Often in neurosurgical planning a dual echo acquisition is performed that yields proton density (PD) and T2-weighted images to evaluate edema near a tumor or lesion. The development of vessel segmentation algorithms for PD images is of general interest since this type of acquisition is widespread and is entirely non-invasive. Whereas vessels are signaled by black blood contrast in such images, extracting them is a challenge because other anatomical structures also yield similar contrasts at their boundaries. / In this thesis we present a novel multi-scale geometric flow for segmenting vasculature from PD images which can also be applied to the easier cases of MR angiography data or Gadolinium enhanced MRI. The key idea is to first apply Frangi's vesselness measure [Frangi et al. (1998)] to find putative centerlines of tubular structures along with their estimated radii. This multi-scale measure is then distributed to create a vector field which is orthogonal to vessel boundaries so that the flux maximizing flow algorithm of Vasilevskiy and Siddiqi (2002) can be applied to recover them. We carry out a qualitative validation of the approach on PD, MR angiography and Gadolinium enhanced MRI volumes and suggest a new way to visualize the segmentations in 2D with masked projections. We also validate the approach quantitatively on a data set consisting of PD, phase contrast (PC) angiography and time of flight (TOF) angiography volumes, all obtained for the same subject. A significant finding is that over 80% of the vasculature recovered in the angiographic data sets is also recovered from the PD volume. Furthermore, over 25% of the vasculature recovered from the PD volume is not detectable in the TOF angiographic data. / Thus, the technique can be used not only to improve upon results obtained from angiographic data but also as an alternative when such data is not available.

An interactive 3D target localisation and delineation tool has been developed for radiosurgery planning of arteriovenous malformations (AVMs). With this system, magnetic resonance (MR), MR angiography (MRA) and computed tomography (CT) volumes can be fused in stereotactic space. Stereotactic angiography (SA) images can be linked to the MRA volume by recovering the SA acquisition geometry. The MRA and SA images can be correlated (1) by ray-tracing through the MRA volume with the recovered SA acquisition geometry and overlaying the images onto the SA images and (2) by localising the AVM onto a volume rendered representation of the MRA with a 3D cursor and projecting its position onto the SA images. Target contours can then be drawn on the MRA/MR/CT images and simultaneously projected onto the SA images. / The plans of patients who had previously undergone radiosurgery at our institution employing SA images for localisation and MR images for delineation were investigated. MRA datasets were also acquired at the time of MR scanning employing the 3D TOF technique. Some ray-traced MRA images correlated well visually with the SA images, others presented inconsistencies which suggest that MRA should be used only as complement to SA images. / The role of the different modalities (M-RA, MR and SA) in the definition of target volumes is investigated by defining the target contours with different combinations of modalities within the interactive system. The target volumes drawn with different modalities were compared to a reference volume, drawn using MRA, MR and SA images, and presented underestimation and overestimation of target volumes ranging from 20% to 92% and from 3% to 40%. / The dosimetric implications of image fusion for target delineation are investigated by retrospective evaluation of the dose coverage of the reference target volume by the original treatment plan. Target coverage inferior to 60% of the reference target volumes by the original treatment plans was obtained for the patients. Treatment plan optimisation was performed to evaluate the possible dosimetric improvements resulting from image fusion for AVM target delineation. The plans were improved by employing three, eight and four isocentres for the 3 patients, and resulted in target coverage equal or superior to 98% for all three patients.

Magnetic resonance imaging (MRI) is conventionally used for macroscopic qualitative observations. However, increasingly there is a need for quantitative MRI measures, which may lead to enhanced detection sensitivity. Two quantitative techniques that may be used to make neuroanatomical inferences about a population or between different populations are magnetization transfer ratio (MTR) and voxel-based morphometry (VBM). / VBM involves the statistical analysis of smoothed segmented white or gray matter maps to reflect increases or decreases in the probability of classifying a voxel as either white or gray matter. MTR provides a measure of the interaction of water and semi-solids within tissue, and thus is indicative of its macromolecular density and microstructural integrity. An MTR group analysis may detect variations of these semi-solid tissue characteristics within or between populations. / This thesis investigates the relationship between information attained from VBM and MTR population studies carried out in the context of the Saguenay Youth Study. Additionally, through this study, the effects of age and gender on brain neuroanatomy are explored using the above techniques. The observed age and gender VBM and MTR effects were consistent with existing literature, but also offered new findings. Overall, applying MTR in conjunction with VBM allows for further insight into the origins of specific anatomical changes, and the discovery of areas that undergo within-tissue development without corresponding white or gray matter volume changes.

