Frameless neuronavigation with computer vision and real-time tracking for bedside external ventricular drain...

Area of Science:

• Neurosurgery
• Medical Imaging
• Computer Vision

Background:

• Image guidance in neurosurgery is hindered by the lack of real-time tracking for moving patients.
• Freehand placement of external ventricular drains carries risks of inaccuracy, hemorrhage, and brain injury.

Purpose of the Study:

• Introduce and validate a novel image registration and real-time tracking system.
• Enable frameless stereotactic neuronavigation and catheter placement in non-immobilized patients.

Main Methods:

• Developed a computer vision algorithm for automatic, marker-less image registration.
• Fused pre-procedure CT scans with live 3D camera images (Snap-Surface).
• Incorporated patient movement via AI-driven recalibration (Real-Track) and calculated surface and target registration errors (SRE/TRE).

Main Results:

• Achieved submillimetric accuracy (mean SRE: 0.429 ± 0.108 mm) in cadaveric specimens.
• Maintained TRE under 1.2 mm during rapid patient movements (roll, pitch, yaw).
• Demonstrated consistent accuracy regardless of surgical draping or lighting conditions, with a recalibration time under 0.23 seconds.

Conclusions:

• The computer vision-based system offers real-time tracking with submillimetric accuracy and millimetric catheter placement accuracy.
• This technology can improve safety and reduce complications in bedside ventriculostomy.
• The system has potential applications in other frameless stereotactic procedures for awake, non-immobilized patients.