Experimental and theoretical investigations of near infrared tomographic imaging methods and clinical applications

This thesis describes investigations performed using a time-resolved optical tomographic medical imaging system. Optical tomography involves transilluminating a volume of tissue (e.g. an infant's head or an adult breast) using near infrared (NIR) light. The differing absorption characteristics of oxy- and deoxy-haemoglobin in this wavelength range can then be exploited to yield information about spatial variations in oxygenation and blood volume deep within the tissue. The distinctive scatter properties of tissues such as tumors, gray matter and white matter in the NIR range also provide diagnostic information. A 32 channel time-correlated single photon counting instrument (MONSTIR), designed and built at University College London (UCL), is at first evaluated, and then used to perform a number of investigations into the capabilities and limitations of optical tomographic imaging. Simulations and image reconstructions are performed using the TOAST algorithm, developed at the UCL Department of Computer Science. The progression of the imaging system from laboratory prototype to clinical instrument is detailed. Specifically, studies are presented demonstrating techniques for calibration of time- resolved data, along with validation of the TOAST forward model. Early phantom imaging trials are described with reference to problems associated with using 2D and 3D algorithms to reconstruct measured data. An investigation into the effects of static background structure on images of changes in properties is presented. Methods for derivation of functional parameters from time-resolved optical tomography are explored and demonstrated on clinical data acquired on the breast, forearm and neonatal head. An experimental study of the effects of non-scattering regions within diffusing media is presented, highlighting issues relating to refractive index mismatches. Finally, practical aspects of clinical data acquisition are addressed. These investigations predominantly refer to general optical tomography issues rather than specifically relating to MONSTIR or TOAST. Findings and conclusions should be relevant to the wider biomedical optics community