Physical Sciences of Imaging in the Biomedical Sciences CDT
Completed in 2016 and progressed into a Pre-registration Clinical Scientist position.
Thesis project - "Development of a Multi-Modal, Optical Imaging System for use in Pre-Clinical Studies"
Dr Hamid Dehghani, School of Computer Science
Dr Iain Styles, School of Computer Science
Professor Phil Newsome, Institute of Immunology and Immunotherapy
This work involves development of a multi-modal optical imaging system (BLDOT), which is capable of bioluminescence tomography (BLT), diffuse optical tomography (DOT) and three-dimensional surface capture (SC). BLT is an extension of bioluminescence imaging (BLI), a planar imaging technique widely used in pre-clinical research to monitor the distribution of bioluminescent sources within an imaging subject. The sources, such as luciferases can be used to label different cell types, for example: cancer cells, enabling tumour growth monitoring or novel drugs to be tested; stem cells, bacteria and immune cells, enabling monitoring of their migration and location within the imaging subject. However, as BLI only provides two-dimensional information, it is impossible to determine the true location of the source within the subject. Additionally, due to the unknown underlying tissue attenuation, which affects the path of the bioluminescent light through the tissue, the size and intensity of the source cannot be determined accurately. Therefore, we aim to develop, test and utilise novel tomographic methods through the development of novel physical systems and computational algorithms to provide quantitative, three dimensional, volumetric information about the bioluminescent source.
BLT reconstruction algorithms currently ignore the fact that multi-spectral image data is taken using band pass filters with a finite bandwidth, assuming that data is collected at the central filter wavelength only. As data is collected at all wavelengths which are transmitted through the filter (defined by its bandwidth), the assumption that it originates from a single wavelength introduces unknown and previously not understood errors. We will develop, test and evaluate novel computational algorithms that will incorporate filter bandwidth modelling, increasing the quantitative accuracy of BLT. This will be evaluated using simulations, phantom and real model studies.
To increase the quantitative accuracy of BLT further, we aim to perform simultaneous BLT and DOT. BLT relies on the use of a 'model' based parameter reconstruction algorithm, which requires some prior knowledge about the underlying tissue attenuation. DOT will enable subject-specific tissue attenuation to be calculated and incorporated into the BLT source recovery, enabling a higher degree of accuracy when modelling the propagation of light through the tissue, and therefore more accurate source reconstructions. We will develop novel systems and instrumentation to incorporate DOT within our current system. Using multi-spectral transmission data from the near infrared range, novel algorithms will be used to calculate tomographic tissue attenuation to be used as prior knowledge for BLT. These will be evaluated using phantoms and real models and improvements in accuracy will be quantified.