Thèses sur le sujet « Imaging Medicale »

Les tissus biologiques sont des milieux fortement diffusant pour la lumière. En conséquence, les techniques d'imagerie actuelles ne permettent pas d'obtenir un contraste optique en profondeur à moins d'user d'approches invasives. L'imagerie acousto-optique (AO) est une approche couplant lumière et ultrasons (US) qui utilise les US afin de localiser l'information optique en profondeur avec une résolution millimétrique. Couplée à un échographe commercial, cette technique pourrait apporter une information complémentaire permettant d'augmenter la spécificité des US. Grâce à une détection basée sur l'holographie photoréfractive, une plateforme multi-modale AO/US a pu être développée. Dans ce manuscrit, les premiers tests de faisabilité ex vivo sont détaillés en tant que premier jalon de l'imagerie clinique. Des métastases de mélanomes dans le foie ont par exemple été détectées alors que le contraste acoustique n'était pas significatif. En revanche, ces premiers résultats ont souligné deux obstacles majeurs à la mise en place d'applications cliniques.Le premier concerne la cadence d'imagerie de l'imagerie AO très limitée à cause des séquences US prenant jusqu'à plusieurs dizaines de secondes. Le second concerne le speckle qui se décorrèle en milieu vivant sur des temps inférieurs à 1 ms, trop rapide pour les cristaux photorefractif actuellement en palce. Dans ce manuscrit, je propose une nouvelle séquence US permettant d'augmenter la cadence d'imagerie d'un ordre de grandeur au moins ainsi qu'une détection alternative basée sur le creusement de trous spectraux dans des cristaux dopés avec des terres rares qui permet de s'affranchir de la décorrélation du speckle
Biological tissues are very strong light-scattering media. As a consequence, current medical imaging devices do not allow deep optical imaging unless invasive techniques are used. Acousto-optic (AO) imaging is a light-ultrasound coupling technique that takes advantage of the ballistic propagation of ultrasound in biological tissues to access optical contrast with a millimeter resolution. Coupled to commercial ultrasound (US) scanners, it could add useful information to increase US specificity. Thanks to photorefractive crystals, a bimodal AO/US imaging setup based on wave-front adaptive holography was developed and recently showed promising ex vivo results. In this thesis, the very first ones of them are described such as melanoma metastases in liver samples that were detected through AO imaging despite acoustical contrast was not significant. These results highlighted two major difficulties regarding in vivo imaging that have to be addressed before any clinical applications can be thought of.The first one concerns current AO sequences that take several tens of seconds to form an image, far too slow for clinical imaging. The second issue concerns in vivo speckle decorrelation that occurs over less than 1 ms, too fast for photorefractive crystals. In this thesis, I present a new US sequence that allows increasing the framerate of at least one order of magnitude and an alternative light detection scheme based on spectral holeburning in rare-earth doped crystals that allows overcoming speckle decorrelation as first steps toward in vivo imaging

Oggigiorno, la mole di dati biomedicali eterogenei è in continua crescita grazie alle nuove tecniche di sensing e alle tecnologie ad high-throughput. Relativamente all'analisi di immagini biomedicali, i progressi relativi alle modalità di acquisizione di immagini agli esperimenti di imaging ad high-throughput stanno creando nuove sfide. Questo ingente complesso di informazioni può spesso sopraffare le capacità analitiche sia dei medici nei loro processi decisionali sia dei biologi nell'investigazione di sistemi biochimici complessi. In particolare, i metodi di imaging quantitativo forniscono informazioni scientificamente rilevanti per la predizione, la prognosi o la valutazione della risposta al trattamento, prendendo in considerazione anche approcci di radiomica. Pertanto, l'analisi computazionale di immagini medicali e biologiche svolge un ruolo chiave in applicazioni di radiologia e di laboratorio. A tal proposito, framework basati su tecniche avanzate di Machine Learning e Computational Intelligence permettono di migliorare significativamente i tradizionali approcci tradizionali di Image Processing e Pattern Recognition. Tuttavia, le tecniche convenzionali di Intelligenza Artificiale devono essere propriamente adattate alle sfide uniche imposte dai dati di imaging biomedicale. La presente tesi mira a proporre innovativi metodi assistiti da calcolatore per l'analisi di immagini biomedicali, da utilizzare anche come strumento per lo sviluppo di Sistemi di Supporto alle Decisioni Cliniche, tenendo sempre in considerazione la fattibilità delle soluzioni sviluppate. In primo luogo, sono descritti gli algoritmi classici di Image Processing realizzati, focalizzandosi sugli approcci basati su regioni e sulla morfologia matematica. Dopodiché, si introducono le tecniche di Pattern Recognition, applicando il clustering fuzzy non supervisionato e i modelli basati su grafi (i.e., Random Walker e Automi Cellulari) per l'elaborazione di dati multispettrali e multimodali di imaging medicale. In riferimento ai metodi di Computational Intelligence, viene presentato un innovativo framework evolutivo basato sugli Algoritmi Genetici per il miglioramento e la segmentazione di immagini medicali. Inoltre, è discussa la co-registrazione di immagini multimodali utilizzando Particle Swarm Optimization. Infine, si investigano le Deep Neural Network: (i) le capacità di generalizzazione delle Convolutional Neural Network nell'ambito della segmentazione di immagini medicali provenienti da studi multi-istituzionali vengono affrontate mediante la progettazione di un'architettura che integra blocchi di ricalibrazione delle feature, e (ii) la generazione di immagini medicali realistiche basata sulle Generative Adversarial Network è applicata per scopi di data augmentation. In conclusione, il fine ultimo di tali studi è quello di ottenere conoscenza clinicamente e biologicamente utile che possa guidare le diagnosi e le terapie differenziali, conducendo verso l'integrazione di dati biomedicali per la medicina personalizzata. Difatti, i metodi assistiti da calcolatore per l'analisi delle immagini biomedicali sono vantaggiosi sia per la definizione di biomarcatori basati sull'imaging sia per la medicina e biologia quantitativa.
Nowadays, the amount of heterogeneous biomedical data is increasing more and more thanks to novel sensing techniques and high-throughput technologies. In reference to biomedical image analysis, the advances in image acquisition modalities and high-throughput imaging experiments are creating new challenges. This huge information ensemble could overwhelm the analytic capabilities needed by physicians in their daily decision-making tasks as well as by biologists investigating complex biochemical systems. In particular, quantitative imaging methods convey scientifically and clinically relevant information in prediction, prognosis or treatment response assessment, by also considering radiomics approaches. Therefore, the computational analysis of medical and biological images plays a key role in radiology and laboratory applications. In this regard, frameworks based on advanced Machine Learning and Computational Intelligence can significantly improve traditional Image Processing and Pattern Recognition approaches. However, conventional Artificial Intelligence techniques must be tailored to address the unique challenges concerning biomedical imaging data. This thesis aims at proposing novel and advanced computer-assisted methods for biomedical image analysis, also as an instrument in the development of Clinical Decision Support Systems, by always keeping in mind the clinical feasibility of the developed solutions. The devised classical Image Processing algorithms, with particular interest to region-based and morphological approaches in biomedical image segmentation, are first described. Afterwards, Pattern Recognition techniques are introduced, applying unsupervised fuzzy clustering and graph-based models (i.e., Random Walker and Cellular Automata) to multispectral and multimodal medical imaging data processing. Taking into account Computational Intelligence, an evolutionary framework based on Genetic Algorithms for medical image enhancement and segmentation is presented. Moreover, multimodal image co-registration using Particle Swarm Optimization is discussed. Finally, Deep Neural Networks are investigated: (i) the generalization abilities of Convolutional Neural Networks in medical image segmentation for multi-institutional datasets are addressed by conceiving an architecture that integrates adaptive feature recalibration blocks, and (ii) the generation of realistic medical images based on Generative Adversarial Networks is applied to data augmentation purposes. In conclusion, the ultimate goal of these research studies is to gain clinically and biologically useful insights that can guide differential diagnosis and therapies, leading towards biomedical data integration for personalized medicine. As a matter of fact, the proposed computer-assisted bioimage analysis methods can be beneficial for the definition of imaging biomarkers, as well as for quantitative medicine and biology.

Breast cancer is the most crucial cancer type among all other cancer types. There are many imaging techniques used to screen breast carcinoma. These are mammography, ultrasound, computed tomography, magnetic resonance imaging, infrared imaging, positron emission tomography and electrical impedance tomography. However, there is no gold standard in breast carcinoma diagnosis. The object of this study is to create a hybrid system that uses thermal and electrical imaging methods together for breast cancer diagnosis. Body tissues have different electrical conductivity values depending on their state of health and types. Consequently, one can get information about the anatomy of the human body and tissue&rsquo
s health by imaging tissue conductivity distribution. Due to metabolic heat generation values and thermal characteristics that differ from tissue to tissue, thermal imaging has started to play an important role in medical diagnosis. To increase the temperature contrast in thermal images, the characteristics of the two imaging modalities can be combined. This is achieved by implementing thermal imaging applying electrical currents from the body surface within safety limits (i.e., thermal imaging in active mode). Electrical conductivity of tissues changes with frequency, so it is possible to obtain more than one thermal image for the same body. Combining these images, more detailed information about the tumor tissue can be acquired. This may increase the accuracy in diagnosis while tumor can be detected at deeper locations. Feasibility of the proposed technique is investigated with analytical and numerical simulations and experimental studies. 2-D and 3-D numerical models of the female breast are developed and feasibility work is implemented in the frequency range of 10 kHz and 800 MHz. Temporal and spatial temperature distributions are obtained at desired depths. Thermal body-phantoms are developed to simulate the healthy breast and tumor tissues in experimental studies. Thermograms of these phantoms are obtained using two different infrared cameras (microbolometer uncooled and cooled Quantum Well Infrared Photodetectors). Single and dual tumor tissues are determined using the ratio of uniform (healthy) and inhomogeneous (tumor) images. Single tumor (1 cm away from boundary) causes 55 °
mC temperature increase and dual tumor (2 cm away from boundary) leads to 50 °
mC temperature contrast. With multi-frequency current application (in the range of 10 kHz-800 MHz), the temperature contrast generated by 3.4 mm3 tumor at 9 mm depth can be detected with the state-of-the-art thermal imagers.

