Multiwavelength Regression Algorithm for Eliminating Chamber Surface Effects of Microfluidic Chips.

An ultraviolet visible (UV-Vis) spectrophotometric and multiwavelength linear regression (MLR) method was developed for eliminating the influence of the surface quality of centrifugal microfluidic chips on the accuracy of their absorbance detection. The regression model is based on scalar scattering theory. The method was validated with cuvettes with different surface quality and Orange G (orange gelb) dye. The coefficients of variation (CVs) of the predicted solution concentration ratios in different cuvettes were < 1%, and the relative errors were < 1.5%. The model was shown to have higher accuracy and precision than that of traditional methods.