This paper describes the design and evaluation of the second generation of a robotic system called the light puncture robot for thoracic and abdominopelvic interventional radiology procedures under CT and MRI guidance. It is mounted on the patient's body and positions and inserts the needle according to the trajectory and target chosen by the radiologist in the image. The mechanical design is described in detail along with its forward and inverse kinematics. The robot can be segmented and registered fully automatically in both imaging modalities. Phantom experiments in the CT scanner and preliminary feasibility experiments in the MRI are described, showing a targeting accuracy of 3.3 ± 1.7 mm in gelatin for depths ranging from 30 to 90 mm and needle orientations of -13° to +15° about normal from the patient's skin. A detailed error analysis is discussed along with potential improvements in the system. The potential clinical advantages of the system include higher targeting accuracy for complex dual obliquity trajectories, the need for fewer images, and improved accessibility to MRI-guided interventions.

This paper describes a new 3-D ultrasound robotic prostate brachytherapy system. It uses a stationary 3-D ultrasound probe rigidly fixed to a robotic needle insertion mechanism. The novelty of the system is its ability to track prostate motion intraoperatively to allow the dose planning and needle trajectories or depths to be adapted to take into account these motions. Prostate tracking is done using a fast 3-D ultrasound registration algorithm previously validated for biopsy guidance. The 7-degree-of-freedom (7-DOF) robot and ultrasound probe are calibrated together with an accuracy of 0.9 mm, allowing the needles to be precisely inserted to the seed targets chosen in the reference ultrasound image. Experiments were conducted on mobile, deformable synthetic prostate phantoms using a prototype laboratory system. Results showed that, with prostate motions of up to 7 mm, the system was able to reach the chosen targets with less than 2-mm accuracy in the needle insertion direction. This measured accuracy included extrinsic measurement errors of up to 1.1 mm. A preliminary cadaver feasibility study was also described in preparation for more realistic experimentation of the system.