Image-guided robots for dot-matrix tumor ablation

Abstract

Advances in medical imaging now provides detailed images of solid tumors inside the body and miniaturized energy delivery systems enable tumor destruction through local heating powered by a thin electrode. However, the use of thermal ablation as a first line of treatment is limited due to the difficulty in accurately matching a desired treatment and a limited region of active heating around an electrode. The purpose of this research is to identify and quantify the current limitations of image-guided interventional procedures and subsequently develop a procedure and devices to enable accurate and efficient execution of image-based interventional plans and thus ablation of a tumor of any shape with minimal damage to surrounding tissue. Current limitations of probe placement for ablation therapy were determined by a detailed retrospective study of 50 representative CT-guided procedures. On average, 21 CT scans were performed for a given procedure (range 11-38), with the majority devoted to needle orientation and insertion (mean number of scans was 54%) and trajectory planning (mean number of scans was 19%). A regression analysis yielded that smaller and deeper lesions were associated with a higher number of CT scans for needle orientation and insertion; highlighting the difficulty in targeting. Another challenge identified was repositioning the instrument distal tip within tissue. The first robot is a patient-mounted device that aligns an instrument along a desired trajectory via two motor-actuated concentric, crossed, and partially nested hoops. A carriage rides in the hoops and grips and inserts an instrument via a two degree-of-freedom friction drive. An imagebased point-and-click user interface relates appropriate clicks on the medical images to robot commands. Mounting directly on the patient provides a sufficiently stable and safe platform for actuation and eliminates the need to compensate for chest motion; thereby reducing the cost and complexity compared to other devices. Phantom experiments in a realistic clinical setting demonstrated a mean targeting accuracy of 3.5 mm with an average of five CT scans. The second robot is for repositioning the distal tip of a medical instrument to adjacent points within tissue. The steering mechanism is based on the concept of substantially straightening a pre-curved Nitinol stylet by retracting it into a concentric outer cannula, and re-deploying it at different axial and rotational cannula positions. The proximal end of the cannula is attached to the distal end of a screw-spline that enables it to be translated and rotated with respect to the casing. Translation of the stylet relative to the cannula is achieved with a second concentric, nested smaller diameter screw that is constrained to rotate with the cannula. The robot mechanism is compatible with the CT images, light enough to be supported on a patient's chest or attached to standard stereotactic frames. Targeting experiments in a gelatin phantom demonstrated a mean targeting error of 1.8 mm between the stylet tip and that predicted with a kinematic model. Ultimately, these types of systems are envisioned being used together as part of a highly dexterous patient-mounted positioning platform that can accurately perform ablation of large and irregularly shaped tumors inside medical imaging machines - offering the potential to replace expensive and traumatic surgeries with minimally invasive out-patient procedures.