Noninvasive Diagnostics of Lung Cancer Based on Whole Blood Surface-Enhanced Raman Spectroscopy and Deep Machine Learning.

Combining deep machine learning with silver nanoparticle (Ag NP)-based surface-enhanced Raman spectroscopy (SERS), we have developed a novel method for whole blood analysis for cancer detection applications. The whole blood was collected from two groups: one group of patients (n = 26) with lung cancer and another group of healthy volunteers (n = 45). The logistic regression (LR), k-nearest neighbor (KNN), decision tree (DT), and random forest (RF) algorithms were employed to develop a diagnostic model using the same spectral data. The results show that the diagnostic accuracy of LR, KNN, DT, and RF models was 87, 66, 77, and 83%, respectively. LR is superior to other algorithms in the SERS spectra classification of whole blood. We therefore believe that this proposed strategy will have great clinical potential for SERS technology combined with LR and act as a complementary method for the detection of lung cancer. [ABSTRACT FROM AUTHOR]