Search

Radiotherapy represents an important phase of treatment for a large number of cancer patients. It is essential that resources used to deliver this treatment are used efficiently. This thesis approaches the problem of scheduling treatments in a radiotherapy centre. Data about the daily intake of patients are collected and analysed. Several approaches are presented to create a schedule every day. The first presented are constructive approaches, developed due to their simplicity and low computational requirements. The approaches vary the preferred treatment start, machine utilisation reservation levels, and the frequency and number of days in advance with which schedules are created. An Integer Linear Programming (ILP) model is also presented for the problem and used in combination with approaches similar to the ones above. A generalisation of the constructive utilisation threshold approach is developed in order to vary the threshold level for each day according to how far it is from the current day. In addition, the model is evaluated for different sizes of the problem by increasing the rate of patient arrivals per day and the number of machines available. Different machine allocation policies are also evaluated. An exact method is introduced for finding a set of solutions representing the whole Pareto frontier for integer programming problems. It is combined with two robust approaches: the first considers known patients before they are ready to be scheduled, while the second considers sets of predicted patients who might arrive in the near future. A rescheduling approach is also suggested and implemented. A comparison is made amongst the best results from each group of approaches to identify the advantages and disadvantages of each. The robust approaches are found to be the best alternative of the set.

This thesis describes the design, construction and testing of a dome coil. The dome coil is hemispherical in shape and is intended to be used within a set of hemispherical gradient coils in a seven tesla magnetic resonance imaging magnet. The dome coil has eight independent elements and is designed to be used for parallel transmission and reception. It is shown that the dome coil produces less specific absorption rate than a conventional birdcage coil and is suitable for head imaging. A study of travelling wave magnetic resonance imaging is also presented. In this study two different methods of generating a travelling wave (using a patch antenna and an end-fire helix antenna) are compared, and the power requirements and specific absorption rate of the travelling wave approach are considered. It is concluded that travelling wave magnetic resonance imaging is best performed using a local receive coil and with the travelling wave antenna placed at least 50 cm from the subject. All theory relevant to the design, construction, testing and use of these coils and antennas is also presented.

The microstructure of the lung is complex, containing many branching airways and alveolar sacs for optimal gas exchange. Lung diseases such as cystic fibrosis (CF), asthma, and emphysema lead to a destruction of this microstructure. As such, there is a growing interest in the early identification and assessment of lung disease using non invasive imaging techniques. Pulmonary function tests such as spirometry and plethysmography are currently used for this purpose but can only provide quantitative lung function measurements rather than direct measurements of lung physiology and disease. Computed tomography (CT) has also been used but due to risk of cell damage and mutation from the ionising radiation, long term monitoring of the lungs is severely constrained. Recently, new methods based on magnetic resonance imaging (MRI) have been developed to provide diagnostic imaging of the lung. Conventional MRI is not very well suited for lung imaging due to the very low proton density of the pulmonary airspaces. This problem can be overcome by making the patient inspire noble gases such as 3He whose polarisations have been vastly increased through optical pumping. Therefore 3He MRI permits a non-invasive determination of lung function. The high diffusion coefficient of 3He can be exploited to probe the microstructure of the lung. By measuring how fast 3He diffuses within the lung, the size of the lung microstructure can be assessed. Normally, the airspace walls impede the diffusion of the gas but for diseased lungs where microstructure has been destroyed, diffusion is less restricted and a higher apparent diffusion coefficient (ADC) is observed. The research conducted for this thesis focused on the measurement of ADC using three different MRI pulse sequences with each sequence being designed to assess the peripheral airspaces over different length scales. These sequences were then implemented on three different subject study groups.

