Point-of-Care Using Vis-NIR Spectroscopy for White Blood Cell Count Analysis

Abstract

Total white blood cells count is an important diagnostic parameter in both human and veterinary medicines. State-of-the-art is performed by flow cytometry combined with light scattering or impedance measurements. Spectroscopy point-of-care has the advantages of miniaturization, low sampling, and real-time hemogram analysis. While white blood cells are in low proportions, while red blood cells and bilirubin dominate spectral information, complicating detection in blood. We performed a feasibility study for the direct detection of white blood cells counts in canine blood by visible-near infrared spectroscopy for veterinary applications, benchmarking current chemometrics techniques (similarity, global and local partial least squares, artificial neural networks and least-squares support vector machines) with self-learning artificial intelligence, introducing data augmentation to overcome the hurdle of knowledge representativity. White blood cells count information is present in the recorded spectra, allowing significant discrimination and equivalence between hemogram and spectra principal component scores. Chemometrics methods correlate white blood cells count to spectral features but with lower accuracy. Self-Learning Artificial Intelligence has the highest correlation (0.8478) and a small standard error of 6.92 × 109 cells/L, corresponding to a mean absolute percentage error of 25.37%. Such allows the accurate diagnosis of white blood cells in the range of values of the reference interval (5.6 to 17.8 × 109 cells/L) and above. This research is an important step toward the existence of a miniaturized spectral point-of-care hemogram analyzer.

Reagentless Vis-NIR Spectroscopy Point-of-Care for Feline Total White Blood Cell Counts

“…Blood spectra are governed by multi-scale interference and matrix effects, super-imposing physical (scattering) and chemical (absorbance) information. Unscrambling the information contained in spectral data, due to multi-scale interference, is the major challenge for point-of-care (POC) technologies [ 1 , 2 ] ( Figure 1 a). Multi-scale interference results from overlapping spectral bands of blood constituents, leading to interference at varying intensities and deviations from the Beer–Lambert law (BLL) [ 3 ].…”

“…Reagentless spectral POC demonstrated its effectiveness in quantifying hemogram parameters in dog and cat blood [ 1 , 2 ]: (i) Dog blood—RBCs (6.39%), Hgb (7.14%), hematocrit (HTC) (4.43%); (ii) cat blood—RBCs (5.67%), Hgb (4.08%), and HTC (1.69%). WBCs in dogs were quantified with a correlation of 0.8478, standard error of 6.92 × cells/L and MAPE of 25.37%, allowing us to diagnose WBCs in the reference interval (5.6–17.8 × cells/L) [ 2 ].…”

“…Data augmentation can be used to enhance the diversity of the knowledgebase and improve the accuracy of model predictions [ 2 , 41 , 42 ]. Blood composition representativity is limited in real world data (RWD) due to the large biological variance.…”

“…This is partially solved by self-learning models, which can assess if the measured spectra are predictable, suggesting an update to the knowledgebase [ 12 ]. Synthetic spectroscopy data (SSD) are generated by the random hybridization of the hemogram and spectral information from two samples ( Figure 1 b), creating new spectra and a corresponding hemogram with a 1:1 information mixture, which is comparable to the mixture of the two blood samples, resulting in the average composition of the two RWD samples and assuming no reaction between blood samples (e.g., antigen–antibody of blood type and Rhesus factor) [ 2 ]. SSD improves the knowledgebase by (i) increasing the level of local detail by the interpolation of the existing RWD samples and (ii) expanding the knowledgebase through extrapolation, once the boundary coordinates of the samples in the feature space are convex ( Figure 1 b).…”

Enhancing host-pathogen phenotyping dynamics: early detection of tomato bacterial diseases using hyperspectral point measurement and predictive modeling

“…Spectral information provided by HS is extremely valuable, nonetheless, in biological tissues, it is super-imposed in the recorded spectra at different scales of interference ( Barroso et al., 2022 ; Tosin et al., 2022 ). Moreover, HS data can present substantial amounts of redundant information in contiguous wavelengths, and just some specific spectral features might be relevant to predict and classify diseased tissues ( Caicedo et al., 2014 ; Rivera et al., 2014 ).…”