Ce travail est consacre aux aspects physiques et biophysiques de l'imagerie fonctionnelle cerebrale chez l'homme par resonance magnetique nucleaire. Dans le premier chapitre, le principe des techniques d'imagerie rapide, dont l'imagerie fonctionnelle est une des applications, est expose. Cette partie insiste notamment sur la sensibilite des differentes sequences aux parametres susceptibles de modifier le signal rmn lors d'une activation cerebrale. Les deux chapitres suivants presentent un ensemble de resultats experimentaux concernant l'imagerie des fonctions cerebrales motrices etudiees a 1,5 tesla avec une technique d'echo de gradient. Nous avons en premier lieu evalue les differents phenomenes physiologiques responsables des variations de signal detectees: variation du debit sanguin cerebral (effet de flux) et variation du taux d'oxygenation sanguine (effet de susceptibilite magnetique). Par la suite, nous avons demontre que les variations de signal observees provenaient de maniere predominante de certaines veines de drainage situees a la surface des zones corticales d'interet. Ce resultat nous a permis d'introduire le concept d'angiographie fonctionnelle. Le dernier chapitre est independant du sujet principal. Il concerne le developpement d'une technique d'acquisition rapide de profils d'excitation d'impulsions selectives

In ultrasound array imaging, the beamforming operation is performed by aligning and processing the received echo signals from each individual array element to form a complete image. This operation can be performed in many different ways, where adaptive and non-adaptive beamformers are considered as the main categories. Adaptive beamformers exploit the statistical correlation between the received data to find a weighting value at the focal point, instead of using a fixed weighting window in non-adaptive beamforming. This results in a significant improvement in the image quality in terms of resolution and sidelobes reduction. This improvement is necessary for ultrafast imaging because of the lack of focusing in Plane Wave Imaging (PWI) that results in lowering the SNR, and thus the produced imaging quality is reduced. This thesis analyses different adaptive beamforming techniques for ultrafast imaging. For accurate medical diagnosis, the frame rate, the imaging resolution, contrast and speckle homogeneity are all considered as important parameters that contribute to the final imaging result. To be able to evaluate each technique by minimizing the effect of external parameters, two different analysis were performed. First an empirical expression for PWI lateral resolution is produced after studying the effect of the imaging parameters on this imaging method. Then a method for selecting the suitable steering angles in Compound Plane- Wave Imaging (CPWI) is introduced, with a detailed explanation for the effect of the compound angles on resolution and sidelobes level. In order to add the contrast improvement to the properties of adaptive beamformers, some techniques like the coherence-based factors and Eigenspace-Based Minimum Variance (ESBMV) are produced in the literature. After demonstrating the principle of Minimum Variance adaptive beamformer, a detailed comparison for the types of coherence-based factors is given. In addition, a new technique of Partial-ESBMV is introduced to modify reference ESBMV so that no Black Box Region artefacts nor dark spots appear when using this method in medical imaging. After explaining its background and properties using cystic and wire phantoms, the proposed method is applied to the real RF data of carotid artery, as an application to clarify the efficiency of this method in medical ultrasound imaging.

It is estimated that over half of current adults within Great Britain under the age of 65 will be diagnosed with cancer at some point in their lifetime. Medical Imaging forms an essential part of cancer clinical protocols and is able to furnish morphological, metabolic and functional information. The imaging of molecular interactions of biological processes in vivo with Positron Emission Tomography (PET) is informative not only for disease detection but also therapeutic response. The qualitative and quantitative accuracy of imaging is thus vital in the extraction of meaningful and reproducible information from the images, allowing increased sensitivity and specificity in the diagnosis and precision of image guided treatment. Furthermore the utilization of complementary information obtained via Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) in integrated PET-CT and PET-MR devices offers the potential for the synergistic effects of hybrid imaging to provide increased detection and precision of diagnosis with reduced radiation dose in a fully comprehensive single imaging examination. With the increasing sophistication in imaging technology respiratory organ motion during imaging has demonstrated itself to be a major degrading factor of PET image resolution. A modest estimate of respiratory motion amplitude of 5mm, results in PET system resolution degrading from ≈ 5mm to ≈8.5mm. This evidently has an impact on cancer lesion detectability. Therefore accurate and robust methods for respiratory motion correction are required for both clinical effectiveness and economic justification for purchasing state of the art hybrid PET scanners with high resolution capabilities. In addition the judicious use of imaging resources from hybrid imaging devices coupled with advanced image processing / acquisition protocols will allow optimization of data used for improving quantitative accuracy of PET images and those used for clinical interpretation. In essence it would prove impractical to use the MR scanner purely for monitoring respiratory motion. Numerous methods exist to attempt to correct PET imaging for respiratory motion. As presented in this thesis many methods demonstrate themselves to be ineffective in the clinical setting where the patients breathing patterns appear irregular in comparison to the idealized situation of regular periodic motion. Advanced respiratory motion correction techniques utilize hybrid PET/CT, PET/MR scanners coupled with an external source of information which serves as a surrogate to build a static correspondence to the estimated internal respiratory motion. Static models however are unable to adapt to their external environment and do not consider time dependent changes in the state of a system. A further confounding factor in the development and assessment of motion correction schemes for medical imaging data is the inability to acquire volumetric data with high contrast and high spatial and temporal resolution which serves as a ground truth for quantifying model accuracy and confidence. This thesis addresses both problems by analysing respiratory motion correspondence modelling under a manifold learning and alignment paradigm which may be used to consolidate many of the respiratory motion estimation models that exist today. A Bayesian approach is adopted in this work to incorporate a-priori information into the model building stage for a more robust, flexible adaptive respiratory motion estimation / correction framework. This thesis constructs and tests the first proposed adaptive motion model to correlate a surrogate signal with internal motion. This adaptive approach allows the relationship between external surrogate signal and internal motion to change dependent upon breathing pattern and system noise. The adaptive model was compared to a state-of the-art static model and allows more accurate motion estimates to be made when the patient is breathing with an irregular pattern. Testing performed on MRI data from 9 volunteers demonstrated the adaptive model was statistically more significant (p < 0.001) in the presence of irregular motion in comparison to a static model. The adaptive Kalman model on average reduced the error in motion by 30% in comparison to the static model. Utilizing the adaptive model during a typical PET study would theoretically result in ≈ 10% increase in PET resolution in comparison to relying on a static model alone for motion correction. The adaptive Kalman model has the capability to increase the performance of PET system resolution from ≈ 8.5mm to ≈ 5.8mm, ≈ 30%. A simulated PET study also demonstrated ≈ 30% increase in tumour uptake when using motion correction. Also demonstrated in the thesis is the first method to acquire volumetric imaging data from sparse MR samples during free breathing to allow the realization of high contrast, high resolution 4D models of respiratory motion using limited acquired data. The developed framework facilitates greater freedom in the acquisition of free breathing respiratory motion sequences which may be used to inform motion modelling methods in a range of imaging modalities as well as informing the development of generalizable models of human respiration. It is shown that the developed approach can provide equivalent motion vector fields in comparison to fully sampled 4D dynamic data. The incorporation of the manifold alignment step into the sparse motion model reduces the error in motion estimates by ≈ 16%. Example images of propagated motion are also presented as supplementary information. The thesis concludes by generalizing the concepts in this work and looking to utilize the developed methods to other problems in the medical imaging arena.

Mestrado em Engenharia de Computadores e Telemática
A aplicação CapView utiliza um algoritmo de classificação baseado em SVM (Support Vector Machines) para automatizar a segmentação topográfica de vídeos do trato intestinal obtidos por cápsula endoscópica. Este trabalho explora a aplicação de processadores gráficos (GPU) para execução paralela desse algoritmo. Após uma etapa de otimização da versão sequencial, comparou-se o desempenho obtido por duas abordagens: (1) desenvolvimento apenas do código do lado do host, com suporte em bibliotecas especializadas para a GPU, e (2) desenvolvimento de todo o código, incluindo o que é executado no GPU. Ambas permitiram ganhos (speedups) significativos, entre 1,4 e 7 em testes efetuados com GPUs individuais de vários modelos. Usando um cluster de 4 GPU do modelo de maior capacidade, conseguiu-se, em todos os casos testados, ganhos entre 26,2 e 27,2 em relação à versão sequencial otimizada. Os métodos desenvolvidos foram integrados na aplicação CapView, utilizada em rotina em ambientes hospitalares.
The CapView application uses a classification algorithm based on SVMs (Support Vector Machines) for automatic topographic segmentation of gastrointestinal tract videos obtained through capsule endoscopy. This work explores the use graphic processors (GPUs) to parallelize the segmentation algorithm. After an optimization phase of the sequential version, two new approaches were analyzed: (1) development of the host code only, with support of specialized libraries for the GPU, and (2) development of the host and the device’s code. The two approaches caused substantial gains, with speedups between 1.4 and 7 times in tests made with several different individual GPUs. In a cluster of 4 GPUs of the most capable model, speedups between 26.2 and 27.2 times were achieved, compared to the optimized sequential version. The methods developed were integrated in the CapView application, used in routine in medical environments.

The goal of this research is to develop computational methods for predicting how a given medical imaging system and reconstruction algorithm will perform when mathematical observers for tumor detection use the resulting images. Here the mathematical observer is the ideal observer, which sets an upper limit to the performance as measured by the Bayesian risk or receiver operating characteristic analysis. This dissertation concentrates on constructing the ideal observer in complex detection problems and estimating its performance. Thus the methods reported in this dissertation can be used to approximate the ideal observer in real medical images. We define our detection problem as a two-hypothesis detection task where a known signal is superimposed on a random background with complicated distributions and embedded in independent Poisson noise. The first challenge of this detection problem is that the distribution of the random background is usually unknown and difficult to estimate. The second challenge is that the calculation of the ideal observer is computationally intensive for non stylized problems. In order to solve these two problems, our work relies on multiresolution analysis of images. The multiresolution analysis is achieved by decomposing an image into a set of spatial frequency bandpass images so each bandpass image represents information about a particular fitness of detail or scale. Connected with this method, we will use three types of image representation by invertible linear transforms. They are the orthogonal wavelet transform, pyramid transform and independent component analysis. Based on the findings from human and mammalian vision, we can model textures by using marginal densities of a set of spatial frequency bandpass images. In order to estimate the distribution of an ensemble of images given the empirical marginal distributions of filter responses, we can use the maximum entropy principle and get a unique solution. We find that the ideal observer calculates a posterior mean of the ratio of conditional density functions, or the posterior mean of the ratio of two prior density functions, both of which are high dimensional integrals and have no analytic solution usually. But there are two ways to approximate the ideal observer. The first one is a classic decision process; that is, we construct a classifier following feature extraction steps. We use the integrand of the posterior mean as features, which are calculated at the estimated background close to the posterior mode. The classifier combines these features to approximate the integral (or the ideal observer). Finally, if we know both the conditional density function and the prior density function then we can also approximate the high dimensional integral by Monte Carlo integration methods. Since the calculation of the posterior mean is usually a very high dimensional integration problem, we must construct a Markov chain, which can explore the posterior distribution efficiently. We will give two proposal functions. The first proposal function is the likelihood function of random backgrounds. The second method makes use of the multiresolution representation of the image by decomposing the image into a set of spatial frequency bands. Sampling one pixel in each band equivalently updates a cluster of pixels in the neighborhood of the pixel location in the original image.

Introduction: Endovascular chemotherapy treatment allows localized delivery adjacent to the target tumor; allowing an increased dosage and decreased leakage to other areas. It also allows for the opportunity to filter chemotherapy escaping the target tumor and entering the bloodstream. The ChemoFilter - a temporarily deployable, endovascular device will do just that; reducing systemic toxicity thus reducing adverse side effects from chemotherapy treatment. This will allow further increased dosage, increased tumor suppression, and increased tolerance to treatment. ChemoFilter has successfully filtered the chemotherapeutic Doxorubicin, but had yet to be tested in other chemotherapeutics. This study evaluates binding with new chemotherapeutics: Cisplatin, Carboplatin, and a cocktail comprised of Cisplatin and Doxorubicin.

Materials and Methods: ChemoFilter prototypes based on: 1.) Genomic DNA and 2.) Dowex (ion-exchange) resin, were evaluated for their ability to bind chemotherapy in vitro in phosphate-buffered saline (PBS). ChemoFilter was tested free in solution and encapsulated in nylon or polyester mesh packets of various dimensions. Concentrations were quantified using inductively coupled plasma mass spectrometry (IPC-MS), ultraviolet-visible spectrophotometry (UV-Vis), or fluorospectrometry. ¹¹C, ¹³C, and/or ¹⁴C radiolabeling Carboplatin began for in vitro and in vivo ChemoFilter quantification. In vitro quantification can include scintillation and/or gamma counting. In vivo may include Positron Emission Tomography (PET) imaging, Hyperpolarized ¹³C Magnetic Resonance Imaging (MRI), and/or Magnetic Resonance Spectroscopy (MRS) for real-time visualization. Reactions were verified using High Performance Liquid Chromatography (HPLC) for chemical species identification.

Results and Discussion: Results indicate significant and nearly complete, ~99% (p<0.01) clearance of Cisplatin using the DNA ChemoFilter sequestered in Nylon mesh, quantified with gold standard ICP-MS (evidenced at 214 and 265 nm). The Ion-exchange ChemoFilter has significant clearance, within seconds, of both Doxorubicin and Cisplatin mixed in a cocktail solution. However, it appears some Cisplatin is binding to the Nylon Mesh itself. Size, shape, and material of the mesh have been optimized. A potential mechanism for ¹¹C, ¹³C, or ¹⁴C radiolabeling of Carboplatin has been developed and early results have been successful. ChemoFilter works much more efficiently when sequestered in nylon packets of specific geometries. Significant improvements have been made to ChemoFilter, moving the device closer to clinical trials.

The work described in this thesis concentrates on two aspects of Proton NMR imaging: development and evaluation of new/old experimental sequences and application of those techniques to study some non-medical systems that are of industrial importance. Two-dimensional Fourier transform spin warp imaging technique has been evaluated. Importantly, the adaptation of a conventional high resolution spectrometer to perform imaging has been demonstrated with means of "phantoms". This includes calibration of magnetic field gradients, mapping the static magnetic field and radiofrequency field distributions and intensity measurements related to proton spin densities. In addition, a preliminary study describes microscopic imaging of glass capillary tube phantoms containing water. Several different sequences related to Chemical Shift imaging including the one developed during the study have been described. A brief insight into chemical shift artifacts as well as some experimental methods of minimizing some of them have also been presented. The potential of NMR imaging to study non-medical systems has been explored in three different areas of interest: Chromatography columns. Porous rock samples and Wood samples. A variety of NMR imaging sequences have been used to study some interesting and challenging features of these systems which clearly extends the scope of NMR imaging science.
Science, Faculty of
Chemistry, Department of
Graduate

This thesis addresses two problems in medical imaging, the development of a system for 3D imaging with ultrasound and a system for making titanium prostheses for cranioplasty. Central to both problems is the construction and depiction of surfaces from volume data where the data is not acquired on a regular grid or is incomplete. A system for acquiring 3D pulse-echo ultrasound data using a conventional 2D ultrasound scanner equipped with an electro-magnetic spatial locator is described. The non-parallel nature of 2D B-scan slices acquired by the system requires the development of new visualisation algorithms to depict three dimensional structures. Two methods for visualising iso-valued surfaces from the ultrasound data are presented. One forms an intermediate volume reconstruction suitable for conventional ray-casting while the second method renders surfaces directly from the slice data. In vivo imaging of human anatomy is used to demonstrate reconstructions of tissue surfaces. Filtering and spatial compounding of scan data is used to reduce speckle. The manifestation of 2D artefacts in 3D surface reconstructions is also illustrated. Pulse-echo ultrasound primarily depicts tissue boundaries. These are characterised by incomplete acoustic interfaces contaminated by noise. The problem of reconstructing tissue interfaces from ultrasound data is viewed as an example of the general problem of reconstructing an object's shape from unorganised surface data. A novel method for reconstructing surfaces in the absence of a priori knowledge of the object's shape, is described and applied to 3D ultrasound data. The method uses projections through the surface data taken from many viewpoints to reconstruct surfaces. Aspects of the method are similar to work in computer vision concerning the determination of the shape of 3D objects from their silhouettes. This work is extended significantly in this thesis by considering the reconstruction of incomplete objects in the presence of noise and through the development of practical algorithms for pixel and voxel data. Furthermore, the reconstruction of realistic, non-convex objects is considered rather than simple geometric objects. 2D and 3D ultrasound data derived from phantoms, as well as artificial data, are used to demonstrate reconstructions. The second problem studied in this thesis concerns designing cranial implants to repair defects in the skull. Skull surfaces are extracted from X-ray CT data by ray-casting iso-valued surfaces. A tensor product B-spline interpolant is used in the ray-caster to reduce ripples in the surface data due to partial voluming and the large spacing between CT slices. The associated surface depth-maps are characterised by large irregular holes which correspond to the defect regions requiring repair. Defects are graphically identified by a user in surface-rendered images. Radial basis function approximation is introduced as a method of interpolating the surface of the skull across these defect regions. The fitted surface is used to produce CNC milling instructions to machine a mould in the shape of the surface from a block of hard plastic resin. A cranial implant is then formed by pressing flat titanium plate into the mould under high pressure in a hydraulic press. The system improves upon current treatment procedures by avoiding the manual aspects of fashioning an implant. It is also suitable when other techniques which use symmetry to reconstruct the skull are inadequate or not possible. The system has been successfully used to treat patients at Christchurch Hospital. Radial basis function (RBF) approximation has previously been restricted to problems where the number of interpolation centres is small. The use of newly developed fast methods for evaluating radial basis interpolants in the surface interpolation software results in a computationally efficient system for designing cranial implants and demonstrates that RBFs are potentially of wide interest in medical imaging and engineering problems where data does not lie on a regular grid.

Mestrado em Engenharia de Computadores e Telemática
Hoje em dia, as instituições de cuidados de saúde, utilizam a telemedicina para suportar ambientes colaborativos. Na área da imagem médica digital, a quantidade de dados tem crescido substancialmente nos últimos anos, requerendo mais infraestruturas para fornecer um serviço com a qualidade desejada. Os computadores e dispositivos com acesso à Internet estão acessíveis em qualquer altura e em qualquer lugar, criando oportunidades para partilhar e utilizar recursos online. Uma enorme quantidade de processamento computacional e armazenamento são utilizados como uma comodidade no quotidiano. Esta dissertação apresenta uma plataforma para suportar serviços de telemedicina sobre a cloud, permitindo que aplicações armazenem e comuniquem facilmente, utilizando qualquer fornecedor de cloud. Deste modo, os programadores não necessitam de se preocupar onde os recursos vão ser instalados a as suas aplicações não ficam limitadas a um único fornecedor. Foram desenvolvidas duas aplicações para tele-imagiologia com esta plataforma: repositório de imagens médicas e uma infraestrutura de comunicações entre centros hospitalares. Finalmente, a arquitetura desenvolvida é genérica e flexível permitindo facilmente a sua expansão para outras áreas aplicacionais e outros serviços de cloud.
Healthcare institutions resort largely, nowadays, to telemedicine in order to support collaborative environments. In the medical imaging area, the huge amount of medical volume data has increased over the past few years, requiring high-performance infrastructures to provide services with required quality. Computing devices and Internet access are now available anywhere and at anytime, creating new opportunities to share and use online resources. A tremendous amount of ubiquitous computational power and an unprecedented number of Internet resources and services are used every day as a normal commodity. This thesis presents a telemedicine service platform over the Cloud that allows applications to store information and to communicate easier, using any Internet cloud provider. With this platform, developers do not concern where the resources will be deployed and the applications will not be restricted to a specific cloud vendor. Two tele-imagiologic applications were developed along with this platform: a medical imaging repository and an interinstitutional communications infrastructure. Lastly, the architecture developed is generic and flexible to expand to other application areas and cloud services.

This thesis investigated novel deep learning techniques for advanced medical imaging applications. It addressed three major research issues of employing deep learning for medical imaging applications including network architecture, lack of training data, and generalisation. It proposed three new frameworks for CNN network architecture and three novel transfer learning methods. The proposed solutions have been tested on four different medical imaging applications demonstrating their effectiveness and generalisation. These solutions have already been employed by the scientific community showing excellent performance in medical imaging applications and other domains.

L’obiettivo di questo lavoro è quello di contribuire allo sviluppo di un’epistemolo-gia dell’imaging medico, intendendo con questo termine sia le immagini utilizzate a fini diagnostici, sia le tecnologie che le producono. La mia tesi principale è che le tecnologie di imaging medico non si limitano a produrre immagini più o meno accurate degli organi interni e di alcuni processi fisiologici, ma piuttosto trasformano il corpo in un oggetto scientifico, operando un cambiamento profondo della sua visibilità. Gli strumenti di imaging mutano il corpo in un oggetto visivo che può essere osservato in condizioni sperimentali. A differenza del corpo reale, tale oggetto può essere archiviato, consultato, condiviso, misurato e manipolato in varie maniere. Questa tesi di fondo è accompagnata da altre due: (1) Le immagini diagnostiche, come tutte le immagini scientifiche, sono veri e propri strumenti cognitivi, strumenti epistemici integrati in un quadro teorico-pratico specifico; (2) Un’immagine che rivela l’interno dell’organismo ha significato e valore diagnostico solo nell’ambito di una specifica concettualizzazione del corpo e della malattia, di conseguenza uno studio sull’epistemologia dell’imaging medico non si potrà limitare a esaminare le immagini diagnostiche in quanto immagini, ma dovrà analizzarle anche nella loro veste di strumenti di diagnosi medica. Per questo motivo nel primo capitolo della dissertazione traccio le linee generali delle condizioni di possibilità storiche e concettuali della radiografia -- la prima tecnologia di imaging medico -- inventata nel 1895. Lo scopo è quello di comprendere quali teorie e pratiche mediche dovessero essere vigenti alla fine del XIX secolo, affinché immagini che parevano ombre del corpo interno potessero essere considerate strumenti diagnostici. La spiegazione da me proposta è che la rilevanza diagnostica della radiografia si fonda sulla concettualizzazione di corpo, malattia e diagnosi resa operativa dall’anatomia clinica già alla fine del XVIII secolo. Seguendo e supportando questa linea di ragionamento mostro che lo stetoscopio, inventato nel 1816, può essere considerato il predecessore materiale e intellettuale dell’imaging medico perché introdusse una primitiva forma di mediazione sensoriale nel campo della diagnostica e permise al medico di esplorare dall’esterno le profondità del corpo del paziente, estraendone segni di malattia. Lo stetoscopio è solo il primo di una vasta famiglia di strumenti inventati nel XIX secolo per visualizzare diversi aspetti della morfologia interna e della fisiologia del vivente. Sebbene ciascuno di questi strumenti rispondesse a specifiche necessità diagnostiche e ponesse specifici problemi epistemologici, si possono identificare alcune caratteristiche comuni: tutti avevano come obbiettivo quello di sostituire le sensazioni soggettive dei pazienti e dei medici con indici oggettivi di salute e malattia; tutti creavano registri visivi dell’interno del corpo umano che potevano essere archiviati, recuperati e condivisi da diversi medici; tutti richiedevano la creazione di un linguaggio specializzato, condiviso da una comunità medico-scientifica; tutti creavano una progressiva separazione tra il corpo del paziente e il corpo del medico. È in questo complesso scenario di pratiche, oggetti, raffigurazioni e idee che la radiografia fece la sua comparsa e acquisì la sua funzione diagnostica. Nel secondo capitolo prendo in esame la nascita della fotografia, al fine di comprendere in che modo la prima tecnologia di produzione meccanica di immagini influenzò la medicina. I principali riferimenti teorici di questo capitolo sono dati dalla semiotica di Charles Sanders Peirce, in particolare la sua classificazione dei segni in indici, icone e simboli, e dalla riflessione di Walter Benjamin sulla serie fotografica (produzione e riproduzione meccanica di un’immagine e del corpo in essa rappresentato), sull’intrinseco potenziale analitico e di dissezione della fotografia (il fotografo come chirurgo), e sull’inconscio ottico (fotografia come protesi che arricchisce e trasforma l’esperienza sensibile). Basandomi su questi autori e esaminando i lavori dei primi medici-fotografi nell’ambito della psichiatria, dermatologia, fisiologia e neurologia, mostro che le serie fotografiche raccolte in riviste mediche, manuali di studio e archivi ospedalieri produssero uno sguardo clinico in senso foucauldiano. Sostengo, inoltre, che la serie fotografica faceva parte di un più ampio apparato sperimentale che includeva il paziente, la macchina fotografica e l’osservatore il cui scopo era trasformare il corpo e la malattia in oggetti visivi che potessero essere sottoposti ad analisi scientifica. Nel terzo capitolo discuto il problema del referente invisibile, ossia analizzo i processi attraverso cui le immagini fotografiche di oggetti invisibili vengono dotate di significato. Probabilmente questo è il problema fondamentale di qualunque tipo di imaging scientifico. Quando il referente di una fotografia è invisibile, la modalità iconica di significazione non può essere messa in atto, perché nell’immagine prodotta dallo strumento (sia esso meccanico o elettronico) non possiamo riconoscere nessuna similitudine con l’oggetto rappresentato. Di fatto, potremmo dire che in questi casi l’immagine non assomiglia a nulla. Come sappiamo, dunque, se l’oggetto che vediamo nella fotografia – per esempio una cellula o una lesione tubercolare – è davvero là, e possiede davvero l’aspetto mostrato dall’immagine? Sulla scorta dell’analisi teorica sviluppata nel capitolo precedente, difendo l’idea che la visualizzazione dell’invisibile richieda una peculiare combinazione delle modalità di significazione indicale, iconica e simbolica. La mia argomentazione è costruita in opposizione al concetto di oggettività meccanica proposto da Lorraine Daston e Peter Galison. In particolare, dimostro che l’idea di oggettività meccanica come soppressione moralizzante del soggetto proposta dai due storici è una caricatura delle idee e pratiche sviluppate dagli scienziati del XIX secolo per risolvere il problema della visualizzazione dell’invisibile. La mia argomentazione si articola in tre momenti, corrispondenti all’analisi del problema dell’oggettività e della significazione delle immagini in tre diversi ambiti: microfotografia, cronofotografia e radiografia. Nel quarto capitolo affronto il problema del valore cognitivo delle immagini, sostenendo che le immagini sono strumenti epistemici (nel senso forte, non metaforico della parola strumento) e che rappresentazione e osservazione non sono mai atti puramente automatici, perché richiedono sempre una componente creativa. Come nel capitolo precedente, parte del mio discorso è una refutazione della posizione di Daston e Galison, in particolare per quanto riguarda le loro affermazioni sulla natura passiva di certe rappresentazioni visive. Secondo Daston e Galison, infatti, fino allo sviluppo delle tecnologie digitali, le immagini scientifiche erano mere ri-presentazioni [re-presentations] del mondo, miranti a copiare la natura. Con la comparsa del digitale, invece, si è passati a un’epoca in cui le immagini sono presentazioni [presentations], perché attraverso di esse l’osservatore può visualizzare l’oggetto in mutevoli forme, manipolandolo virtualmente. La mia critica a questa posizione è basata su argomenti storici e teorici. Sul piano storico mostro che i primi tentativi di creare immagini mediche manipolabili risalgono almeno al XVI secolo. Sul piano teorico, ricorrendo alla letteratura prodotta in campi così diversi come la teoria dell’arte e le neuroscienze, dimostro che la nozione di ricezione passiva di un’immagine è insostenibile, perché le immagini coinvolgono sempre l’osservatore in un atto corporeo di percezione che sollecita non solo sensazioni visive, ma anche sensazioni tattili e reazioni motorie. Inoltre, sostengo che l’enfasi posta da Daston e Galison sul nanoimaging come l’unica tecnologia che permette di manipolare l’oggetto durante la fase di produzione di un’immagine è fuorviante. Infatti, anche nei casi in cui non raggiungono le vette di sofisticazione tecnologica proprie delle nano-immagini, le immagini scientifiche sono sempre il risultato di una manipolazione dell’oggetto naturale rappresentato. Un’immagine scientifica non può essere una mera copia della natura, perché è sempre parte di una praxis sperimentale il cui obiettivo è comprendere un fenomeno naturale, non solo riprodurlo. Per corroborare questa idea analizzo alcune pratiche concrete di significazione di immagini scientifiche, prendendo in esame documenti scritti (analisi semiotica di un articolo di radiologia) e pratiche materiali (etnografia di laboratorio riguardante l’interpretazione di immagini di elettroforesi in biologia molecolare e descrizione di un caso di significazione di immagini di microscopia elettronica). Questa analisi permette di fare tre osservazioni: (1) Il processo di significazione delle immagini scientifiche è un processo distribuito; (2) Le immagini scientifiche possono essere considerate strumenti di ricerca, nel senso che scienziati e medici le manipolano in varie forme al fine di esplorare aspetti diversi del loro oggetto di studio; (3) Le immagini scientifiche vanno comprese come fenomeni artificiali controllati prodotti allo scopo di ridefinire la visibilità degli oggetti naturali. Per approfondire meglio quest’ultima idea, nel capitolo finale introduco il concetto di fenomenotecnica sviluppato da Gaston Bachelard. La nozione di fenomenotecnica non può essere applicata direttamente all’imaging medico, ma alcuni degli elementi che caratterizzano il concetto bachelardiano offrono spunti importanti per pensare l’imaging medico. Il primo di questi elementi è l’idea che per studiare un fenomeno naturale, lo scienziato deve innanzitutto trasformarlo in un oggetto scientifico. Il secondo elemento, strettamente legato al primo, è che l’esperienza scientifica è necessariamente mediata, e tale mediazione ha un carattere intellettuale e materiale. Questo significa che la costruzione di strumenti e lo sviluppo di tecnologie non sono un prodotto della scienza, ma piuttosto un elemento interno al processo scientifico. La tecnologia è integrata nella scienza, perché la nostra apprensione? scientifica del mondo è necessariamente mediata da strumenti. Gli strumenti, a loro volta, sono materializzazioni di un vasto corpo di conoscenze e pratiche scientifiche (nel caso dell’imaging digitale tale sapere ha un carattere eminentemente matematico). Scienza e tecnologia, dunque, si costituiscono reciprocamente. A partire da queste considerazioni propongo un descrizione dell’imaging medico in termini di fenomenotecnica, utilizzando tale concetto come parola chiave attorno alla quale riorganizzare le idee discusse in precedenza. In primo luogo ricorro al concetto di fenomenotecnica per spiegare come le immagini diagnostiche mediano l’esperienza sensoriale e intellettuale del medico. Successivamente descrivo le immagini diagnostiche in termini di fenomeni artificiali (riconfigurazione visiva di segnali non visivi) che funzionano come simulazioni del corpo del paziente e che materializzano ambiti della conoscenza differenti (dalla medicina alla fisica, passando per l’ingegneria). Infine, mostro che la significazione corretta ed efficace di un’immagine diagnostica richiede una fenomenotecnica dell’osservatore. Per riconoscere i segni di malattia in un’immagine dell’interno del corpo è necessario padroneggiare le regole implicite ed esplicite che permettono di dare senso al nuovo dominio sensoriale prodotto dalla tecnologia. Ciò implica un abbandono dei modi spontanei di percezione-significazione e il passaggio attraverso un processo educativo che modula le capacità percettive. L’osservatore specializzato è un osservatore che ha preso parte a un processo di formazione che trasforma profondamente la visione naturale, inserendo l’atto del guardare all’interno di una vasta rete epistemica che include conoscenze teoriche e pratiche concrete.
