Dill (Anethum Greveolens L.), a biennial or annual herb is globally known for its utilization in various food industries, including refrigeration and food concentrates, to impart its distinctive flavor and seasoning to a range of dishes. Dill seed also offers nutritional benefits as it contains various essential nutrients such as vitamin A, calcium, magnesium, sodium, potassium, fiber, protein and niacin. Dill seed is usually adulterated with fennel, cumin seeds, Indian dill (Anethum sowa) and other resembling dill cultivars or varieties for economic gain. In this study, shortwave infrared (SWIR) hyperspectral imaging was deployed to differentiate three dill varieties based on reflectance features. Three hundred hyperspectral images captured in the wavelength range of 900 to 2500 nm were processed, segmented and subjected to machine learning algorithms. Multiclass classifier was used to classify the dill varieties based on reflectance spectra. All the three dill varieties were classified correctly with a classification accuracy of 100%.

In recent years, frequent reports of food adulteration have increased the concerns of consumers, food industries and regulatory agencies. Food adulteration is termed as a deliberate offensive act to delude people for monitory benefits. Chickpea is broadly consumed pulse in the world. It is rich in protein and also a good source of dietary fiber, carbohydrates, vitamins, minerals, and several bioactive components. It is often adulterated with low cost and low protein resembling materials like corn flour. In the present study, chickpea flour was adulterated with different percentage levels (1, 3, 5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 95, 97, 99, and 100%) of corn flour. Images of mixed flour samples were acquired with 25 replications using shortwave infrared (SWIR) hyperspectral imaging (HSI) in the wavelength range between 900 and 2500 nm. The captured images were processed, segmented and subjected to machine learning algorithms. The classification of different levels of corn flour adulteration was achieved with 98% accuracy using multiclass classifier. Variations in protein and starch contents of adulterated samples can be attributed to the higher classification accuracy.

The significant shift in customer eating habits towards high-protein, low- carb, gluten-free alternatives has forced the food processing industry to come up with more innovative solutions to meet rising demands. Specifically, the pasta industry is currently developing techniques to incorporate more protein per serving while ensuring the palatability of the product. Potato flour (PF) is a gluten- free, versatile flour that can be incorporated into baked goods such as pasta and gnocchi, while chickpea flour (CF), known for its low glycemic index, high protein and essential vitamins which is ideal for fortification. In this research, potato and chickpea flour blends of various concentrations were prepared and analyzed for their physiochemical properties; specifically starch and protein content, which could help pasta- making. The NIR hyperspectral imaging system was used in the range of 900-2500 nm to scan 17 different flour blends in reflectance mode. Spectral features were extracted from the acquired images which were further processed and analyzed using machine learning approaches. The regression analysis predicted protein and starch content with a correlation coefficient of 0.999 and 0.996 respectively.

Potatoes are the widest grown vegetable in Canada, in addition to being the fifth largest primary agricultural crop. Due to their unique end-uses and value addition to the market as processed potatoes, it becomes vital to ensure potato quality from harvest to the consumption. This study intended to understand the effect of storage on dry matter (DM) and accumulated sugars. Hence, the tubers were stored at two different temperatures 2℃ and 15℃, in boxes that admitted little to no light. For the DM study, 277 Russet Burbank (RB) potatoes were scanned in the short-wave infrared (SWIR) of 900-2500 nm range using a hyperspectral imaging system. The relationship between tuber specific gravity (SG) vs DM for individual potatoes was also explored by using a contraption attached to the texture analyzer, that was developed in-house. Similarly, a separate batch containing 130 RB potatoes (stored at 2℃ and 15℃) was scanned for their sugar content; followed by wet chemical techniques to estimate total glucose and sucrose. A 10-fold cross-validation test conducted on the dataset revealed a 90.2% and 82.8% classification accuracy, with a root mean square error of 0.105 and 0.2403, respectively, for observed DM and sugar values, imaged on various days.

Glyphosate, a widely used organophosphate herbicide, poses significant concerns globally due to its persistent presence in pulse crops, leading to food safety challenges. This research investigates the potential of Near-Infrared (NIR) hyperspectral imaging, operating between 900 to 2500 nm wavelengths, for identifying glyphosate residues in red lentil flour. Red lentil flour samples were tested at five different glyphosate concentrations (0 ppm, 5 ppm, 10 ppm, 15 ppm, and 20 ppm) to assess residue levels. Spectral data preprocessing involved smoothing and first derivate techniques. Partial Least Square Regression (PLSR) models were developed incorporating chemical reference measurements and spectral data, refined using a variable importance in projection (VIP)-based method to identify crucial wavelengths for glyphosate detection. Results demonstrated robust predictive capabilities, with a correlation coefficient of 0.930 and a root mean square error of cross-validation (RMSEP) of 0.871 across the concentration range. These outcomes underscore the potential of NIR hyperspectral imaging in accurately quantifying glyphosate residue levels in red lentil flour, thereby mitigating food safety risks associated with agricultural products. This study highlights the efficacy of this non-destructive technique, providing swift and precise detection methods. Future research could explore integrating this approach into routine monitoring practices in food processing facilities to ensure strict adherence to food safety standards.

Rapid, non-destructive and precise prediction of yellow pea flour adulteration in chickpea flour is crucial for food inspection agencies. In present study, short wave infra-red (SWIR) hyperspectral imaging (HSI) in the wavelength range between 900 nm and 2500 nm was deployed to predict the protein and starch in the chickpea flour adulterated with yellow pea. Adulteration was formulated at different levels of yellow pea, i.e. 1, 3, 5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 95, 97, 99, and 100%. Gaussian process regression (GPR), support vector machine regression (SVMR) and neural network models based on different spectral preprocessing techniques were developed by correlating the hyperspectral data with measured reference protein and starch values of the adulterated samples. Competitive adaptive reweighted sampling (CARS) and iteratively retaining informative variables (IRIV) algorithms were used to select the effective wavelengths with full spectrum. The optimum prediction model for protein was obtained using GPR yielding correlation coefficient of prediction (R2p) and root mean square error of prediction (RMSEP) values of 0.99 and 0.089, respectively. Similarly, the same GPR model predicted the starch content with R2p and RMSEP values of 0.86 and 0.549, respectively.

Natural polymers, including proteins and polysaccharides, have been receiving increasing attentions recently as sustainable materials for the development in biomedical applications. Natural polymers extracted from plants or animal products provide great biocompatibility and biodegradability, which are highly desirable in the biomedical field. Hydrogels are highly hydrated, porous, and soft materials with a touch similar to that of the human skin and have therefore attracted extensive interest in the development of wearable biosensors. Egg white is a major nutrient source that has been utilized for extraction of functional proteins. However, the costly and complex process of isolation and purification limits the utilization of egg white and its derivatives. Recently, there has been immense interests in using raw egg white with advanced applications such as supercapacitors and semiconductors, although they still involve complex fabrication methods. Here, we prepared a physically crosslinked egg white (EW) hydrogel with extraordinary stretchability using a simple method. The prepared EW hydrogel showed shear-thinning and self-healing properties that enabled its 3D-printing to fabricate the complex architectures of the biosensor. Incorporation with carbon nanotubes, the versatile EW hydrogel-based biosensors can capture a wide range of human motion including finger bending, wrist pulse, and respiration rate. The biosensor also effectively discerned the cardiovascular system's radial augmentation and stiffness indexes. A highly sensitive humidity-responsive reversible actuator was fabricated through a design of gradient crosslinking density structure. This low-cost natural material provides an easy and effective way for the tailored fabrication of multifunctional biosensors and actuators.