Ultrasound Super Resolution using Vision Transformer with Convolution Projection Operation

Ultrasound Localization Microscopy (ULM) refers to a technique using ultra-fast consecutive frames to create a super resolution image beyond the diffraction limit. As one of general steps for ULM, ultrasound microbubble (MB) localization directly affects the image performance. The traditional localization methods suffer from the lake of robustness and computational inefficiency. To solve these problems, we propose a transformer based convolutional neural network to make an end-to-end mapping to localize the microbubbles. The performance of the proposed method is validated on data from Ultrasound Localisation and TRacking Algorithms for Super Resolution (ULTRA-SR). The results showed that our proposed network achieved high precision and Jaccard index. These benefits can be used to further improve the image visualization and processing efficiency.