# Predicting Brain Morphogenesis via Physics-Transfer Learning

arXiv: https://arxiv.org/abs/2509.05305 · HTML: https://mlsanity.com/paper/2509.05305/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 12 of its 70 references that are in the map.
- First posted on arXiv 22 Aug 2025.

## Builds on (most cited first, 12 of 12)

- [DeepONet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operators](https://mlsanity.com/paper/1910.03193/) (2019)
- [Learning Mesh-Based Simulation with Graph Networks](https://mlsanity.com/paper/2010.03409/) (2020)
- [Deep Learning and the Information Bottleneck Principle](https://mlsanity.com/paper/1503.02406/) (2015)
- [E(3)-Equivariant Graph Neural Networks for Data-Efficient and Accurate Interatomic Potentials](https://mlsanity.com/paper/2101.03164/) (2021)
- [Physics-informed learning of governing equations from scarce data](https://mlsanity.com/paper/2005.03448/) (2020)
- [Weak baselines and reporting biases lead to overoptimism in machine learning for fluid-related partial differential equations](https://mlsanity.com/paper/2407.07218/) (2024)
- [How I failed machine learning in medical imaging -- shortcomings and recommendations](https://mlsanity.com/paper/2103.10292/) (2021)
- [Neural tuning and representational geometry](https://mlsanity.com/paper/2104.09743/) (2021)
- [Brain Latent Progression: Individual-based Spatiotemporal Disease Progression on 3D Brain MRIs via Latent Diffusion](https://mlsanity.com/paper/2502.08560/) (2025)
- [A Data-Driven Approach to Morphogenesis under Structural Instability](https://mlsanity.com/paper/2308.11846/) (2023)
- [Discovering High-Strength Alloys via Physics-Transfer Learning](https://mlsanity.com/paper/2403.07526/) (2024)
- [Validating the predictions of mathematical models describing tumor growth and treatment response](https://mlsanity.com/paper/2502.19333/) (2025)

## Abstract (arXiv)

Brain morphology is shaped by genetic and mechanical factors and is linked to biological development and diseases. Its fractal-like features, regional anisotropy, and complex curvature distributions hinder quantitative insights in medical inspections. Recognizing that the underlying elastic instability and bifurcation share the same physics as simple geometries such as spheres and ellipses, we developed a physics-transfer learning framework to address the geometrical complexity. To overcome the challenge of data scarcity, we constructed a digital library of high-fidelity continuum mechanics modeling that both describes and predicts the developmental processes of brain growth and disease. The physics of nonlinear elasticity from simple geometries is embedded into a neural network and applied to brain models. This physics-transfer approach demonstrates remarkable performance in feature characterization and morphogenesis prediction, highlighting the pivotal role of localized deformation in dominating over the background geometry. The data-driven framework also provides a library of reduced-dimensional evolutionary representations that capture the essential physics of the highly folded cerebral cortex. Validation through medical images and domain expertise underscores the deployment of digital-twin technology in comprehending the morphological complexity of the brain.