Laser Doppler blood flow imaging is well established as a tool for clinical research. The technique has considerable potential as an aid to diagnosis and as a treatment aid in a number of situations. However, to make widespread clinical use of a blood flow imager feasible a number of refinements are required to make the device easy to use, accurate and safe. Existing LDBF systems consist of 2D imaging systems, and single point scanning systems. 2D imaging systems can offer fast image acquisition time, and hence high frame rate. However, these require high laser power to illuminate the entire target area with sufficient power. Single point scanning systems allow lower laser power to be used, but building up an image of flow in skin requires mechanical scanning of the laser, which results in a high image acquisition time, making the system awkward to use. A new approach developed here involves scanning a line along a target, and imaging the line with a 1D sensor array. This means that only one axis of mechanical scanning is required, reducing the scanning speed, and the laser power is vastly reduced from that required for a 2D system. This approach lends itself well to the use of integrated CMOS detectors, as the smaller pixel number means that a linear sensor array can be implemented on an IC which has integrated processing while keeping overall IC size, and hence cost, lower than equivalent 2D imaging systems. A number of front-end and processing circuits are investigated in terms of their suitability for this application. This is done by simulating a range of possible designs, including several logarithmic pixels, active pixel sensors and opamp-based linear front-ends. Where possible previously fabricated ICs using similar sensors were tested in a laser Doppler flowmetry system to verify simulation results. A first prototype IC (known as BVIPS1) implements a 64x1 array of buffered logarithmic pixels, chosen for their combination of sufficient gain and bandwidth and compact size. The IC makes use of the space available to include two front-end circuits per pixel, allowing other circuits to be prototyped. This allows a linear front-end based on opamps to be tested. It is found that both designs can detect changes in blood flow despite significant discrepancies between simulated and measured IC performance. However, the signal-noise ratio for flux readings is high, and the logarithmic pixel array suffers from high fixed pattern noise, and noise and distortion that makes vein location impossible. A second prototype IC (BVIPS2) consists of dual 64x1 arrays, and integrated processing. The sensor arrays are a logarithmic array, which addresses the problems of the first IC and uses alternative, individually selectable front-ends for each pixel to reduce fixed-pattern noise, and an array of opamp-based linear detectors. Simulation and initial testing is performed to show that this design operates as intended, and partially overcomes the problems found on the previous IC - the IC shows reduced fixed pattern noise and better spatial detection of blood flow changes, although there is still significant noise.

The increased contrast to noise ratio available at Ultrahigh (7T) Magnetic Resonance Imaging (MRI) allows mapping in space and time the brain's response to single trial events with functional MRI (fMRI) based on the Blood Oxygenation Level Dependent (BOLD) contrast. This thesis primarily concerns with the development of techniques to detect and characterize single trial event-related BOLD responses without prior paradigm information, Paradigm Free Mapping, and assess variations in BOLD sensitivity across brain regions at high field fMRI. Based on a linear haemodynamic response model, Paradigm Free Mapping (PFM) techniques rely on the deconvolution of the neuronal-related signal driving the BOLD effect using regularized least squares estimators. The first approach, named PFM, builds on the ridge regression estimator and spatio-temporal t-statistics to detect statistically significant changes in the deconvolved fMRI signal. The second method, Sparse PFM, benefits from subset selection features of the LASSO and Dantzig Selector estimators that automatically detect the single trial BOLD responses by promoting a sparse deconvolution of the signal. The third technique, Multicomponent PFM, exploits further the benefits of sparse estimation to decompose the fMRI signal into a haemodynamical component and a baseline component using the morphological component analysis algorithm. These techniques were evaluated in simulations and experimental fMRI datasets, and the results were compared with well-established fMRI analysis methods. In particular, the methods developed here enabled the detection of single trial BOLD responses to visually-cued and self-paced finger tapping responses without prior information of the events. The potential application of Sparse PFM to identify interictal discharges in idiopathic generalized epilepsy was also investigated. Furthermore, Multicomponent PFM allowed us to extract cardiac and respiratory fluctuations of the signal without the need of physiological monitoring. To sum up, this work demonstrates the feasibility to do single trial fMRI analysis without prior stimulus or physiological information using PFM techniques.

