An improved registration method based on ICP for image guided prostate seed implanting surgery

Abstract

Minimally invasive image guided surgery (IGS) has been increasingly used in interventional radiology. One of the key technologies in IGS is tracking the surgical instruments during operation, which makes it possible to visualize the position and orientation relationships between the tumor and the puncture needle based on the 3D model reconstructed from pre-operative images. In order to improve the accuracy of instruments tracking, and simplify registration process, we proposed a registration method based on virtual frame. The registration was performed in two steps. In the first stage, the rigid frame registration using the iterative closest point algorithm was completed by the producer when the surgery system was installed and debugged in the operating room, during which a standard transformation matrix from real space to model space was obtained. In the second stage, getting more than three points from the patient’s skin and converting them to the image space using the matrix acquired in the first step, and then working out the transformation matrix between the converted points and the corresponding points in the image space, the rotation and translation parameters were calculated by the second matrix and were used to adjust the model’s location and pose. This intra-operative registration was completed by surgeons. In the surgery system, we employed the electromagnetic measurement system to locate the puncture needle, and it was calibrated by laser tracker to improve the position accuracy. Contrasting experiments were conducted to verify the effectiveness of the registration method. The results showed that the accuracy of conventional paired-point-based registration method was 0.85 ± 0.10 mm, while the proposed registration method was 0.44 ± 0.07 mm, indicating that the proposed registration method could improve the registration quality.