The objective of this dissertation is to contribute to the development of an epistemology of medical imaging. My central thesis is that medical imaging does not merely produce more or less accurate pictures of the inner organs, it rather transforms the living body into a scientific object by changing its very visibility. The imaging apparatus turns the body into a visual object that can be observed under experimental conditions: unlike the real body, it can be filed, retrieved, shared, measured and manipulated in several ways. This main thesis is accompanied by two others: first, diagnostic images, as all scientific images, are actual cognitive instruments, epistemic objects inscribed within theoretical contexts and experimental practices. Second, an image of the inner body has diagnostic meaning and value only in the scope of a specific conceptualization of the body and its ailments. Accordingly, if we are to develop an epistemology of medical imaging, we cannot limit our analysis to diagnostic images qua images, we also have to understand them qua diagnostic instruments. This is why at in the first chapter of the dissertation I take into examination the historical and conceptual conditions of possibility of radiography -- the first medical imaging technology, invented in 1895. My aim is to understand what medical theories and practices had to be at work in the nineteenth century, for those shadow-images produced by the X-ray apparatus to be perceived and employed as diagnostic devices. I argue that the diagnostic relevance of radiography is rooted in the conceptualization of body, disease and diagnosis put forward by clinical anatomy already at the end of the eighteenth century. I also defend the idea that the stethoscope, developed in 1816, was the material and intellectual predecessor of medical imaging, because it introduced a primitive form of mediated perception in medical diagnosis, and allowed the clinician to explore from the outside the inner body of the living patient, extracting signs of illness. The stethoscope was only the first of a vast array of instruments invented in the nineteenth century to visualize different aspects of the inner morphology and physiology of the living body. Each of these instruments fulfilled specific diagnostic aims and posed distinct epistemological problems, but all of them shared some commonalities: they were meant to replace the subjective sensations of patients and doctors with objective indices of health and disease; they created visual records of the inner body that could be filed, retrieved and shared among physicians; they required the development of a specialized language agreed upon by a community of experts; they created a progressive physical separation between the body of the patient and the body of the physician. It was in this complex scenario of medical practices, objects, images and ideas that radiography appeared and progressively acquired its diagnostic function. In the second chapter I take into account the early developments of medical photography in order to understand how the first technology for the production of mechanical images entered and influenced the domain of medicine. The main theoretical references in this chapter are Charles Sanders Peirce's semiotics, in particular, his classification of signs in indices, icons and symbols, and Walter Benjamin's reflections on the photographic series (mechanical production and reproduction of an image and of the body it represents), on the intrinsic analytic and dissecting potential of photography (the photographer as a surgeon), and on the optical unconscious (photography as a prosthesis that enriches and transforms our sensorial experience). Drawing on these authors, and analyzing the works of early physicians-photographers in psychiatry, dermatology, neurology and physiology, I show that the photographic series collected in medical journals, manuals and hospital archives, produced a clinical gaze in the Foucauldian sense. I also argue that the photographic series was part of a larger experimental apparatus, which encompassed the patient, the camera and the observer, and whose aim was to turn the body and disease into a visual object available for scientific analysis. In the third chapter I discuss the problem of the invisible referent, that is, I analyze the processes whereby photographs that reveal invisible phenomena are endowed with meaning. This is likely to be the fundamental problem of all scientific imaging. When the referent of a picture is invisible, the iconic mode of signification fails, because in this case the image produced by the mechanical or electronic apparatus does not look like anything we already know, it resembles nothing. So, how do we know that the object we see in the photograph -- e.g., a cell or a tubercular lesion -- is really there and does really look like that? Drawing on the theoretical analysis developed in the previous chapter, I maintain that the visualization of the invisible entails a peculiar combination of the indexical, iconic and symbolic modes of signification. My reasoning opposes Lorraine Daston and Peter Galison's idea of mechanical objectivity, and demonstrates that their notion of mechanical objectivity as the moralizing suppression of subjectivity is a caricature of the actual ideas and practices developed by the scientists of the nineteenth century to deal with the problem of visualizing the invisible. The argument is articulated in three moments, corresponding to the analysis of the problem of objectivity and image signification in microphotography, chronophotography, and radiography. In the fourth chapter I argue that images are cognitive tools and that representation and observation are never an act of automated repetition, they always entail a creative component. As in the previous chapter, part of my discourse is built in contrast with Daston and Galison, challenging their claims concerning the passive nature of representation. For these authors, until the development of digital technologies for image manipulation, scientific images were mere re-presentations of the world, focused on copying nature. Computer images, on the contrary, are presentations, because the observer can virtually manipulate them so that they show the object in ever changing ways. I criticize this classification of scientific images with historical and theoretical arguments. From the historical point of view, I show that at least since the sixteenth century there have been attempts to create images that can be actually manipulated by the observer. From the theoretical perspective, I draw on a variety of literature spanning from art theory to neuroscience, to demonstrate that the very notion of a passive representation is unsustainable, because images always engage the observer in an embodied act of perception, which elicits not only visual, but also tactile sensations and motor reactions. Moreover, I argue that Daston and Galison's emphasis on nanoimaging as the only technology that allows manipulating the object of study during the process of image production is misleading. In fact, even when they do not reach the peaks of technological sophistication that characterizes nanoimages, scientific images are the result of some manipulation of the natural object they represent. A scientific image cannot be a passive copy of nature, because it is part of an experimental praxis, whose goal is to understand natural phenomena, not just to reproduce them. To corroborate this idea I explore actual scientific practices of image signification, taking into account written documents (semiotic analysis of a radiology article) and material practices (laboratory ethnography describing the interpretation of electrophoresis images in a molecular biology laboratory, and description of an example of signification of electron microscopy pictures). From this analysis three remarks can be put forward: (1) the process of signification of scientific images has a distributed character, because it can involve different persons, objects and activities; (2) scientific images can be considered experimental tools, in the sense that scientists and physicians handle them in several forms in order to explore different aspects of their object of study; (3) scientific images are to be understood as controlled, artificial phenomena produced with the aim of redefining the visibility of natural objects. In order to clarify this latter idea, in the final chapter I introduce Gaston Bachelard's concept of phenomenotechnique. Although the idea of phenomenotechnique cannot be directly applied to medical imaging, there are two characterizing elements of this concept that provide important insights for conceptualizing medical imaging. The first is the idea that in order to study a natural phenomenon, scientists must previously transform it into a scientific object. The second, closely related to the former, is that scientific experience is by necessity mediated, and such mediation has both an intellectual and material character. This means that the development of instruments and new technologies is not a second-order product of science, it is part and parcel of the scientific process. Technology is embedded into science, because our scientific grasping of the world is necessarily mediated by instruments; scientific instruments, in turn, are materializations of a vast body of scientific knowledge and practices (in the case of digital imaging this knowledge has an eminently mathematical character). Thus, science and technology are reciprocally constituted. On these grounds I propose a description of medical imaging in terms of phenomenotechnique, using this concept as a key-word around which to reorganize the ideas previously discussed. Firstly, I resort to the concept of phenomenotechnique to gain insights into how diagnostic images mediate the physician's sensory and intellectual experience. Second, I give an account of diagnostic images as artificial phenomena (visual reconfigurations of non-visual signals) that work as simulations of the patient's body, and that reify different domains of knowledge (from medicine to physics and engineering). Finally, I argue that the proper and efficient signification of a diagnostic image requires a phenomenotechnique of the observer. To recognize the signs of disease in an image of the inner body, one has to master the explicit and implicit rules necessary to make sense of the novel sensory domain produced by the technological apparatus. This implies abandoning spontaneous modes of perception and signification to engage in a process of educated perception. The expert viewer goes through a formal and informal training that deeply transforms natural vision, by placing the act of watching within a wide epistemic network that encompasses both theoretical and practical knowledge.

