Elastic parameter identification of three-dimensional soft tissue based on deep neural network.

Area of Science:

• Biomedical Engineering
• Computational Mechanics
• Medical Imaging

Background:

• Accurate soft tissue elastic parameter identification is crucial for virtual surgery and deformation simulation.
• Current methods often assume fixed elasticity, limiting patient-specific modeling and clinical applicability.
• Irregular soft tissue structures pose challenges for existing elasticity modeling techniques.

Purpose of the Study:

• To develop a novel method for identifying human soft tissue elastic parameters using finite element analysis and deep neural networks.
• To enable patient-specific elasticity distribution prediction from full-field displacement data.
• To improve the accuracy of virtual surgery simulations and soft tissue deformation modeling.

Main Methods:

• Utilized the finite element method (FEM) combined with a deep neural network (UNet).
• Input required: full-field displacement data of soft tissues under external loads.
• Validated using experimental data and clinical data from rhinoplasty surgeries.

Main Results:

• Achieved over 99% accuracy in predicting elastic parameters.
• Clinical validation demonstrated a reduction in finite element deformation simulation error by over 80% compared to traditional methods.
• Successfully predicted elastic distribution for irregularly structured soft tissues.

Conclusions:

• The proposed FEM-UNet method effectively identifies human soft tissue elastic parameters.
• This approach significantly enhances computational accuracy in virtual surgery and soft tissue deformation modeling.
• The method offers a promising solution for patient-specific modeling in clinical practice.