Current methods for treating and diagnosing spinal deformities caused by scoliosis are both surgically intensive and rarely allow for complete correction. This is mainly due to the fact that the diagnostic techniques used are rough estimates made by angles defined by observations of 2-D radiographs. By utilizing the latest software, our research is based on designing a tool that creates a 3-D representation of the spine. When creating a three-dimensional spinal model, it becomes possible to determine local curvature and local torsion values at each specific vertebrae. By manipulating these values at discrete locations on the spine, one can generate "virtual" spines in a three-dimensional environment. The Scoliosis Learning Tool includes algorithmic steps that determine the Lenke Classification of the "virtual" spines. The Lenke Classification is the most commonly accepted method for diagnosing spinal deformities. This patient building program will produce a group of spines with random values for curvature, torsion and initial spinal orientation. An algorithm within the software determines the Lenke Classification of each, and discards any curves that appear unnatural. By defining a metric that places an emphasis on certain geometric similarities, the software is able to define diameters of classification groups and separations between different classification groups. In turn it is possible to determine minor to major differences between spines within the same classification. In doing so, the opportunity exists to possibly find an undiscovered deformity that had previously fallen under another classification category. / Master of Science

This project was born out of the work performed by a group of researchers at the New Jersey Institute of Technology (NJIT) [1] [2] [3] working on interferometric imaging with a spiral array. Their investigation stopped at two dimensional imaging with a two dimensional array. In this thesis, their idea was developed further into the significantly more complex imaging with a three dimensional array. The general design of the NJIT [1] [2] [3] experiment was reproduced, studied, and modified in a manner that was theorized to enhance the experiment with the added ability to perform three dimensional imaging. The NJIT team [1] [2] [3] has developed their experiment to where they were able to accurately perform two dimensional imaging of two sources of equal intensity located at different distances from a spiral array. In this thesis, the equations used for two dimensional imaging are extrapolated into a three dimensional array application. This three dimensional imaging concept is simulated with MATLAB and the results presented and compared to the NJIT experimental results. [1] [2] [3] A proof of concept physical experiment is conducted and the results are compared to the MATLAB simulation. The results show that additional spatial information can be obtained from a three dimensional array that can enhance the information gleaned from images. / Master of Science

No description available.

University of Macau / Faculty of Science and Technology. / Department of Computer and Information Science

"Ultrasound has developed from 2D into 3D ultrasound in recent years. 3D ultrasound gives enhanced diagnostic capabilities and can make it easier for less trained people to interpret ultrasound images. In general there are two ways of getting a 3D ultrasound image : By using a 2D array scanner (giving 3D images directly) or by using a series of 2D scans and combine these scans to build a 3D volume. The only practical scanning technique that can be used for portable systems is freehand scanning that combines a series of 2D images. 3D images acquired by using a conventional ultrasound transducer and the freehand scanning technique are, however, often misaligned laterally and have unevenly spacing. These errors can be corrected if the position associated with each 2D image is known. Commercially available positioning systems use magnetic or optical tracking, but these systems are very bulky and not portable. We have proposed another way to get the position by tracking on the skin surface. This is done by obtaining digital images of the surface at a very high rate and then cross correlating each image to reveal the change in position. Accumulating these changes will then give the correct location (in two dimensions) relative to a starting point. Correct volume and surface rendering can therefore be achieved when a scan is done. A custom-made housing was made to mount an optical sensor to the ultrasound transducer. The optical sensor was placed in the housing and the hardware circuit from an optical mouse was used to interface to a USB interface. An implementation with an optical fiber was also made since this could fit easily to the transducer handle. In Windows a custom-made mouse driver was used to extract the position information from the sensor. This driver allowed multiple mouse devices in the system and removed the acceleration of the mouse, giving a correct transfer of the position. A DLL (Dynamic Link Library) was used to interface to a 3D ultrasound software called Sonocubic. Using the DLL and a custom modified version of Sonocubic 3D construction software has allowed a correct compensation of the acquired ultrasound images. To validate the accuracy of the optical sensor an optical mouse was placed in an XY-recorder to compare the acquired position with the actual position. The test revealed that the accuracy of the optical sensor is very high. A 55 mm movement of the sensor gave a deviation of 0.56 mm which is well within the expected result. A computer generated phantom was made to see if the compensation algorithm was working. The test revealed that the compensation algorithm and the software is working perfectly. Next a vessel phantom was scanned to see that the compensation algorithm (lateral compensation) was working in real life. The test showed that a correct lateral compensation was made. Finally 3D phantoms were custom made to test the accuracy of the system by estimation of a known volume. The system was able to estimate the volume in a phantom within an accuracy of 6 %. Performance of the system with direct imaging, using the optical sensor and a lens, was compared to an implementation with an optical fiber, two lenses and the optical sensor. The optical fiber was difficult to implement since the image contrast was degraded severely through the optical fiber and the lenses. This made it difficult for the correlation algorithm to function correctly and tracking could therefore not be done on a skin surface. Code for an FPGA was made in VHDL to extract the actual images from the optical sensor and display them directly on a computer screen. This was necessary to see how well the sensor was in focus. This proved to be a really useful tool for adjusting the optical system for maximal contrast. The optical tracking on a skin surface is a good way to assist a user doing a freehand scanning to get images without geometric distortion. Furthermore, it is the only real positioning system for a portable system. One requirement for this system is, however, that the object being scanned is flat and does not curve or vary vertically. For most applications this is not the case, and we are therefore proposing an implementation with microgyros that is able to give angle information as well. This would give the system a total of up to 5 instead of just 2 degrees of freedom. The status of this is currently that it can be easily implemented in the DLL, but it is not implemented in the 3D reconstruction software, Sonocubic."