The growing worldwide prevalence of cardiovascular diseases (CVD) such as chronic heart failure (CHF) highlights the need for an effective treatment method. Annually, there are an estimated 17.5 million deaths caused by cardiovascular diseases worldwide, representing 30% of all global deaths; of these deaths approximately 50% are due to CHF [1] and about 80% occur in low- and middle-income countries [2]. If current trends are allowed to continue, by 2015 an estimated 20 million people annually will die from CVD [3]. Each year, only about 3,000 people receive a heart transplant as the only current definitive long-term treatment for end-stage CHF. To compound the severity of the situation, organ donations are decreasing (4). Implantable blood pumps offer an effective treatment to CHF, either as a bridge to transplantation / recovery, or as destination therapy i.e. use of long term mechanical circulatory support in patients with end-stage heart failure without the intention of eventual heart transplantation. With the number of sufferers of CHF rising in both the developed and developing world it becomes pertinent to design a cost effective device. It is the objective of this work to investigate the proposal of a new cost-effective Centrifugal Rotary Blood Pump (CRBP), which employs previously unutilized design methodology. Through the replacement of those complex, custom components seen within existing CRBPs with standard off-the-shelf components, and the implementation of high-throughput manufacturing processes, such as injection moulding, a reduction in component parts allows for a reduced profit margin and hence a reduced total cost of device. It is proposed by the author that the current production cost of LVAD devices may be reduced by up to 95%. The work presented in this thesis identifies the principal difference between current pump designs; this is their bearing system. It is proposed here to form a new classification of bearing type that combines a passive magnetic bearing and a hydrodynamic bearing such that the relative potential merits of both systems may be exploited. Through the amalgamation of established design techniques with other more modern design practices a rigorous, adaptive design tool has been produced that CRBP designers may use to quickly obtain full impeller and volute geometry from few input parameters. The geometry output provides a platform from which a new conceptual Left Ventricular Assist Device (LVAD) has been envisaged. Through experimental and computational analysis of the magnetic coupling, this investigation has shown that it is possible to integrate the magnetic bearing and the drive system into one component of design; it is possible to design a magnetic coupling that not only acts as the drive system for a CRBP, but as a bearing system that offers both axial and radial bearing forces. A spiral groove bearing (5GB) has been implemented as the hydrodynamic bearing as part of the hybrid system. Experimental investigation has shown the spiral groove bearing to be anti traumatic, which may be attributed to the short residence time of blood in the bearing. However, a reduction in the anticipated load capacity shows that the bearings are operating on a reduced viscosity; this is an indication of cell exclusion within the 5GB. Comparisons to aqueous glycerol tests of known viscosities have shown that the blood bearing is operating on a viscosity close to that of plasma. It is suggested that a "shear front differential" is the mechanism behind cell exclusion, in which RBCs migrate away from areas of high shear stress into areas of relatively low shear stress. This investigation has demonstrated the suitability of the hybrid magnetic / hydrodynamic bearing for use in a new CRBP. It has been shown that the electromagnetic drive system intended for implementation in to this CRBP can be used as an effective passive magnetic bearing. It is intended that the axial and radial bearing forces produced by the drive system are balanced by a conical spiral groove thrust bearing. The incorporation of the hydrodynamic bearing into the magnetic bearing transforms the previously unstable passive magnetic bearing to a stable hybrid bearing. The stability of the system has been predicted through numerical analysis of the stiffness matrix and through satisfaction of the stability criteria. The natural frequencies of the system have been calculated; these are shown to be sufficiently different from any excitation frequencies identified that may cause the system to behave in an unstable fashion at the operational speed of the pump. The main point to be realized from the analysis of the hybrid bearing system, however, is that the proposed set-up of the hybrid bearing is not feasible due to the effect of cell exclusion, which causes the SGB to operate on a reduced viscosity. The reduction in viscosity reduces the load capacity of the SGB; the magnetic preload on the impeller cannot be balanced by the chosen SGB geometry. Recommendations have been made as to the design parameters that may be altered such that the design intent of the proposed system may be realised. Future work must concentrate on the realization of that design intent through the manufacture of a prototype, which can provide a proof-of-concept for the proposed system. The work presented here provides a feasibility study for the novel hybrid bearing / drive system and provides sound foundation upon which a prototype may be based.