Real-time ultrasonic imaging systems have been available for more than fifty-five years and are becoming an important tool in the practice of modern medicine. During this time much has occurred to the basic architecture and functions of these clinical systems and their beamformers, which are, in many ways, the most important components of these systems. Throughout most of the 30 years of real time imaging, analog beamformers have been the mainstay of all ultrasonic instruments. But at the present time the industry is undergoing a major shift toward digital beamformation with the introduction of several commercial systems. The thesis will look at the evolution of digital ultrasound beamformers, some of the changes that have occurred and will discuss current trends in beamformer design. Typical analog and digital beam formers have been simulated and their performance compared in terms of beam width, side lobe levels and signal-to-noise ratio. Also the effect of apodization on images have been examined. Experimental investigations have been carried out to compare delay-sum-add and synthetic aperture imaging on different phantoms. The technical challenges in digital beamformation win be reviewed, as also the constraints introduced by today's market place. Finally, the future of digital beamformation in the context of advances in computer and microelectronics technologies is discussed.

Electron beams are used to treat superficial lesions in radiation oncology by taking advantage of the sharp dose fall-off and the limited range of the particles in tissue. The irregular shape of individual tumors, however, often requires custom made shielding in order to geometrically shape the radiation field to the target, while minimizing the dose to surrounding tissues. In many institutions, low melting alloy or lead cutouts are used for electron beam shaping. In this work, electron dosimetry beam parameters such as percentage depth dose (PDD), outputs, and beam profiles, were measured with ten different electron beams from two linear accelerators. The dependence of beam characteristics on field size and shape, particularly for small cutouts, was investigated. In addition, this project examined different methods for measuring electron PDDs, including film densitometry, ion chambers, and diode dosimetry. / The work presented here demonstrates that the depth dose effect is significant when one of the field dimensions of the cutout is less than R p, the practical range of electrons. For these cutouts, it was observed that both PDD and outputs vary significantly due to the lack of lateral electronic equilibrium. As the cutout becomes smaller, the depth of dose maximum (dmax) shifts towards the surface, while the output at dmax decreases. Therefore, it is crucial that PDDs and outputs are either measured or calculated for small field electron cutouts used in the clinical setting.

In PET activation studies, brain function is detected by identifying regional differences in tracer concentration between images acquired while a subject(s) executes different tasks. When imaging volunteers or patients, the differences produced by executing higher-order cognitive processes have magnitudes similar to those of the noise found in the measurements. Therefore, the development of techniques that enhance the signal to noise (S/N) of activation images is of great importance. / The work described here was performed to investigate and evaluate three strategies designed to enhance the S/N in activation images of cognitive function. / First, the applicability of switched protocols to the study of cognitive paradigms was analysed. Switched protocols enhance the S/N of subtracted images by manipulating tracer kinetics during the washout phase of the input function. This allows the acquisition of images up to 4-min long. Activation foci detected with switched and standard rCBF protocols were compared using two language tasks. The switched protocol improved the signal statistical significance up to 38% by reducing image noise. / Second, a protocol designed to augment the S/N enhancements yielded by switched protocols was evaluated. This protocol enhances the S/N of H 215O bolus PET activation images by altering the relative distribution of tracer in the uptake and washout phases of the input function. The protocol combines task-switching with the release of a large pool of activity-free blood (cold-bolus). The combined protocol was first tested using computer simulations of the uptake of H215O into the brain and then evaluated by imaging subjects performing two language tasks. In the PET experiments, activation foci obtained using the combined protocol had significantly higher t-statistic values than equivalent foci detected using a conventional switched protocol (mean improvement 36%). The improvements resulted from increases in signal magnitude. / Third, the performance of the ECAT EXACT HR + 3D PET scanner when employed to measure brain function using H215O bolus activation protocols in single sessions was evaluated. Using two sensory tasks and rCBF measurements, the scanner Performance was studied at different count-rates. Optimal dose fractionation strategies for one session imaging studies employing standard rCBF, switched, and cold-bolus/switched protocols were determined.

<p> Fluorescence molecular imaging will have an important clinical impact in the area of guided oncology surgery, where emerging technologies are poised to provide the surgeon with real-time molecular information to guide resection, using targeted molecular probes. The development of advanced surgical systems has gone hand in hand with probe development, and both aspects are analyzed in this work. A pulsed-light fluorescence guided surgical (FGS) system has been introduced to enable video rate visible light molecular imaging under normal room light conditions. The concepts behind this system design are presented and performance is compared with a commercial system in both phantom and <i> in vivo</i> animal studies using PpIX fluorescence. </p><p> The second critical advance in the emergence of these technologies has been the development of targeted near infrared (NIR) probes. A small, engineered three-helix protein was analyzed for imaging of glioma tumors. The blood brain barrier affects delivery of probes and the superior delivery of a smaller targeted protein (anti-EGFR Affibody) as compared to a full sized antibody is shown using a murine model, <i>ex vivo</i> tissue slices and a commercial imaging system. This small targeted probe is examined further for its possible application in FGS using the pulsed light imaging system. A concentration sensitivity analysis to determine the lower bounds on concentration needed for effective imaging is performed with this culminating analysis carried out in a murine orthotopic glioma tumor model.</p>

