서강대학교

초록/요약

Clinical usage of high-intensity focused ultrasound(HIFU) has been attracting notable interests for it’s potential as a completely non-invasive alternative to open surgery. Since HIFU therapy is capable of generating precise, homogeneous lesion with low risk of complications and shorter postoperative recovery time, it is widely used for oncological applications. However, in case of upper abdominal organs like liver, treatment target keeps drifting due to breathing motion. Thus, it is important to continuously monitor and track this movement. Ultrasound imaging is superior than other monitoring modalities in that it is available in real-time and has high penetration rate in human body. Multiple studies have shown that tracking anatomical land-marks such as vessels in ultrasound image is capable of accurately estimate breathing motion. This study suggests a new 2D ultrasound vessel tracking strategy utilizing region-based convolutional neural network and kernelized correlation filter tracking. Evaluations on human liver data shows that this method can estimate liver vessel movement with high accuracy and real-time capability.