<p>Healthy infants in the age group of 9 to 20 weeks post term have a distinct movement pattern called fidgety movements. The abscense of, or anomalies in these movements are indications that the infant may suffer from Cerebral Palsy. Data from video clips of infants have been extracted and used as features for classifying these movements as normal or abnormal. An earlier project used the concept of motiongrams for extracting the centroid of motion and quantity of motion of the image from each frame of the video clips. This project attempts to improve the dataset by extracting the same features using the concept of optical flow. Using the classification methods that yielded the best results from the motiongram project, the new dataset based on optical flow was constructed and tested for comparison. The image was as before divided into four quadrants with a circle of varying radius in the middle. Tracking how the centroid of motion moved between the separated areas of the image generated a transition probability matrix. By classifying on the entropy and variance of this matrix, a sensitivity of 90,0% and a specificity of 94,6% was reached, which was an improvement compared to the best classification result based on motiongrams with a sensitivity of 90,0% and specificity of 86,6%. The overall performance of both methods are around a sensitivity of 90% and a specificity of 70%. Using motiongrams and optical flow to extract features from video data for classification has turned out to be a promising approach for a simple and non-invasive objective tool for diagnosing infants with Celebral Palsy. The movement information in an optical flow field does however have a much higher potential for generating discriminative features than just as the simple representation a centroid of motion actually is.</p>

<p>The presence of clustered micro-calcifications is an important indicator of early stage breast cancers, and an efficient real-time ultrasound imaging support is needed for conducting needle biopsies. SURF is a dual-frequency band ultrasound technique capable of imaging nonlinear scattering using the superposition of a low-frequency pulse to manipulate the tissue characteristics and an imaging pulse to extract the informations. This new technique has already demonstrated its superiority in imaging high intensity nonlinear scatterers like ultrasound contrast agents compared to the classical methods. However, imaging lower intensity nonlinear scatterers like micro-calcifications requires the use of higher manipulation pressures, causing the apparition of complex distortions of the imaging pulse neglected hitherto. In order to achieve a sufficient level of tissue suppression and detect the micro-calcifications using SURF imaging, these nonlinear forward propagation effects must be characterized and corrected for. In this thesis, simulations have been conducted based on the geometry of the first manufactured prototypes of SURF probes, named Viglen and Okla. The nonlinear propagation effects were described as the regroupment of a nonlinear propagation delay due the average level of LF manipulation pressure experienced by the HF pulse, a mix of compressions/expansions of the imaging pulse resulting from the fluctuations of the LF manipulation pressure over the HF pulse, and a phenomenon of SURF aberration due to the difference in the focusing and diffraction pattern of the HF and LF beams. The build-up of these nonlinear propagation effects were shown to be fully characterized by the combination of the profiles of the LF beam and the phase-relation between the two pulses. The different beamforming strategies, with the superposition of a focused HF beam on a plane or co-focused LF beam, were also discussed: a net trade-off appears between the minimization of delays and SURF aberrations ensured by the co-focal configurations, and the minimization of compressions/expansions effects by using plane LF beams. A strategy to limit the nonlinear propagation effects would be the use of an especially designed transducer providing an excellent HF focusing in the study range and a sufficiently large LF aperture. A filter was then designed and tested to ensure a correction of the pulse-form distortions in the received signals. The filter improved the level of suppression of the linear scattering from tissue by 15 dB in the numerical simulations, allowing a theoretical detection of the micro-calcifications and raising the resolution of the imaging system up to 1 dB close to the noise floor. The same filter demonstrated a smaller suppression of only 7 dB with the experiments conducted on phantoms, 2 dB below the limit required for detecting the smallest micro-calcifications. Further work studying the pulse-form correction filter should be conducted to understand this difference in performances, along with a real-time implementation on the scanner, before achieving a correct detection of micro-calcification using SURF imaging.</p>

<p>High quality images with strong contrast, good resolution and geometrical consistency are of crucial importance in magnetic resonance imaging, where the relatively low intrinsic sensitivity of MR methods places high demands on the imaging hardware. One of many key components in the imaging system is the radiofrequency coil, responsible for transmitting excitation signals and/or listening for response from the object. In this project a number of such coils are developed for specific applications (namely proton imaging of rat and fish brains), then evaluated against phantoms and empirically in simulated imaging situations. Evaluation of the produced coils shows promising initial results, with various opportunities for further refinement into a device suitable for regular use.</p>

