Our previous research demonstrated that, under ideal conditions, high-precision image registration between real-time ultrasound (US) images and preoperative CT/MR images could be achieved using real-time US guidance for needle insertion. In this study, we enhanced a constructed phantom to better simulate the structure of human abdominal tissue, aiming to verify the accuracy of surgical needle placement guided by real-time US images registered with preoperative CT/MR images. Specifically, the pelvic phantom was improved using animal abdominal tissue to mimic the properties of human abdominal tissue. In the improved phantom experiments, the target registration error (TRE) for the real-time needle trajectory of four implanted needles in the registered US-CT images was 0.81 ± 0.11 mm (mean ± SD), with an average trajectory angle error of 0.83 ± 0.063 degrees (mean ± SD). For the registered US-MR images, the average registration error was 0.76 ± 0.014 mm (mean ± SD), and the average trajectory angle error was 0.825 ± 0.018 degrees (mean ± SD). These results demonstrate that, even in simulated human tissue, rigid registration between real-time US images and preoperative CT/MR images maintained high accuracy, supporting the feasibility of real-time US-guided needle insertion.
Keywords: Cervical carcinoma; Gynecologic brachytherapy; Image-guidance; Registered ultrasound image; Rigid registration.
© 2025. The Author(s).