Developing biophotonic techniques for the rapid and non-invasive diagnosis of cancer from biofluids

Abstract

Vibrational spectroscopy relates to the specific optical techniques of Fourier-transform infrared (FTIR) and Raman spectroscopy (RS). These techniques probe molecular vibrations of the sample when light interacts with it, which present ‘fingerprints’ of the global chemical composition. Both techniques hold great promise in disease diagnostics, especially with ‘liquid biopsies’ for biofluids. This study developed bio-spectroscopic methodologies to query the serum biochemistry towards rapid diagnosis and detection of diseases. The aim was two-fold; i) to develop FTIR and RS methodologies to analyse sera for pre-analytical variation. Secondly, to use liquid RS combined with chemometric analysis to interpret pathological data for its diagnostic potential. Beyond the proof-of-concept, with investigations into preanalytical variations (which proved no effect is seen on the serum profile) via serum freeze-thawing and environmental drying, three diagnostic studies were sought; from patient cases, i.e. cirrhotic sera with and without hepatocellular carcinoma, sera with different levels of fibrosis, and with varying stages of brain tumours. Throughout the thesis, a suite of FTIR and Raman spectroscopy techniques were developed/employed, such as attenuated total reflectance, high throughput FTIR, and Raman spectroscopy on liquid and dried human sera. Advanced chemometric approaches were employed such as clustering (PCA, HCA), partial least squares, and forward linear discriminant analysis, radial basis function support vector machine with leave-one-out cross validation, random forest classifiers, all towards developing a robust disease classifier. Across all diagnostic studies, results showed moderate-to-good diagnostic abilities. It was shown that vibrational spectroscopy combined with advanced chemometric methods can provide a good adjunct to clinical screening settings, such as point-of-care areas.