Orientador: Alex Sandro Campos Maia / Coorientador: Marcos Chiquitelli Neto / Banca: Euclides Braga Malheiros / Banca: Tomas Norton / Banca: William Koury Filho / Banca: Vinicius de França Carvalho Fônseca / Resumo: A fim de utilizar a modelagem tridimensional (3D) para estimar área, volume e massa, bem como avaliar o impacto dos desvios envolvidos nas equações empíricas disponíveis na literatura, em animais, foram utilizados seis ovinos Corriedale tosqueados com massa entre 10 e 105 kg e cinco bovinos Nelore com massa entre 600 a 800 kg. Os animais foram treinados utilizando os princípios de habituação e condicionamento, a fim de reduzir a reatividade dos animais. Para medição da área de superfície, utilizaram-se basicamente três procedimentos: equação empírica (EE) para cada espécie, instrumento manual de medição (IMM) e modelagem 3D do animal por fotogrametria (AR) e escaneamento com sensor kinect (AS_KS). Foi realizada uma análise de covariância, sendo a massa a covariável e uma análise de regressão não linear foi ajustada como uma função potência (alometria). Analisando as médias de área de superfície, para ovinos, em relação aos procedimentos utilizados não houve diferença entre IMM e modelos 3D, bem como entre os modelos 3D, enquanto entre IMM e EE houve diferença significativa. Para os bovinos, observou-se que os procedimentos 3D foram, em média, superiores às estimativas de área de superfície pela EE (a=5%). Quando se plotou os dados de área de superfície, de ovinos, dos métodos testados em função da massa corporal, obteve-se aproximações empíricas distintas daquelas estimadas pela EE em mais de 11% para animais acima de 100kg, enquanto que para os bovinos a diferença média entr... (Resumo completo, clicar acesso eletrônico abaixo) / Abstract: In order to use three-dimensional (3D) modeling to estimate area, volume and mass, as well as to evaluate the impact of the deviations involved in the empirical equations available in the literature, in livestock, six shorn Corriedale sheep were weighed with mass between 10 and 105 kg and five Nelore cattle with mass between 600 and 800 kg. The animals were trained using the principles of habituation and conditioning in order to reduce the reactivity of the animals. For the measurement of the surface area, three procedures were used: empirical equation (EE) for each species, manual measurement instrument (IMM) and 3D animal modeling by photogrammetry (AR) and kinect sensor (AS_KS) scanning. A covariance analysis was performed, with the mass being covariable and a non-linear regression analysis was adjusted as a power function (allometry). Analyzing the surface area averages for sheep, in relation to the procedures used there was no difference between IMM and 3D models, as well as between 3D models, while between IMM and EE there was a significant difference. For cattle, it was observed that the 3D procedures were, on average, higher than the surface area estimates by EE (a=5%). When data were plotted on the surface area of sheep and the methods tested for body mass, empirical approximations were obtained, different from those estimated by EE in more than 11% for animals over 100 kg, while for cattle the difference between 3D and EE models was over 18%. As the adjustments of t... (Complete abstract click electronic access below) / Doutor

