An estimated 33,190 new cases of primary liver cancer and an even larger number of new cases of metastatic cancer to the liver will be diagnosed in the United States in 2014. Eradicating those tumors improves patient survival. Percutaneous radiofrequency ablation (RFA), which kills tumors via thermal damage, is a minimally invasive treatment appropriate for the many patients who are not eligible for surgical resection, particularly those with certain tumor type, location, or size; poor liver function; or oher comorbidities. However, technical limitations have precluded broader adoption of RFA as a means of eliminating liver tumors. Tumors in some portions of the liver cannot be accessed because the straight paths currently taken by RFA electrodes would traverse vasculature, lung, or other sensitive structures. Large tumors require multiple electrode insertions; the increased bleeding risk with each puncture through the liver capsule can be too great in patients with advanced liver disease or severe comorbidities. The long-term goal of this work is to significantly improve access, accuracy, and precision during percutaneous liver RFA procedures using robot-assisted needle steering. The following specific aims will enable the development of this technology and demonstrate feasibility, as a path to clinical adoption: Specific Aim 1: Optimize ultrasound-based localization of highly curved needles and closed-loop control of robotic needle steering. Specific Aim 2: Develop an interactive user control interface and enable teleoperated needle steering intervention. Specific Aim 3: Design a clinical robot-assisted needle steering system for percutaneous liver RFA, validated in an ex vivo liver model. Specific Aim 4: Validate the ultrasound-guided needle steering system in vivo in porcine liver. Needle steering enables the insertion of highly flexible needles or guidewires along controlled, curved, 3D paths through tissue. Thus, the needle can be actively steered to acquire tumors not accessible by a straight- line path, and unexpected deviations from the desired path can be corrected as the needle advances toward the tumor. A functional needle steering system appropriate for in vivo experimentation must be compatible with widely available medical imaging (here, ultrasound imaging), have the ability to deliver treatment (here, RFA through the use of a steerable needle electrode), and an appropriate user interface including visualization and human-in-the-loop teleoperation. This technology could extend the survival advantages of RFA to patients who may otherwise not have good treatment options, and in the long term, be used for accurate placement of diagnostic devices and treatment of diseases in other organs.