Use this URL to cite or link to this record in EThOS:
Title: Investigating the use of attenuated total reflection Fourier-transform infrafred (ATR-FTIR) spectroscopy for the rapid diagnosis of brain tumours using human blood serum
Author: Hands, James R.
ISNI:       0000 0004 5921 7212
Awarding Body: University of Strathclyde
Current Institution: University of Strathclyde
Date of Award: 2015
Availability of Full Text:
Access from EThOS:
Access from Institution:
The ability to diagnose brain cancer rapidly from human serum would allow for short testing times and prompt results providing a responsive diagnostic environment. This study demonstrates a new method for primary and metastatic brain cancer diagnosis using 1 μl volumes of human serum and attenuated total reflection Fourier-transform infrared (ATR-FTIR) spectroscopy. To the best of our knowledge, this is the largest study on mid-infrared spectroscopy in relation to cancer research with a 433 patient cohort consisting of 3,897 ATR-FTIR spectra. Spectral data from whole serum, 100 kiloDalton (kDa), 10 kDa and 3 kDa molecular weight cut-off filtrate samples to investigate which fraction allowed for optimum differentiation of disease state and brain tumour severity (high grade vs. low-grade glioma) from non-cancer. The fingerprint region (1800-1000 cm-1) of the acquired data was combined with an RBF-SVM and achieved optimum sensitivities and specificities from whole serum averaging 93.75 and 96.53 % respectively when distinguishing between brain tumour severities. The 1 μl serum spots dried after 8 minutes and the acquired spectra exhibited minimal variance, especially after preprocessing. Expanding the research to, for the first time, detect from cancer vs. noncancer to organ of origin of metastatic disease from the same serum sample achieved optimum sensitivities and specificities of between 80.0 and 100 % respectively. Furthermore, feature extraction fed SVM analysis of the cancer vs. non-cancer spectral model was performed to maximise classification accuracies to achieve improved sensitivities and specificities, in contrast to fingerprint region based SVM, of 91.5 and 83.0 % respectively.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral