Multidimensional light field endoscope for robotic and AI guided surgery

Area of Science:

• Biomedical Engineering
• Surgical Technology
• Optical Imaging

Background:

• Robotic surgery and AI-driven diagnostics require advanced imaging for precision.
• Current endoscopes are integrating 3D, near-infrared (NIR) fluorescence, and AI capabilities.
• Mimicking biological vision systems offers potential for enhanced endoscopic imaging.

Purpose of the Study:

• To develop a multidimensional light field endoscope for improved robotic and AI-guided surgery.
• To integrate stereoscopic vision, polarized spectral vision, and NIR fluorescence imaging.
• To enhance surgical site visualization for navigation and tissue differentiation.

Main Methods:

• Designed a multidimensional light field endoscope capturing multi-view, multi-polarization color images and NIR fluorescence.
• Developed algorithms for color restoration, 3D reconstruction, polarization extraction, and electronic-stain-free fluorescence imaging.
• Conducted experiments to validate imaging performance across different channels.

Main Results:

• The endoscope successfully acquired color 3D images with polarization information and ICG NIR fluorescence data.
• Demonstrated capability for scene color 3D reconstruction and polarization degree computation.
• Achieved electronic-stain-free color fluorescence imaging, distinguishing lesions from vital organs.

Conclusions:

• The multidimensional light field endoscope significantly enhances the intelligence of robotic and AI-guided surgeries.
• Provides crucial data for surgical navigation, lesion detection, and vital organ identification.
• Represents a significant advancement towards "wisdom-eyes" for future surgical systems.