Related Experiment Video
Updated: Jan 12, 2026
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.3K
Surrogate modeling of Cellular-Potts agent-based models as a segmentation task using the U-Net neural network
Tien Comlekoglu1,2, J Quetzalcóatl Toledo-Marín3,4, Tina Comlekoglu5
1Department of Biomedical Engineering, University of Virginia, Charlottesville, Virginia, United States of America.
Plos Computational Biology
|November 3, 2025
Summary
We developed a deep learning surrogate model to speed up complex biological simulations. This convolutional neural network accelerates Cellular-Potts model (CPM) simulations by 562x, capturing key biological behaviors like vessel growth.
Area of Science:
- Computational Biology
- Biophysics
- Deep Learning Applications
Background:
- Cellular-Potts models (CPMs) are widely used for simulating multicellular biological systems.
- CPMs are computationally intensive due to explicit agent interactions and partial differential equations (PDEs).
Purpose of the Study:
- To develop a computationally efficient surrogate model for CPMs using deep learning.
- To accelerate the simulation of in vitro vasculogenesis using a U-Net based convolutional neural network (CNN).
Main Methods:
- Developed a CNN surrogate model with a U-Net architecture incorporating periodic boundary conditions.
- Trained the surrogate model to predict 100 Monte-Carlo steps (MCS) ahead.
- Evaluated the surrogate model's ability to capture emergent behaviors in CPM simulations.
Main Results:
- Achieved a 562-fold acceleration in simulation speed compared to single-core CPU execution.
- The surrogate model accurately predicted emergent behaviors like vessel sprouting, extension, anastomosis, and lacunae contraction up to 300 MCS.
- Deep learning demonstrated potential for efficient CPM simulation acceleration.
Conclusions:
- Deep learning-based surrogate models offer a significant speedup for computationally expensive CPM simulations.
- This approach enables faster investigation of complex biological processes like vasculogenesis.
- The U-Net CNN surrogate model shows promise for advancing computational biology research.