Computer vision tasks such as three dimensional (3D) registration, 3D modelling, and 3D object recognition are becoming more and more useful in industry, and have application such as reverse CAD engineering, and robot navigation. Each of these applications use correspondence algorithms as part of their processes. Correspondence algorithms are required to compute accurate mappings between artificial surfaces that represent actual objects or scenes. In industry, inaccurate correspondence is related to factors such as expenses in time and labour, and also safety. Therefore, it is essential to select an appropriate correspondence algorithm for a given surface matching task. However, current research in the area of surface correspondence is hampered by an abundance of applications specific algorithms, and no uniform terminology of consistent model for selecting and/or comparing algorithms. This dissertation presents a correspondence framework for surface matching algorithms. The framework is a conceptual model that is implementable. It is designed to assist in the analysis, comparison, development, and implementation of correspondence algorithms, which are essential tasks when selecting or creating an algorithm for a particular application. The primary contribution of the thesis is the correspondence framework presented as a conceptual model for surface matching algorithms. The model provides a systematic method for analysing, comparing, and developing algorithms. The dissertation demonstrates that by dividing correspondence computation into five stages: region definition, feature extraction, feature representation, local matching, and global matching, the task becomes smaller and more manageable. It also shows that the same stages of different algorithms are directly comparable. Furthermore, novel algorithms can be created by simply connecting compatible stages of different algorithms. Finally, new ideas can be synthesised by creating only the stages to be tested, without developing a while new correspondence algorithm. The secondary contribution that is outlined is the correspondence framework presented as a software design tool for surface matching algorithms. The framework is shown to reduce the complexity of implementing existing algorithms within the framework. This is done by encoding algorithms in a stage-wise procedure, whereby an algorithm is separated into the five stages of the framework. The software design tool is shown to validate the integrity of restructuring existing algorithms within it, and also provide an efficient basis for creating new algorithms. The third contribution that is made is the specification of a quality metric for algorithms comparison. The metric is used to assess the accuracy of the outcomes of a number of correspondence algorithms, which are used to match a wide variety of input surface pairs. The metric is used to demonstrate that each algorithm is application specific, and highlight the types of surfaces that can be matched by each algorithm. Thus, it is shown that algorithms that are implemented within the framework can be selected for particular surface correspondence tasks. The final contribution made is this dissertation is the expansion of the correspondence framework beyond the surface matching domain. The correspondence framework is maintained in its original form, and is used for image matching algorithms. Existing algorithms from three image matching applications are implemented and modified using the framework. It is shown how the framework provides a consistent means and uniform terminology for developing both surface and image matching algorithms. In summary, this thesis presents a correspondence framework for surface matching algorithms. The framework is general, encompassing a comprehensive set of algorithms, and flexible, expanding beyond surface matching to major image matching applications.

Thesis (Ph.D.). / Written for the Dept. of Electrical & Computer Engineering. Title from title page of PDF (viewed 2008/01/12). Includes bibliographical references.

Thesis (Ph.D.) - University of Glasgow, 2008. / Ph.D. thesis submitted to the Faculty of Information and Mathematical Sciences, Department of Statistics, University of Glasgow, 2008. Includes bibliographical references. Print version also available.

Thesis (Ph. D.)--University of Chicago, Dept. of Radiology, August 2000. / Includes bibliographical references. Also available on the Internet.