Virtual-force-guided intraoperative ultrasound scanning with online lesion location prediction: A phantom study.

Abstract

In ultrasound-guided minimally invasive surgery (MIS) of tumours, it is crucial to discover the optimal scanning plane (OSP) and organise the MIS scalpel work trajectory in this plane. The OSP can be altered and is challenging to track when the scalpel interacts with deformed tissues. Therefore, tracking the OSP becomes critical in MIS. In master-slave control, virtual force (VF) is used to assist the operator in completing the task. However, most literature assumes that the environment is sufficiently stable. No specific method focuses on tracking the OSP of the lesion within largely deformed tissues. This paper used the improved artificial potential field method to establish the VF that could guide the operator to track the OSP. When tissue deformation occurred, an artificial neural network (ANN) was used to predict the target position, guiding the operator to find the new OSP. An experimental robot platform was built to verify the proposed algorithm's effects. Experiments to track the OSP were performed on a phantom. The results showed that the presented method could reduce the trajectory redundancy of ultrasonic scanning, shorten the time of OSP discovery and tracking, and decrease the deviation between the ultrasonic scanning position and the OSP. This method has significance for the accurate localization and successful removal of tumours. Future work will focus on improving the adaptability of the proposed ANN prediction model in different phantoms.