A level-wise spine registration framework to account for large pose changes.

Area of Science:

• Medical Imaging
• Surgical Navigation
• Biomedical Engineering

Background:

• Spine registration is vital for image-guided surgery.
• Intervertebral motion due to pose changes causes significant registration errors.
• Existing methods struggle with large pose variations.

Purpose of the Study:

• To develop a geometrical rectification technique for robust level-wise vertebral registration.
• To overcome registration errors caused by significant changes in spine pose.
• To improve accuracy and efficiency in spine image guidance.

Main Methods:

• Utilized nonlinear principal component analysis (NLPCA) for geometrical rectification.
• Employed a parameterized moving-frame approach to normalize spinal postures.
• Generated 2D depth projection images for point-to-pixel correspondences.
• Performed level-wise rigid registration using 3D surface point pairs.

Main Results:

• Achieved average target registration errors (TRE) of 1.70 ± 0.15 mm (ex vivo) and 1.85 ± 0.16 mm (in vivo).
• Demonstrated robustness to significant pose changes (up to 87.5 mm and 32.7 degrees).
• Automated registration process completed within 3-5 minutes.

Conclusions:

• The proposed level-wise spine registration is accurate and efficient for human open spine surgeries.
• The framework shows potential for application with other intraoperative imaging modalities like ultrasound and MRI.
• This approach can enhance the utility of spine registration in various clinical settings.