Robust Ultrasound-Based Visual Servoing for Beating Heart Intracardiac Surgery

Abstract

Abstract-This paper presents a robust visual servoing approach for automatic guidance of an instrument. The visual sensor is an ultrasound probe that observes an instrument which is inserted into the beating heart of a patient and manipulated by a robot. The present paper provides stability analysis, robustification of the control law and an in vivo experiment.

Hand-eye self-calibration of an ultrasound image-based robotic system

“…Ultimately, since the precision is measured by matching a model to a set of measurements, a simple inverse model can be used to reach any desired position located in the image within a millimetric precision. Note that the presented setup was also used for real time control of the instrument using visual servoing, during in vivo beating heart experiments, [4].…”

“…In other words, although the shape of the instrument image is modified, its center of gravity is unchanged. This is important because during in vivo experiments, high gains may be required to see anatomic features, [4].…”

“…A typical application described in [3], [4] consists of intracardiac surgery, where an instrument, manipulated by a robot, is inserted into the beating heart, where it can move with four degrees of freedom while its motion is observed through a trans-esophagus ultrasonic probe. By no way the probe can be physically linked to the instrument.…”

“…In this case, corrections are constantly made so that the instrument location in the image is finally correct, whatever the geometrical errors are, thanks to the capacity of rejecting static disturbances. However, apart from the specific visual servoing technique that we proposed in [4], all the other existing techniques (see e.g. [9], [10]) require the knowledge of the location of the probe with respect to the instrument.…”

Improving ultrasound intensity-based visual servoing: Tracking and positioning tasks with 2D and bi-plane probes

“…In another work [6], the three translations of a XYZ stage robot equipped with two US probes and a HIFU transducer are controlled to follow a kidney stone while compensating for physiological motions during a lithotripsy procedure. With a 2D US probe, Vitrani et al [7] proposed to use the two image points corresponding to the intersection of a surgical forceps with the probe plane as visual features to control 4 dof of the tool inside a beating heart in an eyeto-hand configuration. In relation with this work, the authors of [8] developed a predictive control scheme to keep the forceps visible in the US image.…”