This thesis investigates the normal appearing brain tissue in multiple sclerosis (MS) using high resolution quantitative MRI measures acquired at high magnetic field. The use of magnetisation transfer ratio (MTR) and longitudinal time (T1) were employed to investigate changes in normal appearing white matter (NAWM) for healthy control subjects and patients with clinically isolated syndromes suggestive of MS (CIS) and relapsing-remitting multiple sclerosis (RRMS). The results showed a significant difference in the median peak position, full width at half maximum, the 25th percentile of the MTR histograms and the 75th of the T1 histograms. The magnetic susceptibility mapping technique was used to quantitatively investigate the accumulation of iron in deep grey matter structures in healthy controls, CIS and RRMS patients. The results showed an increase in iron deposition in the caudate nucleus, putamen, globus pallidus structure in CIS and RRMS when compared to those in healthy controls. Combining functional MRI and magnetic susceptibility mapping was used to elucidate the mechanisms underlying the visual activation in lateral geniculate nucleus (LGN) in healthy controls. The results showed an increase in the accuracy of the LGN delineation. This is in turn, highlights the importance of this method in quantifying the visual disturbances associated with MS and CIS patients.

Over the last 20 years, an increasing number of research studies have shown that ultrasonography can provide a valid and accurate assessment of the median nerve and the pathological changes associated with median nerve disorders. More recently ultrasonographic technology has advanced and it now allows for dynamic imaging of the nerve during physical movement of the hand. However dynamic ultrasonographic imaging is still a relatively new application as is yet to be explored to its full potential in the study of median nerve dynamics and the associated pathological changes. The primary aim of this research is to study median nerve activity during physical hand movement using ultrasonography with 8-16MHz linear array transducer. The objectives are to establish an ultrasonographic technique for the dynamic imaging of the median nerve, along with an analytical method that can quantify its changing size, shape and location during four hand movements. The data obtained from participants with normal median nerve function is compared to values from participants who are either moderately symptomatic or fully symptomatic of a median nerve disorder, to investigate whether the novel method and technique could distinguish between the groups. Three studies were designed to investigate median nerve activity during physical movement using ultrasonography. The aim of the first study was to investigate if the novel quantification method and ultrasonographic technique could measure the changing parameters of the median nerve between the start and end positions of a finger grip, power grip, pinch grip and thumb opposition movement, in non-symptomatic participants who have normal median nerve function. The second study investigated whether the novel quantification method and ultrasonographic technique could distinguish between the non-symptomatic participants and participants who are either moderately symptomatic or fully symptomatic of a median nerve disorder, during each movement. The third study examined the change in size, shape and location of the median nerve at five intervals during the four movements and compared the data obtained from the non-symptomatic group with the values obtained from the moderate and fully symptomatic groups. The findings from the studies show that the dynamic ultrasonographic technique can be used to image the median nerve during physical activity of the hand and that the novel quantification method can be used to measure the change in the parameters of the median nerve during physical hand activity and distinguish significant differences between the non-symptomatic median nerve and those symptomatic of an median nerve disorder.