Healthy infants in the age group of 9 to 20 weeks post term have a distinct movement pattern called fidgety movements. The abscense of, or anomalies in these movements are indications that the infant may suffer from Cerebral Palsy. Data from video clips of infants have been extracted and used as features for classifying these movements as normal or abnormal. An earlier project used the concept of motiongrams for extracting the centroid of motion and quantity of motion of the image from each frame of the video clips. This project attempts to improve the dataset by extracting the same features using the concept of optical flow. Using the classification methods that yielded the best results from the motiongram project, the new dataset based on optical flow was constructed and tested for comparison. The image was as before divided into four quadrants with a circle of varying radius in the middle. Tracking how the centroid of motion moved between the separated areas of the image generated a transition probability matrix. By classifying on the entropy and variance of this matrix, a sensitivity of 90,0% and a specificity of 94,6% was reached, which was an improvement compared to the best classification result based on motiongrams with a sensitivity of 90,0% and specificity of 86,6%. The overall performance of both methods are around a sensitivity of 90% and a specificity of 70%. Using motiongrams and optical flow to extract features from video data for classification has turned out to be a promising approach for a simple and non-invasive objective tool for diagnosing infants with Celebral Palsy. The movement information in an optical flow field does however have a much higher potential for generating discriminative features than just as the simple representation a centroid of motion actually is.

The presence of clustered micro-calcifications is an important indicator of early stage breast cancers, and an efficient real-time ultrasound imaging support is needed for conducting needle biopsies. SURF is a dual-frequency band ultrasound technique capable of imaging nonlinear scattering using the superposition of a low-frequency pulse to manipulate the tissue characteristics and an imaging pulse to extract the informations. This new technique has already demonstrated its superiority in imaging high intensity nonlinear scatterers like ultrasound contrast agents compared to the classical methods. However, imaging lower intensity nonlinear scatterers like micro-calcifications requires the use of higher manipulation pressures, causing the apparition of complex distortions of the imaging pulse neglected hitherto. In order to achieve a sufficient level of tissue suppression and detect the micro-calcifications using SURF imaging, these nonlinear forward propagation effects must be characterized and corrected for. In this thesis, simulations have been conducted based on the geometry of the first manufactured prototypes of SURF probes, named Viglen and Okla. The nonlinear propagation effects were described as the regroupment of a nonlinear propagation delay due the average level of LF manipulation pressure experienced by the HF pulse, a mix of compressions/expansions of the imaging pulse resulting from the fluctuations of the LF manipulation pressure over the HF pulse, and a phenomenon of SURF aberration due to the difference in the focusing and diffraction pattern of the HF and LF beams. The build-up of these nonlinear propagation effects were shown to be fully characterized by the combination of the profiles of the LF beam and the phase-relation between the two pulses. The different beamforming strategies, with the superposition of a focused HF beam on a plane or co-focused LF beam, were also discussed: a net trade-off appears between the minimization of delays and SURF aberrations ensured by the co-focal configurations, and the minimization of compressions/expansions effects by using plane LF beams. A strategy to limit the nonlinear propagation effects would be the use of an especially designed transducer providing an excellent HF focusing in the study range and a sufficiently large LF aperture. A filter was then designed and tested to ensure a correction of the pulse-form distortions in the received signals. The filter improved the level of suppression of the linear scattering from tissue by 15 dB in the numerical simulations, allowing a theoretical detection of the micro-calcifications and raising the resolution of the imaging system up to 1 dB close to the noise floor. The same filter demonstrated a smaller suppression of only 7 dB with the experiments conducted on phantoms, 2 dB below the limit required for detecting the smallest micro-calcifications. Further work studying the pulse-form correction filter should be conducted to understand this difference in performances, along with a real-time implementation on the scanner, before achieving a correct detection of micro-calcification using SURF imaging.

High quality images with strong contrast, good resolution and geometrical consistency are of crucial importance in magnetic resonance imaging, where the relatively low intrinsic sensitivity of MR methods places high demands on the imaging hardware. One of many key components in the imaging system is the radiofrequency coil, responsible for transmitting excitation signals and/or listening for response from the object. In this project a number of such coils are developed for specific applications (namely proton imaging of rat and fish brains), then evaluated against phantoms and empirically in simulated imaging situations. Evaluation of the produced coils shows promising initial results, with various opportunities for further refinement into a device suitable for regular use.

