BibTeX entry

@PHDTHESIS{200907Jorge_Cabello,
  AUTHOR={Jorge Cabello},
  TITLE={High throughput digital Beta autoradiography imaging},
  SCHOOL={University of Surrey},
  MONTH=Jul,
  YEAR=2009,
  URL={http://www.bmva.org/theses/2009/2009-cabello.pdf},
}

Abstract

This thesis presents three main strands of work concerned with developing digital imaging for high throughput  autoradiography. These three strands comprise work with the image sensor technology, Monte Carlo simulation and the use of post-acquisition image analysis based on image registration. In this way, the complete autoradiography imaging chain is addressed. CCD and CMOS imaging technologies are presented as potential imaging alternatives to using conventional film in autoradiography. These digital technologies exhibit enhanced sensitivity, dynamic range and linearity compared to film using imaging methods developed at Surrey. These imaging methods address the different sources of noise typically present in CCD and CMOS technologies. Tissue imaging using 3H, 35S and 121I, the typical radioisotopes used by the Drug Addiction Group in the School of Biomedical and Biological Sciences, is presented. The first successful images of 3H-labeled tissue sections using CCD and CMOS technologies operating at room temperature are presented as one of the main achievements of this work. To better understand the image creation process some preliminary Monte Carlo simulations, using the GEANT4 toolkit, have been undertaken, demonstrating intrinsic and extrinsic key parameters of these digital sensors that can be used to optimize spatial resolution. These simulations demonstrate that each radioisotope requires a different optimum detector architecture. In this work these optimum architectures are analyzed. To support the high sensitivity (i.e. fast) imaging produced by the sensor technology, automated post-acquisition analysis is also considered, using an atlas-based image registration approach, by previously aligning automatically segmented biological landmarks using a feature-based extraction approach, region growing. This has the potential to speed up the post-acquisition analysis aspects of the imaging chain. Thus a computer-based tool designed to semi-automatically elastically register a radiogram with an atlas has been developed.