Abstract
In the present study, efficacy of minimally-invasive serum Raman spectroscopy (SRS) in stratification of congenital heart diseases was explored. Blood was collected from 62 subjects [42 congenital heart defect (CHD) patients (19 with atrial septal defect, 13 with ventricular septal defect and 10 with tetralogy of fallot) and 20 controls], and serum separated. Raman spectra of sera were recorded, pre-processed and subjected to spectral and multivariate analyses. Multivariate curve resolution-alternating least squares (MCR-ALS) analyses indicated alterations in lipid and protein levels between the study groups. Principal Component Analysis (PCA) and Principal Component based Linear Discriminant Analysis (PC-LDA), cross-validated with Leave-one-out cross validation (LOOCV), were employed to study stratification between the different groups. CHD could be classified from controls with 76 % efficiency. The different CHD subtypes could be distinguished with efficiencies as high as ∼90 %. To the best of our knowledge, differentiation between controls and CHDs as well as the stratification between controls and CHDs subtypes was for the first time successfully accomplished by serum-based Raman spectroscopy.