<p> The primary goal of this thesis is to develop a combined MR relaxation (T₂ and T_1&rho;), diffusion (ADC, apparent diffusion coefficient), elastography (shear stiffness) method to fully characterize the development of tissue-engineered cartilage in terms of the changes in its composition, structure, and mechanical properties during tissue growth. We do this for the purpose of understanding how we may better use MR-based methodologies to noninvasively monitor and optimize the cartilage tissue engineering process without sacrificing the constructs.</p><p> While conventional T₂ and ADC have been widely used in the studies of engineered cartilage tissues, there were few T_1&rho; and MRE studies related to it. We begin by demonstrating the potential capabilities of T₂, T_1&rho;, ADC, and shear stiffness in characterization of a scaffold-free engineered cartilage tissue. We examine the correlations between MR parameters and biochemical determined macromolecule contents in tissue-engineered cartilage. We show that, in addition to the conventional T₂ and ADC, T_1&rho; and MRE can also be used as potential biomarkers to assess the specific changes in proteoglycan content and mechanical properties of engineered cartilage during tissue growth. </p><p> Secondly, to increase the efficiency of MR characterization of engineered tissues, we develop two new methodologies for simultaneous acquisition of MRI and MRE data: (1) diffusion and MRE (dMRE) and (2) T_1&rho; and MRE (T_1&rho;-MRE), respectively. Conventional T_1&rho;, diffusion, and MRE acquisitions are performed as separate measurements that prolong the imaging protocols. The dMRE and T_1&rho;-MRE are developed to overcome this problem by acquiring two pieces of information in one temporally resolved scan. This allows the simultaneous characterization of both biochemical and mechanical properties of engineered cartilage tissues. We carry out dMRE and T_1&rho;-MRE experiments on tissue-mimicking phantoms to show the feasibilities of two techniques. The results obtained show a good correspondence between simultaneous acquisitions and conventional separate acquisition methods. We expect that the combined MRI/MRE methods will benefit the optimal cartilage tissue engineering process.</p>

Intensity modulated photon beam radiation therapy often results in dynamically delivered beams with small field sizes and steep dose gradients. This defines a need for an integrating, tissue-equivalent, high resolution dosimeter. 3D ferrous sulfate gel based dosimetry involves the use of magnetic resonance (MR) images of radiosensitive paramagnetic gels. The goal of this work is to create a patient specific quality assurance (QA) procedure that links measured dosimetnc information to clinical goals. / The gel dosimeter system is tested through a set of simple experiments which characterize and confirm the system as a valid QA tool for conformal and intensity modulated radiation therapy. / For this work, dynamic photon beams are created on a commercially available inverse treatment planning system and the treatment is delivered to a gel filled acrylic mold. Software has been developed to quantify dose from the QA MR images, and to register this information to the planning computed tomography (CT) scan. The software displays the measured dose on the planning CT, and calculates dose-volume histograms for the registered measured data and contoured patient structures. This work reveals good agreement between planned and measured dose distributions, with less than 5% difference in the mean doses of the contoured patient structures.

During the past ten years, radiosurgery has moved from an obscure radiation treatment modality practiced in only a few specialized centers in the world, to a mainstream radiotherapeutic technique practiced in most major radiotherapy centers. Currently, the main thrust of development in radiosurgery is aimed at conformal dose delivery to irregular intracranial targets. This thesis deals with theoretical and practical aspects of the use of static, non-coplanar, conformal fields in radiosurgery. / For a typical radiosurgical case involving an irregular target, a comparison was made between treatment plans using the dynamic technique with one and two isocenters and a treatment plan using 7 fixed, non-coplanar, irregularly shaped beams. The static conformal fields plan achieved a target-dose conformation similar to the 2-isocenter dynamic plan, treating 2 to 3 times less healthy tissue to intermediate and high doses that did the 1-isocenter dynamic plan, while delivering a much more uniform dose to the target volume. / A comparison was also made between treatment plans using a varying number of static conformal fields. While the degree of tissue sparing and target-dose homogeneity were both shown to increase with the number of static fields, this increase was found to become smaller and smaller as the number of fields was successively raised from 5 to 7, from 7 to 9 and, ultimately, from 9 to 11. A conclusion is reached that a number of fields between 7 and 9 represents a reasonable compromise between the degree of tissue sparing and target-dose homogeneity achieved, and the ease with which the radiosurgical procedure is planned and delivered.