Diffusion-weighted (DW) magnetic resonance imaging allows the quantification of water diffusion within tissue. Due to the hindrance of water molecules by the various tissue compartments, probing for the diffusive properties of a region can provide information on the underlying structure. This is particularly useful for the human brain, whose anatomy is complex. Diffusion imaging provides currently the only tool to study the brain connectivity and organization non-invasively and in-vivo, through a group of methods, commonly referred to as tractography methods. This thesis is concerned with brain anatomical connectivity and tractography. The goal is to elucidate problems with existing approaches used to process DW images and propose solutions and methods through new frameworks. These concern data from two popular DW imaging protocols, diffusion tensor imaging (DTI) and high angular resolution diffusion imaging (HARDI), or Q-ball imaging in particular. One of the problems tackled is resolving crossing fibre configurations, a major concern in DW imaging, using data that can be routinely acquired in a clinical setting. The physical constraint of spatial continuity of the diffusion environment is imposed throughout the brain volume, using a multi-tensor model and a regularization method. The new approach is shown to improve tractography results through crossing regions. Quantitative tractography algorithms are also proposed that, apart from reconstructing the white matter tracts, assign relative indices of anatomical connectivity to all regions. A fuzzy algorithm is presented for assessing orientational coherence of neuronal tracts, reflecting the fuzzy nature of medical images. As shown for different tracts, where a-priori anatomical knowledge exists, regions that are coherently connected and possibly belong to the same tract can be differentiated from the background. In a different framework, elements of graph theory are used to develop a new tractography algorithm that can utilize information from multiple image modalities to assess brain connectivity. Both algorithms inherently consider crossing fibre information and are shown to solve problems that affect existing methods.

The increased signal-to-noise ratio available at high magnetic field makes possible the acquisition of clinically useful MR images either at higher resolution or for quantitative methods. The work in this thesis is focused on the development of quantitative imaging methods used to overcome difficulties due to high field MRI systems (> 3T). The protocols developed and presented here have been tested on various studies aiming at discriminating tissues based on their NMR properties. The quantities of interest in this thesis are the longitudinal relaxation time T1, as well as the magnetization transfer process, particularly the chemical exchange phenomenon involving amide protons which is highlighted particularly well at 7T under specific conditions. Both quantities (T1 and amide proton transfer) are related to the underlying structure of the tissues in-vivo, especially inside the white matter of the brain. While a standard weighted image at high resolution can provide indices of the extent of the pathology, a robust measure of the NMR properties of brain tissues can detect earlier abnormalities. A method based on a 3D Turbo FLASH readout and measuring reliably the T1 in-vivo for clinical studies at 7T is first presented. The other major part of this thesis presents magnetization transfer and chemical exchange phenomena. First a quantitative method is investigated at 7T, leading to a new model for exchange as well as contrast optimization possibility for imaging. Results using those methods are presented and applied in clinical setting, the main focus being to image reliably the brain of both healthy subjects and Multiple Sclerosis patients to look at myelin structures.

Looking inside the lungs without the danger of ionizing radiation side effects became available with magnetic resonance imaging using hyperpolarised noble gases. This technique has the potential to become a real tool for assessing in vivo ventilation, perfusion and even lung microstructure. The work covered in this research was aimed to improve the existing method for 3He polarisation and open the possibility to develop new modalities to probe the lung microstructure that could then be used in clinical trials. For this purpose, the polarisation facility was remodeled and new components were added. The rest of the work was focused on developing diffusion techniques that are more appropriate for the assessment of lung diseases. The improvement of the 3He polarisation facility consisted in the optimization of the gas flow path, implementation of a new dispensing method and new controlling protocol. The capacity of the polarisation system was increased by using a more powerful laser. The outcome of this was an increase in polarisation rate and a significant reduction of the dispensing time. Altogether this allow for clinical studies to be performed without too much delay. A clinical study aimed to distinguish differences between children born at term and premature was started on 70 volunteers. Three methods for measuring diffusion were used : spin echo diffusion weighted method, SPAMM tagging and MR diffusion spectroscopy. The first was previously used in the group and the last two were developed during this research. The results were correlated with basic pulmonary functional tests(spirometry and plethysmography) and also with the multiple breaths nitrogen wash-out results. No differences were found in the two groups. The results don’t agree with the current theories on lung growth and suggest that alveolarisation occurs even after the age of 8, possibly up to adult age. This is very important to be investigated further due to its clinical importance.

