CERES > Cranfield University at Silsoe (1975-2008) > PhD, EngD, DM and MSc by research theses >

Please use this identifier to cite or link to this item: http://dspace.lib.cranfield.ac.uk/handle/1826/5738

Document Type: Thesis or dissertation
Title: Volatile diagnostic techniques for ventilator associated pneumonia
Authors: Humphreys, Martyn Lee
Supervisors: Kendall, C.
Magan, Naresh
Barr, H.
Issue Date: Feb-2010
Abstract: Ventilator associated pneumonia (VAP) is a significant challenge for the Intensive Care doctors worldwide. It is both difficult to diagnose accurately and quickly and to treat effectively once the diagnosis has been established. Current diagnostic microbiological methods of diagnosis can take up to 48 hours to yield results. Early diagnosis and treatment remain the best way of improving outcome for patients with VAP. In this study we look at novel diagnostic techniques for VAP. Electronic nose (Enose) technology was used to identify to identify the presence of microorganisms in bronchoalveolar lavage (BAL) fluid samples taken from the respiratory tracts of ventilated patients. The results were compared with standard microbiological culture and sensitivities. The Enose was able to discriminate 83% of samples into growth or no growth groups on samples grown in the lab. When the technique was employed to samples taken directly from patients the accuracy fell to 68.2%. This suggests that patient related factors may be reducing the accuracy of the Enose classification. The use of antimicrobial drugs prior to patient sampling is likely to have played a major role. The second part of this study used Gas Chromatography-Mass Spectrometry (GC-MS) analysis of patient’s breath in an attempt to identify patients with VAP. Breath samples were taken at the same time as the bronchoalveolar lavage samples described above. The use of this technique did show differences between the breath samples of patients who did not have any microbiological growth from their BAL samples and those that did. Leave one out cross validation of a PC fed LDA model showed 84% correct classification between healthy volunteers, no growth and growth groups. Finally, we evaluated the Breathotron, which is a breath analysis device designed and built at Cranfield Health. It allows for analysis of breath samples using a single sensor system as opposed to a sensor array employed in traditional Enose devices. This allows it to be more portable and cheaper to build. The Breathotron also allows collection of breath onto sorbent cartridges for GC-MS analysis. Its single sensor did not allow for accurate discrimination between samples.
URI: http://dspace.lib.cranfield.ac.uk/handle/1826/5738
Appears in Collections:PhD, EngD, DM and MSc by research theses

Files in This Item:

File Description SizeFormat
Martyn_Lee_Humphreys_Thesis_2010.pdf2.71 MBAdobe PDFView/Open
Martyn_Humphreys_Table6.3_Thesis_2010.xlsx11.25 kBUnknownView/Open
Martyn_Humphreys_Table 6.1_Thesis_2010.xls256 kBMicrosoft ExcelView/Open

SFX Query

Items in CERES are protected by copyright, with all rights reserved, unless otherwise indicated.