Single photon emission transverse section tomography maps the activity concentration of selected radioactive tracers in a thin slice of an object. In the human body this allows the monitoring of specific physiological functions and has considerable, but hitherto unrealized, quantitative potential. The mathematical foundations of the techniques which permit quantitative measurement are presented and the requirements of the theory established together with their implications for practical systems. In practice not all the requirements can be met so the deviations from the theory and their effect on quantitivity are discussed. The two most serious problems, collimation and attenuation, are considered in more detail. In particular, the design of a new collimator combination for the Aberdeen Section Scanner is presented together with experimental results showing the improved depth independence of response and resultant increase in quantitative accuracy achieved. Methods of compensation for attenuation are reviewed and their effectiveness discussed. The iterative convolution procedure was selected to be most promising for implementation on systems without large computer resources and the high degree of accuracy achieved with this technique is demonstrated on a wide range of source distributions. An investigation of the use of statistical cluster analysis on tomographic images was carried out. The results of application of a conditioned ratio method for analysing spatial patterns to liver tomograms are presented, showing the excellent potential of the technique for improving the diagnostic accuracy of the final images. Its use is recommended for extracting extra information from difficult tomograms where visual inspection alone cannot give satisfactory differential diagnosis.

Attenuation correction by measurement of the transmission properties of the field of view with and without the object has some drawbacks. It will take extra time to produce transmission scans, which then can not be used for processing more patients and may be uncomfortable for the patient being scanned. The search is on then for a replacement for transmission scanning that may be used in clinical practice and many suggestions have been put forward for this purpose. Attenuation correction of this sort must be as accurate as the measured attenuation correction but without placing time dependencies on the scanner. We introduce many of these suggestions in this thesis and then move on to the theory behind the method that we have tested for this purpose. Our aim is to remove these artifacts from images and reproduce as accurately as possible the emission image without attenuation. This is important for image interpretation and quantitative analysis of images. The method developed in this thesis revolves around the consistency conditions of the Radon transform. We test an algorithm based on these consistency conditions for its ability to approximate attenuation distributions in conjunction with a parameterised model of the attenuation and a search algorithm. The model has been a uniform ellipse, an elliptical cylinder, and finally an out of position attenuation distribution. In each case we have found solutions that do not correspond to better PET images mainly because of the unpredictability in the solutions found. However these solutions are limited by the amount of information given to the search, increasing things like the similarity of the model and the actual attenuation, reducing the amounts of transformation allowed increase the chances of a predictable useful solution being found by this method. Finally we remark on the usefulness of the current technique and its dependence on the acuity of the model plus the possibility of its use in future PET scanning.

No description available.

This thesis presents the analysis of low angle X-ray scatter measurements taken with an energy dispersive system for substance identification, imaging and system control. Diffraction measurements were made on illicit drugs, which have pseudo- crystalline structures and thus produce diffraction patterns comprising a se ries of sharp peaks. Though the diffraction profiles of each drug are visually characteristic, automated detection systems require a substance identification algorithm, and multivariate analysis was selected as suitable. The software was trained with measured diffraction data from 60 samples covering 7 illicit drugs and 5 common cutting agents, collected with a range of statistical qual ities and used to predict the content of 7 unknown samples. In all cases the constituents were identified correctly and the contents predicted to within 15%. Soft tissues exhibit broad peaks in their diffraction patterns. Diffraction data were collected from formalin fixed breast tissue samples and used to gen erate images. Maximum contrast between healthy and suspicious regions was achieved using momentum transfer windows 1.04-1.10 and 1.84-1.90 nm_1. The resulting images had an average contrast of 24.6% and 38.9% compared to the corresponding transmission X-ray images (18.3%). The data was used to simulate the feedback for an adaptive imaging system and the ratio of the aforementioned momentum transfer regions found to be an excellent pa rameter. Investigation into the effects of formalin fixation on human breast tissue and animal tissue equivalents indicated that fixation in standard 10% buffered formalin does not alter the diffraction profiles of tissue in the mo mentum transfer regions examined, though 100% unbuffered formalin affects the profile of porcine muscle tissue (a substitute for glandular and tumourous tissue), though fat is unaffected.

