Improved microbubble (MB) Localisation Using Particle Detecting algorithm: Evaluation of Algorithm Performance for Different Beamforming Methods

The performance of image analysis techniques (particle detection) on contrast enhanced ultrasound (CEUS) images could be enhanced by using it in combination with the right beamformer (BF). The current study investigates the best performing combination of a particle detecting algorithm (Kanoulas et al. 2019) with four beamformers (BFs), classical and adaptive. In a series of in silico experiments, adjacent MBs are placed in distances comparable to the lateral resolution limit, the CEUS images of the MBs were simulated in FieldII, and finally beamformed with the four methods. The images were processed with the MB detection algorithm and the results were evaluated by the true detections (TD), missed MBs, spurious detections, and localisation uncertainty (LU). For the smallest distances all methods deteriorate but the MV methods provided 4-12% more TD. For the intermediate distances the TD were comparable for all BFs but the adaptive methods provided lower LU. When a set of evaluation metrics is used, the adaptive methods provide marginally but systematically improved results which suggests that, under the appropriate imaging conditions, they could be used to enhance vessel mapping.