Fiducial optimization for minimal target registration error in image-guided neurosurgery

Area of Science:

• Neurosurgery
• Medical Imaging
• Surgical Navigation

Background:

• Image-guided neurosurgery relies on accurate registration between preoperative images and the patient's anatomy.
• Fiducial markers and anatomical landmarks are crucial for this registration process, but their optimal placement is often suboptimal.
• Target Registration Error (TRE) quantifies the accuracy of the registration, and minimizing it is essential for safe and precise procedures.

Purpose of the Study:

• To develop and validate novel methods for the optimal selection of anatomical landmarks and placement of fiducial markers in image-guided neurosurgery.
• To minimize the expected Target Registration Error (TRE) by optimizing fiducial marker and landmark configurations.
• To reduce localization error without requiring additional imaging or hardware.

Main Methods:

• Developed an optimization framework for selecting anatomical landmarks and planning fiducial marker locations.
• Utilized a novel empirical simulation-based TRE estimation method incorporating actual fiducial localization error (FLE) data.
• Intraoperatively selected optimal sets of fiducial markers and anatomical landmarks to minimize expected TRE.

Main Results:

• Clinical experiments on five neurosurgery patients demonstrated reduced TRE with the proposed optimization method.
• Average TRE decreased from 4.7 mm (conventional setup) to 3.2 mm (optimized method).
• Observed a maximum TRE improvement of 4 mm, indicating significant enhancement in registration accuracy.

Conclusions:

• The proposed methods effectively optimize fiducial marker placement and anatomical landmark selection for image-guided neurosurgery.
• Optimizing these elements significantly reduces Target Registration Error, leading to more accurate surgical navigation.
• This approach has the potential to enhance the safety and precision of minimally invasive neurosurgical procedures.