The signal-to-noise ratio available in Magnetic Resonance Imaging (MRI)is determined by the static magnetic field strength, causing a continued drive toward higher fields to enable faster image acquisition at finer spatial resolution. The work in this thesis is primarily concerned with the development of sequences for Ultra High Field Magnetic Resonance Imaging (7T) which allow the acquisition of images with high spatial resolution for study of the structure and function of the brain. The methods developed here for high spatial resolution structural imaging allow the identification of regions of the cortex which exhibit layers of high myelin concentration within the cortical strip. This permits the investigation of the correspondence of functional regions in the visual cortex to their underlying structure 'in vivo'. A robust methodology for high resolution functional mapping over a restricted field of view is presented and the results of fMRI studies demonstrating 1 mm isotropic resolution in the primary somatosensory cortex S1 using this methodology are shown. BOLD responses to vibrotactile digit stimulation were investigated using a travelling wave paradigm to measure the topographic representation of the digits in S1 and an event related paradigm for characterization of the haemodynamic delay. A spin-echo EPI acquisition has been optimized and tested to compare the BOLD response in GE and SE echo planar images by employing visual and motor tasks. The specificity of the BOLD responses of SE and GE data was found to be similar using a travelling wave paradigm.

Non-invasively characterizing the polarization properties of tissues has potential for in vivo clinical applications such as monitoring the healing state of wounds and burns. Conventional methods, which measure the polarization difference of a tissue and usually involve a co-polarized detection, are either restricted by a sample's surface reflections or can only be performed in off-axial systems with the use of matching fluid and a glass plate applied to the sample's surface. In this thesis a new technique called rotating orthogonal polarization imaging (ROPI) is presented. The technique involves illumination in a single polarization state and detection in the orthogonal polarization state in a coaxial system. Synchronously rotating both the illumination and orthogonal detection states can provide a polarization difference image that is free from surface reflections and sensitive to the polarization property of a target embedded within a scattering medium. The basic theoretical principle of the technique is demonstrated using Mueller calculus. Monte Carlo (MC) simulations are used to simulate the ideal performance when detecting a polarizing target embedded in a scattering medium at different depths. Tissue phantom experiments also investigate the imaging of a polarizing target embedded within a scattering medium at different depths using manual rotation of polarisers and liquid crystal tunable filters. ROPI is far more effective than conventional polarization difference imaging due to the significant reduction in surface reflections. The technique is sensitive to polarizing targets embedded at depths of up to 17 mean free paths within the medium. Accurate quantification of dichroism is difficult due to scattering that occurs between the target and the exit surface of the medium. In addition, preliminary measurements of the linear dichroism of different real tissues (bovine tendon, lamb tendon, chicken breast and human skin) and a tissue engineered tendon orientated at two orthogonal directions are demonstrated. It should be noted that the majority of simulation and tissue phantom results that are presented in this thesis are for performing ROPI for polarizing targets embedded in a scattering solution. Much further work is required before the technique can be taken into clinical practice. Issues such as image alignment and the inversion of the data to produce clinically useful images still need to be addressed.

The aim of this thesis was to test the hypothesis that the detection of plaque haemorrhage by Magnetic Resonance Imaging (MRI PH+) is a clinical valid biomarker of unstable carotid disease in patients with symptomatic high grade carotid disease. This hypothesis was tested by a series of studies in patients with symptomatic high grade stenosis including a comparison of MRI PH+ to histological assessment of plaque instability, and an examination of the relationships between MRI PH+ and cerebral ischaemia as assessed by microembolisation, cerebral white matter hyperintense lesions (WMHL), diffusion weighted cerebral imaging (DWI) and clinical assessment. MRI PH+ carotid plaques were associated with features of active plaque disease, including inflammation. The senstiviity and specificity of MRI PH+ to detect spontaneous and intra-operative microembolic signals (detected by transcranial Doppler) was high (85 and 83%) and moderate (51 and 54%), respectively. Cerebral WMHL were found to be more prevalent (mean number [S.D.]: 3.3[3.3] vs. 2.1[2.5]) and larger (mean volume [S.D.]: 10.3[9.3] vs. 7.0[6.6] ml) in hemispheres ipsilateral to MRI PH+ rather than MRI PH- carotid plaques. MRI PH+ increased the risk of developing cerebral DWI ischaemic lesions (odds ratio 6.2; 95% C.I. 1.7-21.8) and were associated with multiple lesions of different ages. MRI PH + symptomatic carotid plaques increased the short-term risk of ipsilateral ischaemic neurological events with a hazard ratio of 5.9 (95% C.I. 1.4- 25.6); but not in the contralateral, asymptomatic side. This work has demonstrated that MRI PH is a potential valid biomarker of plaque instability in patients with symptomatic carotid stenosis. Further studies are required to se if risk stratification can be improved using other plaque features and improved localization of plaque haemorrhage.

