Biomechanical model for computing deformations for whole-body image registration: A meshless approach

Area of Science:

• Computational mechanics
• Medical imaging
• Biomedical engineering

Background:

• Patient-specific biomechanical models aid medical image registration for large inter-image differences.
• Generating these models is time-consuming due to complex organ geometry.
• Existing meshless methods are limited to 2D and single organs.

Purpose of the Study:

• To develop and apply 3D comprehensive patient-specific nonlinear biomechanical models for whole-body image registration.
• To overcome limitations of traditional segmentation-based approaches.
• To enable accurate prediction of 3D deformation fields.

Main Methods:

• Implementation of meshless Total Lagrangian explicit dynamics algorithms for 3D biomechanical models.
• Application of the fuzzy c-means algorithm for assigning mechanical properties without image segmentation.
• Prediction of 3D deformation fields for whole-body image registration.

Main Results:

• Deformations predicted by the meshless approach closely match those from validated finite element models.
• The fuzzy c-means algorithm successfully assigned mechanical properties without image segmentation.
• Evaluation using Hausdorff distance confirmed successful registration of most image features.

Conclusions:

• 3D meshless biomechanical models offer a viable and efficient alternative for whole-body image registration.
• The proposed method eliminates the need for image segmentation, simplifying the process.
• This approach demonstrates high accuracy in predicting deformations and achieving successful image registration.