When a suspicious lesion is detected in breast tissue, often an MRI-guided biopsy is performed in order to classify said lesion. The current biopsy procedure is however not efficient and not accurate, resulting in relatively large tissue damage. Therefore, over the last few years, an MR-safe robotic system for breast biopsies is developed called the Sunram 7.
For accurate position control, robots use encoders. The Sunram 7 however does not possess such encoders as they are not MR-safe. Therefore, only feedforward can be used. In this way, all the joints of the robot have to be placed in a certain zero-configuration from which the robot is then steered towards a target. This method is however susceptible to errors such as the skipping of steps during operation or non-accurate calibration. Therefore, in this Thesis, a new system, which can optically track the needle-tip of the Sunram 7, is designed and evaluated.
The designed tracking system uses computer vision in conjunction with ArUco markers to track the position of the needle in a coordinate frame defined by the base on which the robot is mounted.
It was shown that this method can place the needle-tip in a range of 0.1 to 6mm from the desired target. The most important source of this error was found to be the contour detection of the marker, which needs to be improved in order to make the system even more accurate.