Improved Curvelet Transform-based Sparsity Promoting Algorithm for Fast Ultrasound Localization Microscopy

Ultrasound localization microscopy (ULM) utilizes the localization of individual microbubbles (MBs) to surpass the conventional resolution limits of ultrasound imaging. Despite its advantages, ULM faces significant challenges related to data acquisition time. This is the main issue of ULM because a large number of MBs need to be detected and accurately localized to reconstruct detailed microvasculature. Consequently, the time required to gather sufficient data for a visualize an ULM image can be prohibitively long. To address this, the curvelet transform-based sparsity-promoting (CTSP) method has been proposed. This approach accelerates ULM imaging by leveraging sparsity in the MB localization signals to reconstruct images more quickly. By recovering missing MB signals, CTSP reduces the need for extensive data acquisition and thus shortens the overall acquisition time. However, this method has its limitations. The challenges include the reliance on post-processing approximations and issues with MB signal sparsity, which may degrade image quality. To overcome these limitations, we introduce an enhanced reconstruction algorithm that iteratively solves the norm optimization problem. This approach avoids post-processing approximations. Additionally, we incorporate a MB separation technique designed to improve signal sparsity. By enabling the detection and separation of MBs with greater precision.