Adaptive infrared patterns for microscopic surface reconstructions

Area of Science:

• Medical Imaging
• Surgical Technology
• Computer Vision

Background:

• Current 3D reconstruction methods struggle with surgical environments due to challenging illumination and homogeneous tissue surfaces.
• Accurate surface reconstructions are crucial for surgical planning, intraoperative guidance, and quality assurance, particularly in complex procedures like ENT surgeries.

Purpose of the Study:

• To develop and validate an adaptive near-infrared (NIR) pattern projection technique for enhanced multi-zoom microscopic surface reconstructions.
• To overcome limitations of existing methods in creating accurate 3D models of surgical sites.

Main Methods:

• An adaptive NIR pattern projector was employed to illuminate surgical scenes, generating optimized patterns for dense, multi-zoom stereoscopic surface reconstructions.
• The novel approach was integrated without disrupting the clinical workflow and validated against state-of-the-art methods and CT data.

Main Results:

• The adaptive NIR pattern method achieved surface reconstruction errors of 0.5–0.7 mm, significantly outperforming other approaches (1–1.9 mm).
• Reconstructions were performed across 5 zoom levels, with error visualizations provided via heat maps.

Conclusions:

• Adaptive NIR pattern projection enables dense and accurate microscopic surface reconstructions at various zoom levels, even on small, homogeneous surfaces.
• This technology holds potential for improving microscopic interventions, such as lateral skull base surgeries, and integrating intraoperative data with preoperative imaging.