The aim of the work described in this thesis was to investigate the value of conventional, perfusion- and diffusion-weighted magnetic resonance imaging (MRI) in patients with histology-proven low-grade gliomas (LGG), and the potential role of these methods in the management of patients with these brain tumours. Thirty-six patients were studied at the National Hospital for Neurology and Neurosurgery using conventional, perfusion-weighted and diffusion-weighted MRI at study entry and 6 monthly intervals thereafter. At each visit, tumour volume, maximum rCBV and ADC histogram measures were calculated. This is a unique cohort, as patients were treatment free until malignant transformation was diagnosed, which translates the natural history of these brain tumours. It is unlikely to find such a specific cohort as most of the patients receive treatment after the initial diagnosis of low grade gliomas. Chapters 1 and 2 of this thesis describe the theoretical basis of the MRI techniques used, and summarise the natural history and imaging aspects of cerebral gliomas. Chapter 3 describes a methodological study relating to tumour perfusion measurement: since the inclusion or exclusion of intratumoural vessels may influence the quantification of relative cerebral blood volume (rCBV), a study was conducted to choose the best ROI placement technique to be used for the rCBV measurements included in this thesis. It was shown that only the approach which excluded intratumoural vessels demonstrated a significant association between rCBV values and tumour subtypes (astrocytomas, oligodendrogliomas and oligoastrocytomas) and therefore this technique was used in all subsequent rCBV measurements. In chapter 4, it was investigated whether rCBV is a useful outcome predictor in patients with low-grade gliomas. We found that in LGG susceptibility-weighted MR perfusion imaging not only allows differentiation between histological tumour subtypes but also between two different outcome groups. More importantly, rCBV at study entry can be used to predict time to transformation in these patients. To complement these studies we assessed the utility of rCBV in predicting clinical response in 2 patient groups with low-grade glioma, studied at two separate institutions. The results presented in chapter 5 suggested that glioma rCB V correlates with time to progression or death, independently of institution, despite different approaches to patient management at the two sites. Finally, in chapter 6 we analyzed which quantitative MRI measure (tumour volume, rCBV and ADC) was the best predictor of outcome (stable, progressive disease or death). We found that tumour volume at study entry, at 6 and at 12 months, and its short-term changes were stronger predictors of outcome in patients with low-grade glioma than, and independent of, rCB V, diffusion histogram parameters, age, gender, histology and treatment. Moreover tumour growth was the best predictor of time to transformation and tumour volume was the only independent predictor of time to death. Since this parameter can readily be obtained in any MRI centre and be quantified with an automated method, it may become a useful prognostic factor in clinical practice.

Inflamed areas of atherosclerosis associated with unstable plaque in arteries have been shown to express specific inter-cell adhesion molecules such as ICAM-1. In order to distinguish between areas of stable and unstable plaque an ultrasonic contrast agent has been developed in-house for imaging with high frequency intravascular ultrasound (IVUS). The contrast agent has been imaged with IVUS at 40 MHz at different stages during development. To assess in vivo applicability of the in-house agent it was necessary to image it attached to surfaces and under flow conditions. To image microbubbles at surfaces, work was undertaken on contrast agents at agar-based boundaries. Contrast agent was attached to agar-based material using the avidin and biotin interaction. The attached microbubbles were imaged with high frequency ultrasound, from 7 – 40 MHz. A flow chamber was developed for use with IVUS. The attached microbubbles were imaged under flow conditions. The microbubbles were found to remain echogenic and attached to the agar at a range of flow rates from 75 – 480 ml min^-1 through a flow area of 9 mm². The peak negative acoustic pressures for a selection of high frequency transducers were determined in order to define the ultrasound imaging field. Laser Doppler anemometry (LDA), a non-invasive high resolution technique for measuring flow velocities in liquids and gases was used to determine the flow profile within the flow chamber at the surface of the agar sample. The shear stress on the agar was calculated from the profile. Attached contrast agent was found to remain attached to agar under shear stresses of up to 3.4 Pa compared to a mean in-vivo arterial shear stress of 1.5 Pa. Free flowing in-house agent was shown to attach to prepared agar under low flow rates.

The objective of this dissertation research is to develop new techniques based on Doppler ultrasound to investigate and visualize changes in tissue dynamic and deformation properties due to atherosclerosis in cardiac and vascular applications. A new technique, to correct for the Doppler angle dependency for tissue motion analysis has been developed. It is based on multiple ultrasound beams, and has been validated <i>in vitro</i> to study tissue dynamic properties. It can measure tissue velocity magnitude with low bias (5%) and standard deviation (10%), and tissue velocity orientation with a bias less then 5 degrees and a standard deviation below 5 degrees. A new Doppler based method, called strain rate, has also been developed and validated <i>in vitro</i> for the quantification of regional vessel or myocardial wall deformation. Strain rate is derived from the velocity information and can assess tissue deformation with an accuracy of 5% and a standard deviation less than 10%. Some examples of cardiac strain rate imaging have been gathered and are described in this thesis. Strain rate, as all Doppler based techniques, suffers from angle dependency limitation. A method to estimate one-component strain rate in any direction in the two-dimensional image not necessarily along the ultrasound beam has been developed. The method allows correcting for the strain rate bias along any user-defined direction. <i>In vitro</i> experiments have shown that qualitative two-component strain rate tensor can be derived. Two-component vector velocity from the moving tissue was acquired and two two-component strain rate images were derived. The images showed agreement with the expected deformation pattern.

