Breast cancer is one of the most common forms of cancer today that is negatively affecting the lives of many people. One very important step in the diagnostics step is performing a biopsy procedure. During a biopsy, the radiologist inserts a needle inside the breast tissue and under ultrasound guidance he moves it towards the target lesion, to extract a sample. However, the biopsy can be time-consuming and uncomfortable for patients. Additionally, the accuracy of the result is heavily dependent on the experience of the practitioner who is performing the process. These same practitioners also tend to suffer from fatigue and work-related musculoskeletal pain. Many of these problems can be alleviated by using a robotic system that would assist the radiologists. Robotics is more capable at performing tasks that require repeatability and precision, compared to humans. This could result in improving the patient's experience, reduce the time of the entire diagnosis process, have a better accuracy rate, and benefit radiologists’ working conditions.
In this thesis, a control strategy is designed for such an assisting robotic system. A specialized end-effector, which contains an ultrasound probe and a small serial manipulator that determines the orientation of the biopsy needle, is used. The control strategy first moves the end-effector so that the ultrasound prove has a visual of the lesion target and so that the needle orientation mechanic can guide the biopsy needle into the correct direction. For this to happen, what is referred to as the "Initialization Phase" was first created. Furthermore, to ensure that the target would be hit, the tissue located at the target position was tracked through the ultrasound images. This was achieved by designing a tracking algorithm, with optical flow as its core. Additionally, a controller that would guide the needle, using this tracking algorithm, was designed and implemented. This controller relied on controlling the actuators of the needle orientation mechanism. However, the controller cannot compensate inaccuracies presented by needle bending. For that reason, the possibility of directly tracking the needle was examined. Lastly, it could be argued that an issue that the above controller has is that it does not give the radiologist any control since the needle orientation mechanism is manipulated with position control. For that reason, an impedance controller was designed that would allow the radiologist to controller the degree in which the robotic system and the user has control over the direction of the needle.
Multiple experiments took place to determine the accuracy of each component. Based on those results, the strengths and weaknesses of the current control strategy are discussed. Additionally, recommendations on how to further improve the current system are also presented.