A machine learning framework for automated diagnosis and computer-assisted planning in plastic and reconstructive...

Area of Science:

• Computer-aided surgery
• Medical imaging
• Machine learning in healthcare

Background:

• Current computational tools for plastic and reconstructive surgery are imprecise and time-consuming, limiting their clinical adoption.
• Existing computer-assisted surgical planning systems, while beneficial, are often complex and require significant manual input, hindering their use in patient communication and decision-making.

Purpose of the Study:

• To introduce the first large-scale clinical 3D morphable model for plastic and reconstructive surgery.
• To develop a machine learning framework for automated diagnosis, risk stratification, and treatment simulation in facial surgery.

Main Methods:

• Developed a machine learning-based framework utilizing supervised learning.
• Trained and validated the model on 4,261 faces from healthy volunteers and orthognathic surgery patients.
• The model processes 3D scans for automated diagnosis and patient-specific treatment planning.

Main Results:

• The model achieved high diagnostic accuracy: 95.5% sensitivity and 95.2% specificity.
• Surgical outcome simulations demonstrated a mean accuracy of 1.1 ± 0.3 mm.
• The framework enables fully automated diagnosis and treatment planning from 3D scans.

Conclusions:

• The 3D morphable model significantly enhances efficiency in clinical decision-making for plastic and reconstructive surgery.
• This AI-driven approach improves understanding of facial morphology in primary and secondary surgical interventions.
• The model has the potential to streamline surgical planning, improve patient outcomes, and facilitate better doctor-patient communication.