Vibrational Spectroscopy and Chemometrics for Lung Cancer Diagnosis

In this thesis, three interconnected studies evaluate the potential of vibrational spectroscopy combined with chemometrics, for lung cancer detection using human blood plasma. The first study develops and compares classification models based on Raman spectroscopy, achieving predictive performance with accuracy near 90%. The second study applies feature selection methods to identify discriminative spectral features, primarily associated with lung cancer-related alterations in specific molecules. The third study explores data fusion strategies to integrate Raman and FTIR datasets, resulting in enhancing model performance with accuracy reaching 99.2%. Together, these studies support the application of photonic and chemometric methods for accurate, rapid, and minimally invasive lung cancer screening.