Implementation of Super-Resolution Ultrasound Imaging for in Vitro Experiments

Sammanfattning: Ultrasound imaging systems are a safe and affordable imaging modality with the disadvantage of low spatial resolution, especially for assessing smaller vessels. However, with the implementation of super-resolution ultrasound imaging techniques, studies have shown promising results in achieving a resolution below the diffraction limit. Super-resolution ultrasound imaging techniques takes advantage of the point spread function to localize the centroid of the ultrasound contrast agents in an image. By superimposing thousands of these images, a super-resolved image of the localized and tracked contrast agents can be created, which presents an image where vessels down to a few micrometers can be resolved. The purpose of this master’s thesis was to implement super-resolution ultrasound imaging, test different localization methods and analyze them by using different ultrasound contrast agent concentrations. Grayscale ultrasound images were acquired using the Verasonics system for three different microbubble concentrations. The super-resolution ultrasound imaging program was executed on the grayscale images using three different localization methods, Gaussian fit, No-shift and Interpolation based scheme. The microbubbles were localized and tracked over several frames to create a super-resolved image which had the pixel resolution of a 10th of the wavelength. Significant improvements were demonstrated in the super-resolved images compared to the grayscale images. The higher microbubble concentrations resulted in a higher number of localized and tracked microbubbles. While the low concentration exhibited lower values. Comparing the methods, Gaussian fit and No-shift detected higher number of microbubbles than the method Interpolation. Although further analysis is needed, the thesis concluded that using Gaussian fit as a localization method and higher microbubble concentrations, a super-resolved image can be produced even if the program is tested on fewer images.