Bayesian Inference Framework to Identify Skin Material Properties in vivo from Active Membranes

Area of Science:

• Biomedical Engineering
• Dermatology
• Computational Mechanics

Background:

• Accurate in vivo characterization of skin mechanical properties is crucial for dermatology and surgery.
• Current noninvasive methods struggle to capture skin's complex nonlinear and anisotropic behavior.

Purpose of the Study:

• To develop a novel Bayesian inference framework for patient-specific skin biomechanics assessment.
• To overcome limitations of existing techniques in characterizing nonlinear and anisotropic skin properties.

Main Methods:

• Utilized active membranes to induce controlled skin deformations for property inference.
• Developed a finite element model (FEM) of skin-membrane interaction, parameterized with the Holzapfel-Gasser-Ogden model.
• Constructed a data-driven surrogate model using principal component analysis and Gaussian process regression to accelerate Bayesian sampling.

Main Results:

• Enabled probabilistic inference of key skin parameters: shear modulus, fiber stiffness, dispersion, and orientation.
• Demonstrated accurate parameter recovery even with moderate noise in synthetic studies.
• Showed that multi-frame or multi-membrane observations significantly improve parameter identifiability.

Conclusions:

• The proposed active membrane platform offers a viable approach for noninvasive in vivo skin biomechanics assessment.
• The method successfully infers skin biomechanics without direct force measurements, relying on known membrane properties and strain field measurements.