Preliminary study of estradiol and progesterone detection based on terahertz spectroscopy

Abstract

Estradiol (E2) and progesterone (PROG) are the two typical reproductive hormones in cows. The concentrations of E2 and PROG in milk are highly correlated with the estrus in cows. In the present work, the feasibility of qualitative and quantitative detection of E2 and PROG by terahertz time-domain spectroscopy (THz-TDS) is preliminarily investigated. 522 sample tablets of E2, PROG, powdered milk, estradiol-powdered milk (E2-PM), progesterone-powdered milk (PROG-PM) and estradiol-progesterone-powdered milk (E2-PROG-PM) are prepared, and the time-domain spectral, frequency-domain spectral and absorption coefficients are collected and analyzed. The density functional theory (DFT) is used to simulate the molecular dynamics of E2 and PROG. The four preprocessing methods, including first order derivative (FD), standard normal variant (SNV), multiple scatter correction (MSC) and Savitzky-Golay (S-G) along with different modeling methods, including support vector regression (SVR), partial least squares regression (PLSR) and convolutional neural network (CNN) are used for quantitative model building and predictions. The prediction set correlation coefficient (Rp) having value 0.9843 for the E2 concentration in the E2-PM, 0.9935 Rp of the PROG concentration in PROG-PM, and 0.9830 Rp for E2 and PROG concentrations in E2-PROG-PM are achieved by using the combination of None & CNN model, None & CNN model and S-G & CNN model respectively. Our study indicates that the use of terahertz technology to determine hormone content has great potential applications in the field of identification of estrus in dairy cows.