Necrotising enterocolitis (NEC) is a potentially devastating intestinal inflammation of multifactorial aetiology in premature or otherwise vulnerable neonates. Because of the broad spectrum of presentations, diagnosis and timing of surgical intervention may be challenging, and imaging needs to be an integrated part of management. The first four studies included in this thesis used routinely collected, nationwide register data to describe the incidence of NEC in Sweden 1987‒2009, its variation with time, seasonality, space-time clustering, and associations with maternal, gestational, and perinatal factors, and the risk of intestinal failure in the aftermath of the disease. Early infant survival increased dramatically during the study period. The incidence rate of NEC was 0.34 per 1,000 live births, rising from 0.26 per 1,000 live births in the first six years of the study period to 0.57 in the last five. The incidence rates in the lowest birth weights were 100‒160 times those of the entire birth cohort. Seasonal variation was found, as well as space-time clustering in association with delivery hospitals but not with maternal residential municipalities. Comparing NEC cases with matched controls, some factors, positively associated with NEC, were isoimmunisation, fetal distress, caesarean section, persistent ductus arteriosus, cardiac and gastrointestinal malformations, and chromosomal abnormalities. Negative associations included maternal pre-eclampsia, maternal urinary infection, and premature rupture of the membranes. Intestinal failure occurred in 6% of NEC cases and 0.4% of controls, with the highest incidence towards the end of the study period. The last study investigated current practices and perceptions of imaging in the management of NEC, as reported by involved specialists. There was great consensus on most issues. Areas in need of further study seem mainly related to imaging routines, the use of ultrasound, and indications for surgery. Developing alongside the progress of neonatal care, NEC is a complex, multifactorial disease, with shifting patterns of predisposing and precipitating causes, and potentially serious long-term complications. The findings of seasonal variation, spacetime clustering, and negative associations with antenatal exposure to infectious agents, fit into the growing understanding of the central role of bacteria and immunological processes in normal maturation of the intestinal canal as well as in the pathogenesis of NEC. Imaging in the management of NEC may be developed through future studies combining multiple diagnostic parameters in relation to clinical outcome.

Objectives: We ought to compare the effect of transcatheter aortic valve replacement (TAVR) and surgical aortic valve replacement (SAVR) on right ventricular systolic function (RVSF) in high risk patients with severe aortic stenosis (AS). Methodology: Data Source. PubMed, EMBASE, Cochrane library, and references of selected articles. Study Endpoints. Transthoracic echocardiography was utilized to assess the change in RVSF post TAVR versus SAVR using tricuspid annular plane systolic excursion (TAPSE), and Fractional area change (RVFAC). Statistical analyses. Random effect model on standardized mean difference (Hedges; g) were used together with heterogeneity assessment. Result: We included 485 patients from five single-center observational studies. Comparing TAVR with SAVR, TAVR resulted in better improvement in RVSF [TAPSE (g=2.88, SE=0.63, P<0.001, Q=73.18, /²=94.53, r=0.65), and RVFAC (g=0.91, SE=0.16, P<0.001, Q=2.39, /²=16.61), r=0.65]. Conclusion: Compared with SAVR, TAVR is preferred aortic intervention in patients with severe symptomatic AS and RV systolic dysfunction.

Autosomal Dominant Polycystic Kidney Disease (ADPKD) is among the most prevalent life-threatening genetic conditions. Despite this, no approved medical therapies exist to treat the disease. Until the recent past, no methods could reliably measure the course of the disease far in advance of end stage renal disease (ESRD). As normal tissue is progressively destroyed or blocked by enlarging cysts, remaining nephrons compensate in a process called hyperfiltration. This beneficial physiological response confounds tests of renal function. Thus, potential interventions could not be tested against a reliable measurement of disease progression.

However, progressive changes are visually apparent on medical imaging examinations throughout the course of ADPKD. The search for ADPKD proxy biomarkers is now focused on quantitative imaging, or the extraction of information from medical images for purposes of diagnosis or disease tracking. Recent studies from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK)- sponsored Consortium for Radiologic Imaging Studies of Polycystic Kidney Disease (CRISP) showed Total Kidney Volume (TKV) is a usable quantitative imaging biomarker which can track disease in the early, asymptomatic phase and register measurable changes in as little as 12 months. These findings launched several new trials into potential ADPKD therapies.

Advanced analysis of polycystic kidney images, however, has never been done. The method CRISP used to extract TKV was stereology, an efficient means to estimate volume. However, stereology was tradi- tionally a dead end for further advanced analysis. TKV is useful for clinical trials and large population-based studies, but cannot accurately predict disease progression or stratify risk due to known out- lier cases. Thus, the utility of TKV for individual patient prognosis is limited. This work builds upon stereology data, describing a reliable and accurate new semi-automatic method to fully segment images us- ing only labeled stereology grids. Then, two new second generation quantitative imaging biomarkers are introduced and analyzed: Cyst- Parenchyma Surface Area (CPSA) and cyst concentration. These new physiologically motivated biomarkers will complement or potentially replace TKV in efforts to bring quantitative imaging to individual patients.

The goal of this body of work is to enable a pathway for efficient advanced image analysis in ADPKD, never before attempted in this dis- order, and to define new quantitative imaging biomarkers which will complement or replace existing ones in hopes of making individualized disease tracking for ADPKD patients a reality.

Prostate and breast cancer are the most common cause of cancers in men and women, respectively. Medical imaging plays an important role in breast and prostate cancer detection and evaluation. Then to prove that our web-based medical application could be applied in different medical disciplines, the main part of this thesis is the implementation of two frameworks as a Java applet interface designed as a web-based tool in the domains of mammography from X-rays in radiology, and of prostate imaging from an Magnetic Resonance Imaging (MRI). This aims to facilitate the diagnosis of new mammographic and prostate cases by providing a set of image processing tools that allow a better visualization of the images, and a set of drawing tools used to annotate the suspicious regions (overlays). Each annotation allows including the attributes considered by the experts when issuing the final diagnosis. The overall set of overlays is stored in a database as eXtensible Markup Language (XML) files associated with the original images. Finally, an exhaustive evaluation of the results is also discussed in this thesis. For the application on mammography, the experimental study is performed in order to evaluate the scalability, complexity and response speed at the proposed tool. For the application on MRI of prostate cancer, the evaluation focused on the decrease of the variability of the expert assessments when collaborative work is performed. To conclude, a new architecture with the main goal of managing patient databases with potentially multi-modal imaging is presented such as for an MRI of the prostate cancer and evaluation from potentially several experts.
En aquesta tesi es realitza una revisió bibliogràfica de les principals publicacions recents en els últims anys en aplicacions mèdiques basades en web. Aquest estudi analitza els avantatges i inconvenients dels treballs d’investigació en el camp de la imatge mèdica, així com les arquitectures de base de dades per a la gestió d’imatges digitals. La part principal d’aquesta tesi és la implementació d’una eina basada en la web amb la finalitat de demostrar la integritat i aplicació en diferents disciplines mediques. En aquest sentit, l’aplicació proposada en aquest projecte de tesis ha sigut implementada com a eina d’ajuda al diagnòstic de càncer de mama i pròstata. L’objectiu és facilitar el diagnòstic proporcionant un conjunt d’eines de processat d’imatge que permetin una millor visualització de les imatges, i un conjunt d’eines d’anotació de regions sospitoses o malignes (superposicions). Cada anotació permet incloure tots els atributs i especificacions considerades pels experts a l’emetre el diagnòstic final. S’han dissenyat diferents arquitectures per a la gestió de base de dades (per exemple PACS per emmagatzemar imatges monogràfiques). Per altra banda, el conjunt global d’anotacions s’emmagatzemen en una base de dades d’arxius XML associats a les imatges originals. Conseqüentment, aquesta nova arquitectura es presenta amb l’objectiu d’obtenir una base de dades de casos diagnosticats i validats per radiòlegs experts per a la formació de radiòlegs novells. Finalment, conclusions i noves línies d’investigació associades al projecte com a treball futur són presentades en aquesta tesi.

Medisinsk ultralydavbildning er et relativt rimelig verktøy som er i utstrakte bruk på dagens sykehus og tildels også legekontor. En underliggende antakelse ved dagens avbildningsteknikker er at vevet som skal avbildes i grove trekk er homogent. Det vil i praksis si at de akustiske egenskapene varierer lite. I tilfeller der denne forutsetningen ikke holder vil resultatet bli betraktlig reduksjon av bildekvaliteten. Prosjektet har fokusert på hvordan man best mulig kan korrigere for denne kvalitetsforringelsen. Arbeidet har resultert i et styrket teoretisk rammeverk for modellering, programvare for numerisk simulering. Rammeverket gir en felles forankring for tidligere publiserte metoder som "time-reversal mirror", "beamsum-correlation" og "speckle brightness", og gir derfor en utvidet forståelse av disse metodene. Videre har en ny metode blitt utviklet basert på egenfunksjonsanalyse av et stokastisk tilbakespredt lydfelt. Denne metoden vil potensielt kunne håndtere sterk spredning fra områder utenfor hovedaksen til ultralydstrålen på en bedre måte enn tidligere metoder. Arbeidet er utført ved Institutt for matematiske fag, NTNU, med professor Harald Krogstad, Institutt for matematiske fag, som hovedveileder og professor Bjørn Angelsen, Institutt for sirkulasjon og bildediagnostikk, som medveileder.

Ultra-wideband (UWB) radio techniques have long promised good contrast and high resolution for imaging human tissue and tumours; however, to date, this promise has not entirely been realised. In recent years, microwave imaging has been recognised as a promising non-ionising and non-invasive alternative screening technology, gaining its applicability to breast cancer by the significant contrast in the dielectric properties at microwave frequencies of normal and malignant tissues. This thesis deals with the development of two novel imaging methods based on UWB microwave signals. First, the mode-matching (MM) Bessel-functions-based algorithm, which enables the identification of the presence and location of significant scatterers inside cylindrically-shaped objects is introduced. Next, with the aim of investigating more general 3D problems, the Huygens principle (HP) based procedure is presented. Using HP to forward propagate the waves removes the need to apply matrix generation/inversion. Moreover, HP method provides better performance when compared to conventional time-domain approaches; specifically, the signal to clutter ratio reaches 8 dB, which matches the best figures that have been published. In addition to their simplicity, the two proposed methodologies permit the capture of a minimum dielectric contrast of 1:2, the extent to which different tissues, or differing conditions of tissues, can be discriminated in the final image. Moreover, UWB allows all the information in the frequency domain to be utilised, by combining information gathered from the individual frequencies to construct a consistent image with a resolution of approximately one quarter of the shortest wavelength in the dielectric medium. The power levels used and the specific absorption rates are well within safety limits, while the bandwidths satisfy the UWB definition of being at least 20% of the centre frequencies. It follows that the methodologies permit the detection and location of significant scatterers inside a volume. Validation of the techniques through both simulations and measurements have been performed and presented, illustrating the effectiveness of the methods.