Hyperpolarised 3He has been used extensively in the neutron community and for magnetic resonance imaging (MRI) of the lungs. 129Xe has many desirable advantages compared to 3He for NMR, namely its chemical sensitivity, solubility, adsorption capability and atmospheric abundance. In this work we present the design and construction of a 129Xe spin exchange polariser that operates at gas pressures between 0.6 and 3 bar. The design incorporates a novel Volume Holographic Grating (VHG) spectrally narrowed, high power laser diode device as the optical pump source for a continuous gas flow cylindrical polarisation cell. Nuclear polarisations of 34 +/- 3 % are achieved in 129Xe gas. Magnetisation production factors in excess of 80 % cm^3 min^-1 are achieved using a pseudo batch-flow operation method. Major parameters affecting 129Xe gas polarisation are discussed and the first measurements of cell heating from laser light absorption by Rb vapour are made. Hyperpolarised 129Xe gas produced from our system is imaged inside a low field, whole body MRI scanner. NMR measurements from inside a human lung are made using hyperpolarised 129Xe. Preliminary work on dissolved solution NMR is presented as are studies on the feasibility of using hyperpolarised 129Xe for porosity characterisation of porous media in low magnetic fields.

The measurement of oxygen saturation SO2 is one of the vital signs relied on by the medical profession. Pulse oximeters are widely used in many branches of medicine; and are the most widely used method of assessing oxygen saturation. However they can only be applied to an extremity (usually a finger or toe), need calibration, and are known to be inaccurate under certain conditions. The object of this research was to develop an oximeter, that does not require a pulsatile signal, (and so can be used anywhere on the body); can be used in either transmission of reflective mode; does not require calibration; and does not suffer from the known problems of pulse oximeters. The instrument must work with reflected light, and so the first step was to develop a Monte Carlo simulation of the Attenuation spectra, for visible light, from a scattering media (tissue). A mathematical model of the attenuation surface had then to be found, and its effect on the absorbtion spectra of oxyhemoglobin HbO2 and de-oxyhemoglobin Hb understood. Then the oxygen saturation the ratio of HbO2 to total haemoglobin could be recovered. Methods of computing oxygen saturation from the raw reflectance spectra were devised and then tested with single reflection spectra, the results indicate that a low cost instrument could be developed. The technique was applied to images from a hyper-spectral camera, this instrument takes a full spectrum at each pixel of an image, and enabled an oxygen saturation map for large areas of the body to be produced. The technique is being used with AstraZenca Ltd as a bio marker skin for irritation studies.

Laser Doppler Blood Flowmetry is a non-invasive technique that has been I developed and used for measuring microvascular blood flow in tissue. The technique utilises the backscattering of the laser light from moving red blood cells and static tissue in order to extract information such as the concentration and flow. This thesis describes the early stages of the development of an integrated optical sensor array to perform full field laser Doppler blood flow imaging. This technique will eliminate the need for mechanical scanning and the data bottleneck that exists between the photodiode array and processing unit, so allowing the direct measurement ofblood flow maps to be obtained in real time. A single channel laser Doppler blood flowmetry device has been implemented using a photodetector linked to a field programmable gate array. Filters (low pass, band pass and frequency weighted) have been developed for processing the Doppler signals to obtain flow and concentration measurements. The responses of these filters are demonstrated using measurements from modulated light signals, a rotating diffusing disc and in vivo measurements of blood flow. Several types of a linear array system and current to voltage converter are considered for the first fabrication run of the project based on the cost of fabrication and performance of the system such as operating frequency, gain, bandwidth and signal to noise ratio. A 16x1 linear array of photodiodes is developed and integrated on the same chip with the laser Doppler processing design, successfully implemented in the single channel laser Doppler system, using the standard 0.35Jlm CMOS technology. The characterisation of each individual part of the design was carried out and compared with the Cadence simulation results. The performance of the system on a single pixel is also evaluated using a modulated laser as a light source. The knowledge gained from the characterisation and the overall performance of the linear array system is then used to develop a full field laser Doppler blood flow camera.

