U-net for auricular elements segmentation: a proof-of-concept study.

Area of Science:

• Medical Imaging
• Computer Vision
• Biomedical Engineering

Background:

• Convolutional neural networks (CNNs) are vital for automating anatomical segmentation in medical imaging, reducing time and costs.
• Accurate segmentation of auricular elements is essential for creating patient-specific surgical guides for ear reconstruction.

Purpose of the Study:

• To implement and evaluate a U-net CNN for segmenting key ear structures (helix, antihelix, tragus-antitragus, concha) from 3D ear models.
• To assess the performance of the U-net architecture in segmenting auricular elements for surgical guide development.

Main Methods:

• A dataset of 131 ear depth map images was curated.
• The U-net convolutional neural network architecture was employed for segmentation.
• Data was split into training (70%), validation (15%), and testing (15%) sets.

Main Results:

• The U-net model achieved high accuracy in segmenting auricular elements.
• The network demonstrated 97% accuracy on the validation dataset.
• Successful segmentation of helix, antihelix, tragus-antitragus, and concha contours was achieved.

Conclusions:

• The U-net CNN is a highly effective tool for the automated segmentation of ear elements from depth map images.
• This automated segmentation process significantly contributes to the development of surgical guides for external ear reconstruction.
• The approach offers a promising solution for reducing manual labor and enhancing precision in reconstructive surgery planning.