Mestrado em Engenharia de Computadores e Telemática
Nos últimos anos, a imagem médica em formato digital tem sido uma ferramenta cada vez mais importante quer para o diagnóstico médico quer para o auxílio ao tratamento. Assim, equipamentos de aquisição digital e repositórios de imagem médica são cada vez mais comuns em instituições de saúde, podendo até haver mais que um repositório numa instituição. No entanto, esta proliferação de repositórios leva a que a informação esteja dispersa nos vários locais. Esta dispersão da informação juntamente com as diferenças no armazenamento entre instituições são claros obstáculos à pesquisa e acesso integrado a essa informação. Esta dissertação visa o estudo da tecnologia Peer-to-Peer de forma a minimizar os problemas associados à dispersão e heterogeneidade da informação.
In the last years, digital medical imaging has been an increasingly important tool for both medical diagnostic and treatment assistance. Therefore, digital image acquisition equipments and medical imaging repositories are more and more common in a healthcare institution, being possible even more than one repository in one institution. However, this proliferation of repositories leads to dispersion of data between many places. This data dispersion associated with differences in the data storage between institutions are evident obstacles to the search for medical data. This dissertation aims to the study of the Peer-to- Peer technology in order to minimize the problems related to the dispersion and heterogeneity of medical data.

Mestrado em Engenharia de Computadores e Telemática
A imagem médica em formato digital é um elemento presente nas mais variadas instituições prestadoras de cuidados de saúde, afirmando-se como um imprescindível elemento de suporte ao diagnóstico e terapêutica médica. Nesta área, os formatos e processos de armazenamento e transmissão são definidos pela norma internacional DICOM. Um ficheiro deste tipo contempla, para além da imagem (ou vídeo), um conjunto de meta-dados que incluem informação dos pacientes, dados técnicos relativos ao estudo, dose de radiação, relatório clínico, etc. Um dos maiores problemas associados aos repositórios de imagem médica está relacionado com a grande quantidade de dados produzidos que impõe desafios acrescidos ao armazenamento e transporte da informação, em particular em cenários distribuídos e de grande produção de estudos imagiológicos. Esta dissertação tem como objetivo estudar e explorar soluções que permitam a integração do conceito de pertença e controlo de acesso em arquivos de imagem médica, possibilitando a centralização de múltiplas instâncias de arquivos. A solução desenvolvida permite associar permissões a recursos e delegação a terceiras entidades. Foi desenvolvida uma interface programática de gestão da solução proposta, disponibilizada através de web services, com a capacidade de criação, leitura, atualização e remoção de todos os componentes resultantes da arquitetura.
The production of medical images in digital format has been growing in the most varied health care providers, representing at this moment an important and indispensable element for supporting medical decisions. In medical imaging area, the formats and transmission processes are defined by the international DICOM standard. A file in this format contains image pixel data but also a set of metadata, including information about the patient, technical data related to the study, dose of radiation, clinical report, etc. One of the biggest problems associated with medical imaging repositories is related to the large amount of data produced that poses additional challenges to the transport and archive of information, particularly in distributed environments and laboratories with huge volume of examinations. This dissertation aims to study and explore solutions for the integration of ownership concept and access control over medical imaging resources, making possible the centralization of multiple instances of repositories. The proposed solution allows the association of permissions to repository resources and delegation of rights to third entities. It was developed a programmatic interface for management of proposed services, made available through web services, with the ability to create, read, update and remove all components resulting from the architecture.

A "chirp" is a frequency modulated signal widely used in ultrasound imaging to increase the signal-to-noise ratio and penetration depth. In medical ultrasound imaging, resolution and penetration are two major criteria that are inversely proportional. Because of this inverse relation, short duration pulses cannot achieve a high resolution with good penetration. The reasons for this trade-off are the decrease in signal energy due to shorter pulse duration and the attenuation in tissue, which increases with the excitation frequency. The chirp coded excitation however can increase the total transmitted energy using longer pulse durations, while the resolution can be recovered by decoding on receive. Therefore, chirp signals offer potential advantages over single carrier short duration pulses for medical imaging. This work addresses the possible problems encountered in medical ultrasound imaging with chirps and offers new solutions to these problems in terms of signal processing. These proposed solutions are then applied to three major categories of medical ultrasound imaging; hard-tissue ultrasound imaging, soft-tissue ultrasound imaging and contrast-enhanced ultrasound imaging. The application of coded excitation in medical ultrasound imaging is the main motivation behind this work. Therefore, the concepts of frequency modulation and matched filtering are introduced first, and ultrasound specicific problems for pulse compression of chirps are discussed. Examples are given on specific applications and circumstances, where the performance of the traditional pulse compression techniques drops significantly. Alternate methods of pulse compression and filtering of frequency modulated chirps using the Fractional Fourier transform (FrFT) and the Fan Chirp transform (FChT) are presented. Rather than restricting the chirp analysis in the time or frequency domain; these proposed methods transform the signal of interest into a new domain, which is more suitable to analyse frequency modulated chirps.

An important goal in medical imaging is the assessment of image quality in a way that relates to clinical efficacy. An objective approach is to evaluate the performance of diagnosis for specific tasks, using ROC analysis. We shall concentrate here on classification tasks. While many factors may confine the performance achieved for these tasks, we shall investigate two main limiting factors: image blurring and object variability. Psychophysical studies followed by ROC analysis are widely used for system assessment, but it is of great practical interest to be able to predict the outcome of psychophysical studies, especially for system design and optimization. The ideal observer is often chosen as a standard of comparison for the human observer since, at least for simple tasks, its performance can be readily calculated using statistical decision theory. We already know, however, of cases reported in the literature where the human observer performs far below ideal, and one of the purposes of this dissertation is to determine whether there are other practical circumstances where human and ideal performances diverge. Moreover, when the complexity of the task increases, the ideal observer becomes quickly intractable, and other observers such as the Hotelling and the nonprewhitening (npw) ideal observers may be considered instead. A practical problem where our intuition tells us that the ideal observer may fail to predict human performance occurs with imaging devices that are characterized by a PSF having long spatial tails. The investigation of the impact of long-tailed PSFs on detection is of great interest since they are commonly encountered in medical imaging and even more generally in image science. We shall show that the ideal observer is a poor predictor of human performance for a simple two-hypothesis detection task and that linear filtering of the images does indeed help the human observer. Another practical problem of considerable interest is the effect of background nonuniformity on detectability since, it is one more step towards assessing image quality for real clinical images. When the background is known exactly (BKE), the Hotelling and the npw ideal observers predict that detection is optimal for an infinite aperture; a spatially varying background (SVB) results in an optimum aperture size. Moreover, given a fixed aperture size and a BKE, an increase in exposure time is highly beneficial for both observers. For SVB, on the other hand, the Hotelling observer benefits from an increases in exposure time, while the npw ideal observer quickly saturates. In terms of human performance, results show a good agreement with the Hotelling-observer predictions, while the performance disagrees strongly with the npw ideal observer.

Cardiovascular disease is one of the leading causes of the morbidity and mortality in the western world today. Many different imaging modalities are in place today to diagnose and investigate cardiovascular diseases. Each of these, however, has strengths and weaknesses. There are different forms of noise and artifacts in each image modality that combine to make the field of medical image analysis both important and challenging. The aim of this thesis is develop a reliable method for segmentation of vessel structures in medical imaging, combining the expert knowledge of the user in such a way as to maintain efficiency whilst overcoming the inherent noise and artifacts present in the images. We present results from 2D segmentation techniques using different methodologies, before developing 3D techniques for segmenting vessel shape from a series of images. The main drive of the work involves the investigation of medical images obtained using catheter based techniques, namely Intra Vascular Ultrasound (IVUS) and Optical Coherence Tomography (OCT). We will present a robust segmentation paradigm, combining both edge and region information to segment the media-adventitia, and lumenal borders in those modalities respectively. By using a semi-interactive method that utilizes "soft" constraints, allowing imprecise user input which provides a balance between using the user's expert knowledge and efficiency. In the later part of the work, we develop automatic methods for segmenting the walls of lymph vessels. These methods are employed on sequential images in order to obtain data to reconstruct the vessel walls in the region of the lymph valves. We investigated methods to segment the vessel walls both individually and simultaneously, and compared the results both quantitatively and qualitatively in order obtain the most appropriate for the 3D reconstruction of the vessel wall. Lastly, we adapt the semi-interactive method used on vessels earlier into 3D to help segment out the lymph valve. This involved the user interactive method to provide guidance to help segment the boundary of the lymph vessel, then we apply a minimal surface segmentation methodology to provide segmentation of the valve.

xi, 56 p. : ill. (some col.) A print copy of this thesis is available through the UO Libraries. Search the library catalog for the location and call number.
Image processing is a powerful tool for increasing the reliability and reproducibility of disease diagnostics. In the hands of pathologists, image processing provides quantitative data from histological images which supplement the qualitative data currently used by specialists. This thesis presents a novel method for analyzing digitized images of hematoxylin and eosin (H&E) stained histology slides to detect and quantify inflammatory polymorphonuclear leukocytes to aid in the grading of acute inflammation of the placenta as an example of the use of image processing in aid of diagnostics. Methods presented in this thesis include segmentation, a novel threshold selection technique and shape analysis. The most significant contribution is the automated color threshold selection algorithm for H&E stained histology slides which is the only unsupervised method published to date.
Committee in charge: Dr. John Conery, Chair; Dr. Matthew J. Sottile