This thesis describes the development and application of quantitative echo planar magnetic resonance imaging techniques to the study of human placental development in normal and compromised pregnancies. Initially, a method of rapidly and accurately measuring the transverse relaxation time is proposed using a multi-echo measurement sequence. The method is described, validated on CUS04 phantoms and applied in the study of the human placenta and gastric dilution. It is shown that the inversion provided by sinc pulsesis insufficient to generate an accurate measurement but using adiabatic refocusing pulses yields a measurement that is comparable with a single spin echo. Subsequently, a rapid magnetisation transfer method is presented that allows the quantification of the relative size of the bound proton pool. An experimental pulse sequence is proposed, along with a theoretical model, that permits the investigation of the bound proton pool's transition towards the steady state. The sequence and model are validated using agar gel phantoms and shown to agree well with literature values. When applied in the study of the human placenta, it is shown that there is no significant variation in the fitted value of the bound proton pool size with increasing gestational age or in compromised pregnancies. Finally, several methods of measuring the oxygenation level of blood within the human placenta are investigated. The signal intensities of cardiac gated T~• and T~ weighted images acquired at various points in the maternal cardiac cycle are explored but no significant variation is shown through the cycle. A pulsed gradient spin echo sequence that utilises anti-symmetric sensitising gradients is validated and then applied in the human placenta. Oxygenation measurements with this technique are shown to be unfeasible but the potential of the sequence to monitor blood flow in the placenta is demonstrated.

Functional Magnetic Resonance Imaging (fMRI) is one of the most popular neuroimaging methods for investigating the activity of the human brain during cognitive tasks. As with many other neuroiroaging tools, the group analysis of fMRI data often requires a transformation of the individual datasets to a common stereotaxic space, where the different brains have a similar global shape and size. However, the local inaccuracy of this procedure gives rise to a series of issues including a lack of true anatomical correspondence and a loss of subject specific activations. Inter-subject parcellation of fMRI data has been proposed as a means to alleviate these problems. Within this frame, the inter-subject correspondence is achieved by isolating homologous functional parcels across individuals, rather than by matching voxels coordinates within a stereotaxic space. However, the large majority of parcellation methods still suffer from a number of shortcomings owing to their dependence on a general linear model. Indeed, for all its appeal, a GLM-based parcellation approach introduces its own biases in the form of a priori knowledge about such matters as the shape of the Hemodynamic Response Function (HRF) and taskrelated signal changes. In this thesis, we propose a model-free data-driven parcellation approach to singleand multi-subject parcellation. By modelling brain activation without an relying on an a priori model, parcellation is optimized for each individual subject. In order to establish correspondences of parcels across different subjects, we cast this problem as a multipartite graph partitioning task. Parcels are considered as the vertices of a weighted complete multipartite graph. Cross subject parcel matching becomes equivalent to partitioning this graph into disjoint cliques with one and only one parcel from each subject in each clique. In order to solve this NP-hard problem, we present three methods: the OBSA algorithm, a method with quadratic programming and an intuitive approach. We also introduce two quantitative measures of the quality of parcellation results. We apply our framework to two fMRI data sets and show that both our single- and multi-subject parcellation techniques rival or outperform model-based methods in terms of parcellation accuracy.