Image-Based Navigation for Functional Endoscopic Sinus Surgery Using Structure From Motion.

Area of Science:

• Medical Imaging
• Computer-Aided Surgery
• Otolaryngology

Background:

• Functional Endoscopic Sinus Surgery (FESS) is crucial for chronic sinusitis but carries risks due to proximity to critical structures.
• Current navigation systems in FESS have tracking errors exceeding 1 mm, hindering accurate overlay of virtual anatomy onto endoscopic views.
• Precise anatomical registration is vital for improving FESS safety and efficacy.

Purpose of the Study:

• To develop and evaluate a novel method for registering endoscopic images to CT data for enhanced FESS navigation.
• To improve the accuracy of overlaying critical anatomical structures onto real-time endoscopic imagery during FESS.
• To reduce surgical errors and enhance patient safety in FESS procedures.

Main Methods:

• Utilized Structure from Motion (SfM) to generate 3D point clouds from endoscopic video sequences.
• Employed the Iterative Closest Point (ICP) algorithm to register SfM-generated point clouds to patient-specific 3D sinus meshes.
• Estimated the scale of the point cloud by measuring endoscope motion within the video sequence.

Main Results:

• Achieved an average registration error of 1.21 mm for erectile tissues and 0.91 mm for non-erectile tissues.
• The SfM + ICP method demonstrated rapid execution, completing in under 7 seconds using as few as 15 frames (0.5 seconds of video).
• The system showed promising accuracy with a reasonable initial registration estimate.

Conclusions:

• The proposed SfM + ICP method offers a viable solution for improving endoscopic image to CT registration in FESS.
• This technique has the potential to significantly enhance surgical navigation accuracy and patient safety.
• Further clinical validation and robustness improvements are warranted for widespread adoption.