| Summary: | Optical topography has recently evolved into a widespread research tool for non-invasively mapping blood flow and oxygenation changes in the adult and infant cortex. The work described in this thesis has focused on assessing the potential and limitations of this imaging technique, and developing means of obtaining images which are less artefactual and more quantitatively accurate. Due to the diffusive nature of biological tissue, the image reconstruction is an ill-posed problem, and typically under-determined, due to the limited number of optodes (sources and detectors). The problem must be regularised in order to provide meaningful solutions, and requires a regularisation parameter (\lambda), which has a large influence on the image quality. This work has focused on three-dimensional (3D) linear reconstruction using zero-order Tikhonov regularisation and analysis of different methods to select the regularisation parameter. The methods are summarised and applied to simulated data (deblurring problem) and experimental data obtained with the University College London (UCL) optical topography system. This thesis explores means of optimising the reconstruction algorithm to increase imaging performance by using spatially variant regularisation. The sensitivity and quantitative accuracy of the method is investigated using measurements on tissue-equivalent phantoms. Our optical topography system is based on continuous-wave (CW) measurements, and conventional image reconstruction methods cannot provide unique solutions, i.e., cannot separate tissue absorption and scattering simultaneously. Improved separation between absorption and scattering and between the contributions of different chromophores can be obtained by using multispectral image reconstruction. A method is proposed to select the optimal wavelength for optical topography based on the multispectral method that involves determining which wavelengths have overlapping sensitivities. Finally, we assess and validate the new three-dimensional imaging tools using in vivo measurements of evoked response in the infant brain.
|
|---|