Et av de mer opplagte problemene som møter en person hvis arm har blitt erstattet med en protese, er reduksjonen i antall mulige bevegelser som en protese har i forhold til en frisk arm. Studier foretatt på friske armer, viser et større repertoar av bevegelser når håndleddets rotasjonsakse er stilt på skrå i forhold til underarmens lengdeakse. Et pågående forskningsprosjekt har som mål å undersøke om disse resultatene kan brukes til å hjelpe protesebrukere, og i den anledning har det blitt laget et håndledd med mulighet for vinkling av rotasjonsaksen.Gjennom bruk av en utviklingsmetodikk med navn V-modellen har denne oppgaven, som fortsettelse på tidligere arbeid, forsøkt å lage et fungerende styresystem for håndleddet. Underveis i arbeidet har utviklingsmetodikkens faser blitt fulgt, og all nødvendig dokumentasjon og planer for testing har blitt testet.Maskinvaren til håndleddets styresystem har blitt modularisert og designet for å håndtere sine tilordnede oppgaver. Kretsskjema for hver modul ble tegnet og lagt ut på flere kretskort som deretter kunne stables for å realiserte en prototyp av hele systemet. Programvare for hver av maskinvarens moduler har også blitt skrevet.Av testingen som har blitt gjennomført, vises det at den realiserte prototypen tilfredsstiller kravene fra funksjons- og teknisk spesifikasjon.

Invasive measurements of cardiac functioning allows for more accurate measures of cardiac functioning than non-invasive measurements. However, invasive measurements is often not available in clinical settings. By comparing invasive and non-invasive measurements collected in an experimental context, better relationships between non-invasive measurements and cardiac functioning may be found. This master thesis describes the development of two computer applications for simultaneous acquisition, calibration and synchronization of these measurements. The developed applications were tested out during operations on pigs with all measurement sources connected. The results shows that all the desired measurements were successfully acquired by the system. Calibration of the different measurement were also achieved. Different methods for synchronization were tested out during the experiments. It was possible to achieve synchronization of all clocks present in the system. Finally, all of the desired parameters were calculated.

Intraoperative ultrasound strain imaging of the brain visualizes brain tissue deformation as an image. The hypothesis is that strain and elastographic values can be used to complement conventional B-mode image in the task of brain tumor segmentation. A key asumption is that the natural pulsation of the cerebral arteries causes deformation in the brain tissue that is measurable with ultrasound.Strain values are found with a least-squares technique that estimates the spatial derivative of axial velocity, which in turn is mea- sured using a phase-based velocity estimator. A correlation coefficient is calculated for each estimate, giving an indicator of estimation accuracy. Additionally a method for hiding estimates of bad quality based on correlation coefficient thresholding is demonstrated. More- over, a novel elastographic processing technique suitable for cineloop display is introduced. This method extracts a stiffness parameter from a series of strain images, producing an elastogram. A graphical user interface allowing the user to change parameters and see the corresponding result in real-time, minimizing the time needed for parameter optimization, has been developed.The method has been tested using an elasticity phantom. The phantom elastogram cineloop shows a live image that visualizes the difference between stiff and soft tissue well, portraying information not found in the B-mode image. The conclusion is that the proposed elastographic technique, combined with correlation coefficient thresholding, produces elastograms that are suitable for real-time display. This technique is not limited to imaging of the brain, and could, with different parameters, be used for imaging other parts of the body as well.Clinical data sets from two brain tumor patients have been studied as well, where the estimated velocity, strain and elastographic values is discussed in detail. In both patients the tissue movement due to arterial pulsation was measurable with ultrasound. For one patient, a correlation was found between tissue pathology and estimated strain and elastographic values. For the second patient the strain and elastographic processing broke down, and no similar correlation was found.

In this thesis a working control system for a 7 degrees of freedom hand prosthesis model controlled by electromyographic and accelerometer signals has been developed.The complete system consists of a wireless EMG and accelerometer measurement system, two National Instruments data acquisition modules, a desktop computer, a Lego Mindstorms NXT brick and a hand model with 7 motorized degrees of freedom.The controller is based on pattern recognition and signal classification.Several different EMG features for this purpose are presented and implemented.Two different linear classifiers were used and their performance studied.The LabVIEW software platform was used for both the computer and the NXT.The developed software has a modular design, facilitating future development and extension.Its design and implementation are presented and discussed.