Review of Robot-Assisted HIFU Therapy.

This paper provides an overview of current robot-assisted high-intensity focused ultrasound (HIFU) systems for image-guided therapies. HIFU is a minimally invasive technique that relies on the thermo-mechanical effects of focused ultrasound waves to perform clinical treatments, such as tumor ablation, mild hyperthermia adjuvant to radiation or chemotherapy, vein occlusion, and many others. HIFU is typically performed under ultrasound (USgHIFU) or magnetic resonance imaging guidance (MRgHIFU), which provide intra-operative monitoring of treatment outcomes. Robot-assisted HIFU probe manipulation provides precise HIFU focal control to avoid damage to surrounding sensitive anatomy, such as blood vessels, nerve bundles, or adjacent organs. These clinical and technical benefits have promoted the rapid adoption of robot-assisted HIFU in the past several decades. This paper aims to present the recent developments of robot-assisted HIFU by summarizing the key features and clinical applications of each system. The paper concludes with a comparison and discussion of future perspectives on robot-assisted HIFU.

Modular force approximating soft robotic pneumatic actuator.

PURPOSE: Soft robots are highly flexible and adaptable instruments that have proven extremely useful, especially in the surgical environment where compliance allows for improved maneuverability throughout the body. Endoscopic devices are a primary example of an instrument that physicians use to navigate to difficult-to-reach areas inside the body. This paper presents a modular soft robotic pneumatic actuator as a proof of concept for a compliant endoscopic device. METHODS: The actuator is 3D printed using an FDM printer. Maximum bending angle is measured using image processing in MATLAB at a gauge pressure level of 35 psi. End-effector displacement is measured using electromagnetic tracking as gauge pressure ranges from 10 to 35 psi, and uniaxial tensile loading ranges from 0 to 120 g. RESULTS: The actuator achieves a maximum bending angle of 145°. Fourth-order polynomial regression is used to model the actuator displacement upon inflation and tensile loading with an average coefficient of correlation value of 0.998. We also develop a feedforward neural network as a robust computer-assisted method for controlling the actuator that achieves a coefficient of correlation value of 0.996. CONCLUSION: We propose a novel modular soft robotic pneumatic actuator that is developed via rapid prototyping and evaluated using image processing and machine learning models. The curled resting shape allows for simple manufacturing and achieves a greater range of bending than other actuators of its kind. A feedforward neural network provides accurate prediction of end-effector displacement upon inflation and loading to deliver precise manipulation and control.