A new robust markerless method for automatic image-to-patient registration in image-guided neurosurgery system

Area of Science:

• Medical Imaging
• Neurosurgery
• Computer-Aided Surgery

Background:

• Traditional point-based registration in image-guided neurosurgery requires fiducial markers and dedicated imaging, increasing complexity.
• Existing surface-based registration methods often necessitate a manual coarse registration step, leading to longer procedure times and potential inaccuracies.
• There is a need for automated, efficient, and accurate registration techniques in image-guided neurosurgery.

Purpose of the Study:

• To develop and validate a novel automatic surface-based registration method for image-guided neurosurgery.
• To eliminate the need for manual intervention in the coarse registration phase.
• To improve the accuracy and efficiency of image-to-patient registration.

Main Methods:

• Proposed an automatic surface-based registration technique utilizing 3D surface feature description and matching algorithms.
• Employed a 3D surface feature description and matching algorithm for initial coarse point correspondence.
• Utilized the iterative closest point (ICP) algorithm for final fine-tuning and image-to-patient registration.

Main Results:

• The automatic registration method demonstrated stable accuracy across various downsampling resolutions (18-26 mm) and support radii (2-6 mm) in phantom experiments.
• Clinical data validation showed robust performance, with mean target registration errors (TREs) of 1.30 mm and 1.85 mm for two patients using full head surface registration.
• The method achieved sufficient registration accuracy on diverse real-world surface regions in both phantom and clinical settings.

Conclusions:

• Introduced a new, robust automatic surface-based registration method leveraging 3D feature matching.
• The proposed method successfully automates the registration process, reducing time and uncertainty.
• Demonstrated high accuracy and practicality for image-guided neurosurgery applications.