The aim of this thesis is to explore the evolution and issues affecting the analysis of ischaemic lesions on diffusion – and perfusion-weighted MRI (DWI and PWI) from acute (< 24 h) to chronic times (3 mo) after stroke onset. This thesis includes a review of previous human studies of acute DWI and PWI appearance versus final outcome, a review of the ‘DWI/PWI mismatch’ model (thought to represent the ischaemic penumbra, or ‘tissue at risk’ of infarction), and a systematic review of previous animal studies of the pathophysiology associated with particular lesion. The methodological problems raised by these reviews are addressed in this thesis using a large cohort of stroke patients with serial DWI and PWI. The interrator variability of manual lesion measurements on acute DWI is investigated, and factors affecting this variability are discussed. The effect of lesion oedema (swelling) on measurements of ischaemic lesions on MRI is investigated. This thesis also investigates the tissue state underlying persistent hyperintensity on late DWI, and whether this is just T₂ ‘shine through’, or indicates distinct features in the evolution of the lesion. A novel grid-based analysis method is developed and employed to track serial DWI and PWI changes more effectively, and the effect of observer variability on diffusion and perfusion parameters measured by this method is assessed. Lastly, this thesis discusses the concept of using ‘threshold’ values to predict tissue infarction or survival.

The theme of the thesis is ‘arterial mechanics’ incorporating blood flow, wall dynamics and wall stress. The underpinning aim has been the development and assessment of techniques for use in clinical diagnosis. These techniques include peak systolic velocity within arterial stenosis for estimation of the degree of stenosis; mean velocity for assessment of volumetric flow; wall stress for prediction of disease development and rupture; and elastic modulus for prediction of disease development and rupture. The thesis is divided into 3 themes; ‘development of phantoms’, ‘velocity measurement’ and ‘wall dynamics and stress measurement’. The author developed ‘phantom’ based methods for simulation of flow and wall motion, principally the flow phantom, but also string and electronic injection phantoms. The author evaluated the magnitude and origin of velocity measurement errors in clinical ultrasound systems. The author developed and was involved in clinical evaluation of simplified methods for estimation of wall shear stress and of arterial elasticity. Concluding the thesis as a while; for the measurement of quantities relevant to arterial mechanics for clinical diagnosis, care and attention must be paid in the development of measurement methods with high accuracy, properly validated using an appropriate tissue-mimicking phantom. There is an ongoing clinical problem on velocity measurement for degree of stenosis with no apparent intention to resolve this on the part of manufacturers. Simplified imaging-based measurement methods, of volumetric flow, wall shear stress and arterial elasticity, may have clinical roles where the model assumptions can be justified, which is principally in healthy arteries and early disease. In advanced disease, where there is complex 3D geometry and anisotropy, estimation of flow field data, wall stresses and elasticity should be performed using image guided modelling.

The most common CT dosimetry approaches are standard method, Monte Carlo (MC) method and direct method. In the standard method CTDI is assessed inside a head and a body perspex phantom giving an index of the CT dose efficiency. In the MC method the patient doses are estimated by applying this mathematical technique to simulate the interaction of CT generated radiation with matter inside a mathematical model of phantom. In the direct method, measurement of organ doses is carried out directly inside an anthropomorphic physical phantom. We have developed the direct method by assessing CT axial and longitudinal dose distribution in an anthropomorphic phantom. Measurement of these dose distribution for any scanner allows us to quickly measure organ doses and effective dose, and hence patient doses for any examination protocol with a limited number of TLDs. This work describes the underlying principles and assumption of the direct CT disometry method we have developed and implemented. In addition the review of CT dosimetry provides a comprehensive and critical analysis of the methods developed to date. All dosimetry related concepts for the developed direct technique are considered. Comprehensive dose measurements were carried out to estimate the level of error involved in the developed approach. The MC approach was also used to compare the results of our proposed direct method with this commonly used method in CT dosimetry for routine examinations of some scanners. The proposed direct CT dosimetry method has now been used for several scanner models. The developed dosimetry method has been successfully implemented to assess the radiation dose resulted from various examination protocols using the recent modern CT modalities, that could not easily be investigated by other dosimetry approaches. We believe that the developed direct CT dosimetry approach overcomes many limitations imposed by other common approaches. It also provides a reliable and practical method for the assessment of patient doses from CT practices for a wide range of scan protocols.