Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.757932
Title: Improving the performance of microscope mass spectrometry imaging
Author: Guo, Ang
ISNI:       0000 0004 7430 7400
Awarding Body: University of Oxford
Current Institution: University of Oxford
Date of Award: 2018
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
Mass spectrometry imaging (MSI) is a powerful tool that provides mass-specific surface images with micron or sub-micron spatial resolutions. In a microscope MSI experiment, large sample surfaces are illuminated with a defocused laser or primary ion beam, enabling all surface molecules to be desorbed and ionised simultaneously before being electrostatically projected onto a position-sensitive imaging detector at the end of a time-of-flight mass analyser. Traditionally only the image of one mass-to-charge ratio can be obtained in a single acquisition, which limits its applicability. However, the development of event-triggered sensors, such as CMOS-based cameras, revives the microscope MSI method by allowing multi-mass imaging. Therefore, the challenges facing microscope have MSI shifted to improving its mass resolution, effective mass range, and mass accuracy. This thesis proposes effective solutions to each of them, and thus significantly improves the performance and applicability of microscope MSI. To increase the mass range, two modified post-extraction differential acceleration (PEDA) techniques, double-field PEDA and time-variable PEDA, were used to demonstrate mass-resolved stigmatic imaging over a broad m/z range. In double-field PEDA, a potential energy cusp was introduced into the ion acceleration region of an imaging mass spectrometer, creating two m/z foci that were tuned to overlap at the detector plane. This resulted in two focused m/z distributions that stretched the mass-resolved window with m/Δm >= 1000 to 165 Da without any loss in image quality; a range that doubled the 65 Da achieved under similar conditions using the original PEDA technique. In time-variable PEDA, a dynamic pulsed electric field was used to maximize the effective mass range of PEDA. By simultaneously focusing ions between 300 to 700 m/z using an exponentially rising voltage pulse, time-variable PEDA provides an effective mass range more than six times wider than the original PEDA method. Although reflectrons are widely used to improve the mass resolving power of ToF-MS, incorporating them in a microscope MSI instrument is novel. A reflectron MSI instrument was designed and implemented. Simulations demonstrated that one-stage gridless reflectrons were more compatible with the spatial imaging goal of the microscope MSI instrument than the gridded reflectrons. Preliminary experimental results showed that coupling the gridless reflectron with single-field PEDA achieved a mass resolution above 8,000 m/Δm while keeping a spatial resolution of 20 um. In conclusion, the gridless reflectron was able to triple the mass resolving power without losing any spatial imaging power. The poor mass accuracy hurdle was overcome by machine learning algorithms, which can construct clinical diagnostic models that recognise the peak pattern of biological mass spectra and classify them accurately without knowing the actual mass of each peak. After a proof of concept "experiment", where the mass spectra of dye molecules were classified by various learning algorithms, three pairs of datasets (ovarian cancer, prostate cancer, chronic fatigue and their respective controls) were used to build classifiers that accurately distinguish blood samples from controls. Possible biomarkers were also discovered by evaluating the importance of each m/z feature, which may assist further studies.
Supervisor: Brouard, Mark Sponsor: China Scholarship Council
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.757932  DOI: Not available
Keywords: Mass Spectrometry Imaging ; Ion Optics ; Machine Learning ; Reflectron
Share: