Human-Computer Vision Collaborative Measurement of Surgical Exposure and Length in Endonasal Endoscopic Skull Base...

Area of Science:

• Neurosurgery
• Medical Imaging
• Computer Vision

Background:

• Endonasal endoscopic approach (EEA) is a minimally invasive technique.
• Quantitative analysis of surgical exposure in EEA is crucial for surgical assessment.
• Current methods for measuring surgical exposure can be time-consuming.

Purpose of the Study:

• To develop and validate a computer vision (CV)-based system for quantitative analysis of surgical exposure in EEA.
• To assess the accuracy and efficiency of the CV system compared to existing methods.
• To investigate the association between surgical exposure metrics and residual tumors.

Main Methods:

• A CV algorithm was developed and trained using EEA operative videos.
• Pixel measurements of length and area were calibrated using a reference instrument.
• The CV model was validated using intercarotid distance and sellar height measurements.
• Accuracy was assessed on training, test-1, and test-2 datasets (95.1%, 95.8%, 96.2%).

Main Results:

• The CV-based system achieved high accuracy in analyzing surgical exposure.
• CV-based analysis reduced the time for area measurement by 89% compared to neuronavigation (p < 0.001).
• Smaller lateral and area surgical exposures were significantly associated with residual tumors (p = 0.001 and p = 0.024, respectively).

Conclusions:

• Computer vision offers an accurate and efficient method for quantifying surgical exposure in EEA.
• The CV system can significantly reduce the time required for surgical area measurements.
• AI and CV applications can expedite quantitative surgical exposure analysis in EEA surgeries, potentially improving patient outcomes.