Single microbubble tracking in medical ultrasound imaging

Currently contrast enhanced ultrasound imaging (CEUS) aims to assess blood perfusion macroscopically with millimetre resolution in the clinic. However, current research presents key limitations that are associated with ultrasound imaging in general, such as the variable point spread function (PSF) across the image and the non-constant velocities of the blood among the vascular tree.
Single microbubbles (MBs) are very efficient point scatterers of ultrasound and they lend themselves as ideal point targets of the imaging field. We utilize an existing particle tracking algorithm that has been implemented to track single organelles and molecules in optical microscopy. The first priority is to translate the above software into an ultrasound specific one, working on the two fundamental ideas of particle detection and linking.
In order to test the efficiency of every aspect of the algorithm, a ground truth of ultrasound data has to be defined. In contrary with other fields, ultrasound synthetic data is not available. As a result, in vivo data from the corpus luteum of the sheep ovary can be used. By visually inspecting the video loop we try to classify MBs from noise or other speckle, without any knowledge of the exact location of each injected MB.
We have managed to introduce a MB tracking software, modifying the aforementioned algorithm according to ultrasound limitations. The manual ground truth definition contributed to improve the software performance and our next step will be to convert that process into a more automatic one.