Structured light for touchless 3D registration in video-based surgical navigation.

Area of Science:

• Medical technology
• Surgical navigation
• Computer-assisted surgery

Background:

• Arthroscopic surgery presents visibility and maneuverability challenges.
• Video-based surgical navigation (VBSN) offers clinical benefits but requires difficult surface digitization.
• Current registration methods using touch probes are time-consuming.

Purpose of the Study:

• To present an off-the-shelf laser scanner for noninvasive registration in arthroscopy.
• To enable an increased area of reachable region during arthroscopic procedures.
• To overcome surface digitization challenges in VBSN.

Main Methods:

• Utilized a standard arthroscope and light projector with visual markers for real-time calibration.
• Employed a deep learning-based semantic segmentation technique to identify structures and mitigate outliers from nonrigid surfaces.
• Integrated a laser scanner for noninvasive surface digitization.

Main Results:

• Laser scanner with segmentation achieved registration errors below 2.2 mm in ex-vivo experiments.
• Errors were below 1 mm in the intercondylar region and for phantom experiments.
• The segmentation model effectively mitigated outliers from nonrigid structures.

Conclusions:

• The laser scanner is a viable, noninvasive alternative for VBSN, overcoming digitization challenges.
• This approach expands the reachable surgical region in arthroscopy.
• Future work will focus on hardware improvements for complex procedures.