showed the ability to identify bacterial strains with 96.9% accuracy, making it a dual-task model with potential applications in identifying live foodborne bacteria prior to foodborne outbreaks. These findings suggest that the use of hyperspectral microscope imaging and deep learning methods could provide a new tool for quickly and accurately identifying bacterial viability, thereby improving the efficiency and reliability of food safety inspection.