Modeling "Textured" Bones in Virtual Human Phantoms.

Area of Science:

• Medical Imaging
• Computational Anatomy
• Biomedical Engineering

Background:

• Current virtual human phantoms (XCAT) lack detailed bone structures.
• Realistic bone modeling is crucial for accurate medical imaging simulations.
• Existing models do not capture the complex textures of cortical and trabecular bone.

Purpose of the Study:

• To create detailed and realistic models of spinal, rib, and sternum bones.
• To integrate these enhanced bone models into the XCAT phantom library.
• To improve the realism of virtual human phantoms for medical imaging applications.

Main Methods:

• Cortical bone modeled using 3D morphological erosion based on CT data.
• Trabecular bone texture synthesized using a power law algorithm and high-resolution images.
• Synthesized textures incorporated into XCAT phantoms for enhanced realism.

Main Results:

• The 3D power spectrum of anatomical images confirmed power law behavior (R² = 0.84).
• The texture synthesis algorithm successfully emulated real anatomical bone images.
• Simulated CT images of phantoms with textured bones were more realistic than original XCAT models.

Conclusions:

• The developed method significantly enhances the realism of bone structures in virtual human phantoms.
• Textured phantoms can enable more accurate virtual clinical trials in medical imaging.
• This approach addresses limitations of real-world trials, such as cost and patient risk.