Artificial intelligence system for automatic maxillary sinus segmentation on cone beam computed tomography images.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Radiology

Background:

• Accurate segmentation of maxillary sinuses (MS) in cone beam computed tomography (CBCT) is crucial for various clinical applications.
• Manual segmentation is time-consuming and prone to inter-observer variability.

Purpose of the Study:

• To develop and evaluate an artificial intelligence (AI) model for automated MS segmentation in CBCT images.
• To assess the performance of the nnU-Net v2 deep learning model for this task.

Main Methods:

• A dataset of 101 CBCT scans was utilized, with 80 for training, 11 for validation, and 10 for testing.
• The nnU-Net v2 deep learning model was employed for MS segmentation.
• Performance was evaluated using F1-score, accuracy, sensitivity, precision, AUC, Dice coefficient (DC), 95% Hausdorff distance (95% HD), and Intersection over Union (IoU).

Main Results:

• The AI model achieved high segmentation performance, with an F1-score of 0.96, accuracy of 0.99, sensitivity of 0.96, and precision of 0.96.
• Additional metrics included AUC of 0.97, DC of 0.96, 95% HD of 1.19, and IoU of 0.93.

Conclusions:

• nnU-Net v2 based models can autonomously and accurately segment maxillary sinuses in CBCT images.
• The developed AI model shows significant potential for improving efficiency and consistency in MS segmentation.