Soft-tissue sound-speed-aware ultrasound-CT registration method for computer-assisted orthopedic surgery

Area of Science:

• Orthopedic Surgery
• Medical Imaging
• Biomedical Engineering

Background:

• Ultrasound (US) is used in computer-assisted orthopedic surgery for bone registration due to its safety and cost-effectiveness.
• However, US image distortion from soft-tissue acoustic property variations limits registration accuracy.
• Accurate bone registration is crucial for effective surgical navigation.

Purpose of the Study:

• To develop a soft-tissue sound-speed-aware registration method to overcome US image distortion challenges.
• To enhance bone segmentation performance for improved registration accuracy.
• To provide surgeons with precise 3D navigation information.

Main Methods:

• A multi-channel overlay feature enhancement strategy for U²-net was proposed to improve bone segmentation.
• Soft-tissue sound speed was estimated using a bone surface distance map to update US-derived points.
• An iterative registration strategy was employed to optimize the final registration outcome.

Main Results:

• Phantom experiments demonstrated high accuracy for femur and tibia/fibula registration.
• Fiducial registration error was 0.98 ± 0.08 mm for the femur and 1.29 ± 0.19 mm for the tibia/fibula.
• Target registration error was consistently below 2.11 mm, indicating superior performance.

Conclusions:

• The proposed soft-tissue sound-speed-aware method significantly improves bone registration accuracy in orthopedic surgery.
• This technique effectively addresses US image distortion caused by soft-tissue variations.
• The method shows potential for integration into surgical navigation systems, offering reliable 3D guidance.