Use this URL to cite or link to this record in EThOS:
Title: The feasibility of Fourier transform infrared imaging spectroscopy in discriminating benign prostatic hyperplasia from prostate cancer in blood serum samples
Author: Monjardez, Geraldine
ISNI:       0000 0004 2737 2808
Awarding Body: University of Manchester
Current Institution: University of Manchester
Date of Award: 2013
Availability of Full Text:
Access from EThOS:
Access from Institution:
The feasibility of Fourier transform infrared (FTIR)-imaging spectroscopy as a tool to discriminate samples from patients suffering from benign prostatic hyperplasia (BPH) and prostate cancer (CaP) samples in blood serum was investigated. Prostate cancer is known to be an age related disease, with the risk of developing the disease dramatically increasing in men past forty years old. Currently the PSA blood test is notoriously unreliable and is non specific for CaP thus leading to overtreatment of the disease. It is important therefore to develop diagnostic method that is non-invasive, reliable, and specific for CaP.In order to achieve the objective of establishing a robust protocol, which could be applied to a clinical study, obtaining optimal sample preparation for the FTIR analysis of serum smears, had to be achieved. A protocol was developed to prepare the serum samples prior to their FTIR analysis. First, the samples were centrifuged with ultrafiltration devices of different sizes to obtain several fractions which were then smeared to obtain thin films of serum. The spectra from the larger (>100 kDa components) and medium (containing the 10–100 kDa components) fractions were utilised for both a pilot and a clinical study, while the spectra from the smaller fractions (containing the 3–10 and <3 kDa components) were affected by fringing and could therefore not be used. A major novelty of this project involved the application of FTIR-imaging to the analysis of serum smears. The use of the Focal Plane Array detector system enabled the collection of a spectral image containing 16,384 spectra, on which a Quality Testing and pre-processing techniques were applied to select the “good spectra” and reject the spectra that failed the Quality Test. Several types of substrates were assessed to determine the most appropriate for the analysis of the smears and it was established that the spectra obtained from the serum smeared on CaF2 windows gave the most reproducible results. 5 BPH and 5 CaP samples were analysed for the pilot study following the developed protocol. While no clear separation was observed in the Principal Component Analysis (PCA) plots between the BPH and the cancerous samples, a trend emerged throughout the results, with the CaP samples clustering together and the BPH samples scattered around them. A larger clinical study was conducted with 60 BPH samples and 60 CaP samples. PCA was applied on the “good spectra” and while the over 100 kDa fraction did not show a clear separation between the two types of samples, the 10–100 kDa fraction showed a distinct classification between the BPH and CaP samples. An artificial neural network was then applied to create a model using patients from the database used for the PCA analysis to determine whether the discrimination between the two types of samples could be increased or highlight different classification trends. For the >100 kDa fraction, the sensitivity value was calculated to be 97.8% and the specificity value was calculated to be 44.3% while the sensitivity and the specificity value for the 10 to 100 kDa fraction were calculated to be 78.9% and 60% respectively. A complementary study using mass spectrometry was carried out on healthy and diseased samples to identify the components contained within the different fractions and determine whether they could be correlated with the components identified from the spectral features of the FTIR data. While no quantitative information was obtained from this study, the components found in the different fractions were identified, confirming the results of the FTIR studies.
Supervisor: Gardner, Peter Sponsor: Engineering and Physical Sciences Research Council ; Royal Society of Chemistry
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: Fourier Transform Infrared imaging ; Mass spectrometry ; Sample preparation ; Blood serum ; Clinical study ; Chemometrics ; Prostate Cancer