Objective evaluation for endoscopic sinus surgeries' proficiency by image analysis for postoperative 3D sinus models.

OBJECTIVE

Objective assessment for endoscopic sinus surgery skill proficiency is challenging, partly due to the great diversity of anatomies. Recently developed 3D-printed sinus models provide standardized anatomy with realistic tissue-feel. In this study, we developed a method to evaluate endoscopic sinus surgery skill proficiency through pre- and post-operative image analysis of CT images of dissected 3D sinus models and validated it by comparing to other surgical proficiency indices.

METHODS

Forty-seven otolaryngologists performed endoscopic sinus surgery for 3D sinus models. CT images of each postoperative 3D model were taken. The region growing method, an image segmentation technique, was utilized to assess the surgical space expansion. The region of interest (ROI) was defined as the space within nasal cavity and paranasal sinuses and continuous from the anterior nostril with a width of >4.00 mm kept. The volume of ROI was estimated and compared with conventional assessment methods such as board-certification, the objective structured assessment of technical skills score, and the number of experienced surgery cases. Further, the 3D-CG images of ROI were created. The completion of surgical steps within the allocated time was evaluated based on the 3D-CG and compared with the recorded endoscopic video.

RESULTS

The volume in postoperative models was 24.48±4.34 cm3 across all participants, respectively. The postoperative volume was significantly larger in board-certified participants (certified; 25.51±4.22 cm3 vs. uncertified; 22.87±4.14 cm3, p = 0.03). The volume is positively correlated to the objective structured assessment of technical skills score (r = 0.72, p < 0.01) and the number of experienced cases (r = 0.51, p = 0.01). The accuracy of the 3D-CG judgement for completion of a surgical middle meatal antrostomy, sphenoidotomy, and frontal sinusotomy were 91.3 %, 69.6 %, and 84.8 %, when compared to recorded endoscopic video, with high sensitivity and specificity.

CONCLUSION

The combination of standardized 3D models and image analysis provides a simple and objective tool for the assessment of surgical skill proficiency.