A prospective intervention study, using clinical practice improvement (CPI) methodology, was undertaken to reduce unnecessary x-ray examinations in the early management of patients presenting to the Emergency Department (ED). This was achieved through raising the awareness of medical and allied health staff to medical radiation by means of clinical education and implementing evidence based diagnostic imaging requisition. The main study was conducted in the ED of a public hospital located in the western Sydney, Australia. A second hospital within the area health service, with similar bed size, activity levels and demographics, was used as the control site. The first phase intervention raised the awareness of the health professionals to medical radiation. The second phase intervention used CPI methodology to attain efficient clinical practices so as to eliminate unnecessary examinations and requests. A multi-disciplinary CPI Project Team involved in the process of imaging examination requisition was empowered to improve the appropriateness of the requested examination utilisation. This it achieved mainly through the implementation of evidence based clinical decision rules and imaging guidelines. An additional method of validating the outcomes was provided through the simultaneous rollout of the interventions at another hospital within the same area health service. At the completion of the study, unnecessary examinations such as Skull, Ribs, Nasal Bone and Kidney Ureter Bladder (KUB) requests at the intervention Hospital site were significantly reduced by 92.6% (p (less than) 0.0001), whereas at the control site Hospital there was minimal reduction which was found to be not statistically significant (p=0.2110). Other frequently requested examinations such as Ankle, Knee and Spine requests were marginally though significantly reduced at the intervention Hospital by 22.7% (p (less than) 0.001), whereas at the control site Hospital the reduction was similarly found to be not significant (p=0.1055). Most importantly, the overall x-ray requisition for every 100 ED presentations at Hospital ‘B’ was reduced by 27%. The results of this study, demonstrated that Radiation Awareness educational programs, targeting medical and allied health staff, will reduce the unnecessary requisition of examinations found not to contribute to the process of the patient’s clinical management. The use of a CPI project approach was found to be important in the process of establishing, implementing and sustaining the achieved improvements, and in particular, the rules and guidelines of evidence based imaging requisition. Importantly, the study also confirmed that the CPI methodology that had been used for the main intervention was adaptable to other organisations when it was found to have been successfully rolled out at another hospital. This confirmed that the main outcomes of the investigation could be generalised to other health facilities. There was an immediate reduction in the requisition of unnecessary examination similar to the results at the main intervention hospital. The implementation and adoption of the CPI intervention across the health care system in general could significantly reduce unnecessary x-ray examinations, saving significant health care resources, and sparing patients from potential cancer risks associated with avoidable exposure to ionising medical radiation.

Thesis (Ph.D. (Health))--University of Western Sydney, 2007.
A thesis presented to the University of Western Sydney, College of Health and Science, School of Biomedical and Health Sciences, in fulfilment of the requirements for the degree of Doctor of Philosophy (Health). Includes bibliographies.

Thesis (M.S.)--Worcester Polytechnic Institute.
Keywords: 3d ultrasound; ultrasound; image processing; image segmentation; 3d image segmentation; medical imaging Includes bibliographical references (p.142-148).

Patient motion either internal (organ motion) or external (body movement) can produce artefacts that can adversely affect nuclear medicine imaging. Motion artefacts can impair diagnostic information and potentially affect the image findings and prognosis for patients. The goal of this work was to investigate the effect of motion on nuclear medicine imaging and to improve image quality, lesion detectability, and tumour volume delineation by applying motion correction techniques. To investigate the effects of motion under controlled simulated conditions, a three dimensional phantom drive system was designed and constructed suitable for use with planar, SPECT, PET and CT scanners. The system was used with a range of nuclear medicine phantoms for testing proof of principle with planar, SPECT and PET imaging prior to undertake further work involving patients. Planar phantom and patient 99mTc_DMSA studies demonstrated improvements in image quality by the application of motion correction techniques. A comparison between the motion correction software using dynamic frame and list mode data showed that "MOCO" software with the use of the list mode data produced the best quantification results with phantom data, whereas determining the best approach was more difficult with patient data. The potential of using list mode data as an improved method of combining data into frames for subsequent analysis was demonstrated. Motion correction techniques would appear to offer great potential in lung imaging. Respiratory gated SPECT phantom studies have been carried out to simulate the visualisation of small defects in the lung. The CNRs and alternative free response receiver operating characteristic (AFROC) analysis have demonstrated that summing the gated data after the application of motion correction software significantly improved image quality, observer confidence and small defect detectability (less than 20 mm, p=O.0002). The results of these studies have shown the promising role of "MCFLIRT" software as a motion correction tool with gated SPECT data. Tumour volume delineation was investigated on PET images both with and without motion. The accuracy and consistency of the gradient-based software method for segmentation in PET images, which is commercially available from Mimvista Ltd was investigated. The results of comparing the measured volumes to the true volumes indicated significant differences (p=O.0005). It was found that the Signal:Background ratio and registering the PET to the CT data have significant effects on volume measurements, whereas, the effect of using different grey scale and plane of orientation were not found to have significant effects on the volume measurement. Motion correction techniques also showed to be potentially beneficial in PET imaging. Improvement in volume measurement as a result of summing the motion corrected gated data was demonstrated. The results of these studies have also shown the promising role of "MCFLIRT" as a motion correction tool with gated PET data.

Early diagnosis of disease and developing targeted therapeutics are two major goals of medical research to which nanotechnology can contribute a variety of novel approaches and solutions. This work utilized an optical phenomenon specific to metallic nanoparticles, surface-enhanced Raman spectroscopy (SERS), as a nanomedicine research tool to aid in the progression toward these goals. Single-particle SERS studies were streamlined to identify particles or aggregates with potentially high enhancement factors (EFs) for applications requiring ultrasensitive and possibly single-molecule detection. SERS was used to probe the changes in surface chemistry of nanoparticles for optimizing nanomedicine applications. Fundamental SERS imaging parameters were identified, and a new algorithm for multiplexed SERS imaging was developed and tested. Novel particle-based contrast agents were also developed. Polystyrene hollow beads with a single hole on the surface were fabricated and used to encapsulate contrast agents for a variety of medical imaging modalities. Saline was encapsulated as a novel contrast agent for thermoacoustic tomography (TAT). Encapsulation of X-ray computed tomography (CT) and magnetic resonance (MR) imaging was also performed and tested.

The motivation for this work is a vision of widespread adoption of a priori quantitative epidemiological information for clinical decision-making, and can be seen as a quantitative large-scale extension of evidence-based medicine (EBM). Medical images can be seen as a spatially encoded map of physiological measurements that can be used to predict prognosis and to drive treatment plans. This paradigm can be very powerful and is driven by the recent big data revolution in computer science as well as the increasing availability of medical imaging modalities due to decreases in manufacturing costs. In order to achieve this overarching goal, three practical requirements must be reached and correspond to the parts of this thesis: Part A: Developing IT infrastructure and technology that enables the dataset to be properly collected and organized for analysis. Part B & C: Generation of functional (Part B) and structural (Part C) medical imaging contrast that are optimized for analysis. Part D: Pattern recognition techniques (including both image processing and machine learning techniques) to mine information from the large imaging datasets generated. As part of the thesis, I discuss my contribution to IT infrastructure (Part A) by developing a Short Message Service (SMS)-based system to control the clinically used Picture Archival and Communication System (PACS) (Ch.2) as well as an imaging study tool that categorizes patient imaging data for use in retrospective studies(Ch.3). I then go on to detail my work with functional neuroimaging of obesity using functional magnetic resonance imaging (fMRI)(Ch.4) and (Ch.5). Chapters 6-9 details my efforts at studying abnormal aging versus normal aging using diffusion MRI as well as applications of diffusion MRI to surgical planning. Chapters 10 discusses my work integrating diffusion MR with FLAIR MRI to investigate the properties of white matter lesions and how it can be used in the clinical setting. Chapter 11 then moves on to talk about my work modifying standard brain parcellation techniques to allow them to work with aged brains with large infarcts. Chapters 6-11 altogether represent my efforts in structural neuroimaging using MRI (Part C). The thesis then closes with capstone work in development staging using hand x-rays using fuzzy logic (Ch. 12 & 13). To close the work with Alzheimer's Disease (AD) and aging, we used machine learning techniques to predict disease progression based on a baseline MRI scan as well as higher order analysis of our diffusion MRI dataset by integrating MRI information with other clinical information such as neuropsychological tests, cardiovascular status. This is all in an effort to computationally explore the relationship between MRI measurements and clinical presentation of disease as measured by neuropsychological scores. Similarly with the Obesity work, we related fMRI activation differences between high and low calorie foods with non-imaging information such as insulin resistance (Ch. 16).

This study was conducted to determine if specialty areas are emerging in the magnetic resonance imaging (MRI) profession due to advancements made in the medical sciences, imaging technology, and clinical applications used in MRI that would require new developments in education/training programs and national registry examinations. In this exploratory study, statistical analysis incorporated the use of factor analysis and chi square. Factor analysis was used to group tasks performed by MRI technologists into factors to better identify emerging specialty areas within the MRI profession. Chi square was used to analyze the association between the tasks performed in (a) the employment setting, and (b) hospital size. Factor analysis identified four meaningful factors. The four named factors were: (a) Routine Imaging non-Central Nervous System Imaging; (b) Advanced Imaging; (c) Routine Imaging with Central Nervous System Imaging; and (d) Musculoskeletal and Spine Imaging. From the four named factors, three emerging specialty areas were identified: (a) central nervous system imaging; (b) vascular/cardiovascular imaging; and (c) musculoskeletal imaging. Chi square analysis identified 47 of the 78 tasks as being significant when finding an association between the employment setting and the frequency of tasks performed. Cramer's V was used to measure the strength of their association. The more complicated the procedure the more likely this procedure is performed in either a university or private hospital. Further, chi square analysis identified 42 of the 78 tasks as being significant when finding the association between the hospital size and the frequency of tasks performed. Gamma was used to measure the strength of their association. This means the larger the hospital, the more frequent the tasks were performed.