Biochemical Blood Sensing Platform With CMOS Image Sensor and Software-Based Wavelength Filter

A fully integrated biochemical sensing platform is developed for blood diagnostics. The biochemical analyses of blood include the enzymatic reaction of the sample with a reagent that produces a colorimetric variation of solutions. The developed platform focuses on providing software solutions to the hardware dependencies in the existing analyzers by replacing the physical optical filter wheel assembly with software-based wavelength filtering. The experiments are performed to sense these variations with the developed image acquisition platform and the software-based wavelength filtering unit. The machine learning approach is used to train the system to predict unknown concentrations using the experimentally generated dataset. To verify the usability of the developed platform, the sensing of glucose, albumin, and cholesterol was carried out in the linear range from 10 to 400 mg/dL, from 1 to 8 g/dL, and from 10 to 800 mg/dL with a limit of detection (LOD) of 2.8 mg/dL, 0.64 g/dL, and 3.14 mg/dL, respectively. The results obtained are matching with the commercial biochemistry analyzer. The system is tested thoroughly, and the results show that the developed blood sensing platform is reliable, portable, and accurate due to the minimization of hardware moving parts, and offers extreme flexibility of operation due to software solutions, facilitating numerous blood analysis all-in-one instruments.