Robust Super-Resolution Ultrasound Microbubble Tracking with Optical Flow Guided Kalman Filter

This paper describes a contribution to the Ultrasound Localization and TRacking Algorithms for Super Resolution (ULTRA-SR) Challenge. As a potential strategy for clinical applications, super-resolution (SR) imaging overcomes the diffraction limit and achieves sub-wavelength spatial resolution imaging by locating microbubble (MB) contrast agents and stacking them in multiple frames. To obtain SR images with higher quality, the effectiveness of MB localization and the robustness of MB pairing and tacking still need to be improved. In this paper, we propose to integrate grayscale morphological reconstruction (MR) into the localization and combine the Kalman filter with optical flow estimation for the pairing and tracking process. The proposed method has been demonstrated on both simulation data and in vivo data. The results show that the proposed method can help remove noise and artifacts, locate more MBs, smooth MB tracks, and ultimately improves the quality of SR images.