Predictive Modeling of Human Skin Deformation and Growth During Tissue Expansion in Postmastectomy Breast Reconstruction

Breast reconstruction using tissue expanders is the primary treatment option following mastectomy. Although skin growth in response to chronic supra-physiological stretch is well-established, individual patient factors such as breast shape, volume, skin prestrain, and mechanical properties, create unique deformation and growth patterns. The inability to predict skin growth and deformation prior to treatment often leads to complications and suboptimal esthetic outcomes. Personalized predictive simulations offer a promising solution to these challenges. We present a pipeline for predictive computational models of skin growth in tissue expansion. At the start of treatment, we collect three-dimensional (3D) photos and create an initial finite element model. Our framework accounts for uncertainties in treatment protocols, mechanical properties, and biological parameters. These uncertainties are informed by surgeon input, existing literature on mechanical properties, and prior research on porcine models for biological parameters. By collecting 3D photos longitudinally during treatment, and integrating the data through a Bayesian framework, we can systematically reduce uncertainty in the predictions. Calibrated personalized models are sampled using Monte Carlo methods, which require thousands of model evaluations. To overcome the computational limitations of directly evaluating the finite element model, we use Gaussian process surrogate models. We anticipate that this pipeline can be used to guide patient treatment in the near future.