Preliminary study of needle tracking in a microsurgical robotic system for automated operations

Surgical needle tracking is an important element of high-level automated operations conducted by surgical robotic systems. However, conventional needle tracking algorithms lack robust performance with different needle postures and are not applicable to the small needles used during microsurgery. This paper discusses a robust, efficient needle tracking algorithm, which is capable of estimating all of the positions and of the postures of a microsurgical needle. In the preoperative preparation stage, contour models of the microsurgical needle are generated using a 3-D CAD model and saved in a database. During the operation, the system extracts the contours of the microsurgical needle from the microscopic image using the edge and the color information. The system then calculates the likelihood of the contour models in the database by matching the contours extracted from the microscopic image. The experimental results indicated that our proposed method has high accuracy when tracking a microsurgical needle, and that it performed robustly with different needle postures.