A review on AI-based medical image computing in head and neck surgery

Area of Science:

• Medical Image Computing
• Artificial Intelligence
• Head and Neck Surgery

Background:

• Head and neck surgery presents significant challenges due to complex anatomy and high risks.
• Accurate preoperative planning using medical image computing (MIC) is crucial for improving surgical outcomes and patient survival.
• Deep learning has emerged as a key technology in advancing MIC research.

Purpose of the Study:

• To review the application of deep learning-based MIC in head and neck surgery.
• To systematically summarize deep learning methods for segmentation, landmark detection, and registration of head and neck medical images.
• To provide guidance for researchers and clinicians in this field.

Main Methods:

• Literature retrieval from Web of Science (Jan 2015 - May 2022).
• Selection of papers from leading journals and conferences (e.g., IEEE TMI, MedIA, MICCAI).
• Systematic review and summarization of deep learning applications in segmentation, landmark detection, and registration.

Main Results:

• Identified 65 references on automatic segmentation (organs, tumors, structures).
• Included 15 references on automatic landmark detection (cephalometric, craniomaxillofacial images).
• Covered 8 references on multimodal image registration using deep learning networks.

Conclusions:

• Deep learning significantly aids in head and neck medical image analysis for surgical planning.
• Current limitations and future research directions in deep learning for MIC are discussed.
• This review serves as a valuable reference for professionals in head and neck surgical image analysis.