Inferring epithelial tissue mechanics through data-efficient multi-fidelity modelling
Lead Research Organisation:
University of Sheffield
Department Name: Mathematics and Statistics
Abstract
How do the organs in our body control their shape and structure? We know that when they fail to do this, we get overgrowth, and frequently cancer. If we can understand these control systems, perhaps we will be able to understand what happens when tissues lose their correct structure, and go on to develop more effective treatments. Proper control of tissue structure comes about due to a balance of mechanical forces acting within, and between, the cells that make up a tissue. Therefore, quantifying the mechanics of cells and tissues is essential to understanding their shape and structure.
Recently, our team members generated a large set of images and videos showing how, in the lab, a special type of tissue forms a single sheet of cells strongly adhering to their neighbours. We also created a computer model that allowed us to explore how different balances of mechanical forces can lead to this sheet of cells having different shapes and structures. However, it is difficult to precisely pin down the strength of each mechanical force in the model from the images, because it takes a while for the model to run.
In this project, we will use new and existing artificial intelligence and machine learning methods to learn these mechanical properties much faster from our data, by bringing together information gleaned from simpler versions of our model in an automated way. As a result, we will come up with better estimates for these mechanical properties, and a better understanding of how cell sheets attain their characteristic shapes and structures. We will write software to do this that can also be used by other scientists studying similar problems in areas of biology ranging from cell and developmental biology to synthetic biology, regenerative medicine, and cancer research.
Recently, our team members generated a large set of images and videos showing how, in the lab, a special type of tissue forms a single sheet of cells strongly adhering to their neighbours. We also created a computer model that allowed us to explore how different balances of mechanical forces can lead to this sheet of cells having different shapes and structures. However, it is difficult to precisely pin down the strength of each mechanical force in the model from the images, because it takes a while for the model to run.
In this project, we will use new and existing artificial intelligence and machine learning methods to learn these mechanical properties much faster from our data, by bringing together information gleaned from simpler versions of our model in an automated way. As a result, we will come up with better estimates for these mechanical properties, and a better understanding of how cell sheets attain their characteristic shapes and structures. We will write software to do this that can also be used by other scientists studying similar problems in areas of biology ranging from cell and developmental biology to synthetic biology, regenerative medicine, and cancer research.
