Automatic Segmentation of Mechanically Inhomogeneous Tissues Based on Deformation Gradient Jump

Area of Science:

• Biomechanics
• Materials Science
• Medical Imaging

Background:

• Tissue mechanical behavior is often heterogeneous due to variations in properties, injury, or disease.
• Accurate full-field displacement tracking via advanced imaging provides strain field data.
• Traditional methods using constitutive equations struggle with spatially varying material properties in heterogeneous tissues.

Purpose of the Study:

• To develop a computational approach for dissecting heterogeneous soft tissue samples into homogeneous subdomains.
• To enable the determination of spatially varying constitutive equations for soft tissues.
• To accurately characterize regions with locally similar material properties in complex biological samples.

Main Methods:

• A novel partitioning technique using betweenness-based graphical analysis on deformation gradient fields.
• Identification of subdomain boundaries based on discontinuities in the strain field.
• Application to simulated tissues, tissue analogs (PDMS, collagen gels), and native soft tissues (tendon, aorta, cardiac tissue).

Main Results:

• Successfully determined the shape, size, and location of regions with similar material properties.
• Validated the method on simulated tissues with controlled variations in anisotropy, alignment, stiffness, and nonlinearity.
• Demonstrated efficacy on tissue analogs and native soft tissues exhibiting natural, pathological, and active variations.

Conclusions:

• The computational dissection method accurately maps heterogeneous material properties in soft tissues.
• This approach allows for the study of small tissue specimens with unknown and irregular inhomogeneities.
• Enables a deeper understanding of tissue mechanics in both healthy and diseased states.