Fig. 5: Paraquat concentration prediction trial using spectra preprocessed by different methods. | Light: Science & Applications

Fig. 5: Paraquat concentration prediction trial using spectra preprocessed by different methods.

From: RSPSSL: A novel high-fidelity Raman spectral preprocessing scheme to enhance biomedical applications and chemical resolution visualization

Fig. 5

a The trial workflow. b Curves and shadows showing the normalized SERS spectra with mean ± SD obtained from paraquat solution with different concentrations (10−4 M, 10−5 M, 10−6 M, and 10−7M). c Preprocessed spectra of b using RSBPCNN#. d Comparison of average paraquat concentration prediction error EQ of spectra preprocessed by different methods at different training data proportions (n = 100 spectra for each concentration). Statistical significance was calculated using the Wilcoxon signed-rank test for two correlated samples between the superior established methods (UNet-1D* for proportions of 0.5:9.5, 1:9, and 2:8; Residual CNN* for proportions of 4:6, 5:5, and 7:3) and RSBPCNN#. An asterisk * indicates P < 0.05, ** indicates P < 0.01, *** indicates P < 0.001, **** indicates P < 0.0001

Back to article page