Transforming body composition of computational phantoms based on form-finding dynamics

Area of Science:

• Medical Imaging
• Computational Anatomy
• Biomedical Engineering

Background:

• Computational phantoms lack diverse body composition representation, crucial for accurate virtual imaging trials.
• Existing methods for phantom modification are time-consuming and resource-intensive.
• Physics engines offer a balance of speed and realism for simulating anatomical variations.

Purpose of the Study:

• To model variations in visceral fat (VF), subcutaneous fat (SF), and skeletal muscle (SM) within the XCAT phantom library.
• To simulate the impact of body composition changes on surrounding organs.
• To enhance the representation of diverse patient body types in computational models.

Main Methods:

• Utilized the Kangaroo2 physics engine for algorithmic modeling of tissue changes.
• Converted phantom image data into triangular meshes for physics-based simulation.
• Applied diffeomorphic displacement fields to simulate large deformations in high-resolution phantom images.

Main Results:

• Successfully altered phantom volumes and appearances, including local organ deformation.
• Quantified significant volume changes for muscle and fat growth and reduction.
• Calculated z-scores to position modified phantoms within population data.

Conclusions:

• Developed a technique to manipulate phantom muscle and fat composition effectively.
• Demonstrated qualitative and quantitative observation of body composition changes.
• Workflow enhances phantom libraries for better patient body composition variability representation.