This project will develop a framework to represent, analyze and interpret shapes extracted from images, supporting a wide range of biological investigations. The primary objectives are: (1) to develop a mathematical framework and computational tools for the quantification and analysis of shapes; (2) to integrate these computational models with machine learning and statistical inference methods to enable new discoveries, transforming imaging data into biological knowledge; (3) to deliver novel quantitative methodologies for shape analysis that start from a biological premise, rather than a purely geometric one. The aim is thus not only to quantitatively describe shape, but to develop methods for linking morphological variation to its underlying biological causes. To ensure that the project focuses on methods that are most promising to biology with significant breadth of application, model and tool development will be guided and supported by a set of diverse case studies, ranging from the sub-cellular to organismal scales.
Shape represents a complex and rich source of biological information that is fundamentally linked to underlying mechanisms and function. However, shape is still often examined on a qualitative basis in many disciplines in biology, an approach that is time consuming and prone to human subjectivity. While ad hoc quantitative methods do exist, they are often inaccessible to non-experts and do not easily generalize to a wide variety of problems. The inability of biologists to systematically link shape to genetics, development, environment, function and evolution often precludes advances in biological research spanning diverse spatial and temporal scales, from the movement of molecules within a cell to adaptive changes in organismal morphology. The primary goal of this project is to develop a new suite of widely applicable quantitative methods and tools into the study of biological shape to address the significant need for consistent and repeatable analysis of shape data.
Shape plays a key role in biological function across a whole range of scales, ranging from the shape of whole organs (the shape of a bat’s ear affects the acoustic sonar signal it hears) down to the spatial organization of proteins within individual cells (the arrangement of filaments within a single heart cell allows it to beat in synchrony with its neighbors). The goal of this project was to develop new computational techniques for measuring and analyzing shapes extracted from images that would be useful across a wide range of scientific settings. In order to extract useful measurements of shape from images, it is necessary to first detect and segment out the objects of interest in the image. Custom software is typically written to work on a particular set of images collected during an experiment, but often that software is not robust and fails to work well on new images collected by a different lab or using different protocols. We developed a more robust approach for automatically detecting cells in 3D microscopy images which can be easily adapted to new datasets. The user teaches the system how to detect cells by simply clicking on examples in their particular images. The software learns a model for the appearance of the cells, allowing it to rapidly and automatically detect them in future images. This software has been used to detect cell nuclei in a large dataset of images of the C. elegans nematode worm germline in order to help analyze how stem cells "decide" whether to renew or differentiate into particular cell types. Extracting locations of cells from microscopy images provides a detailed quantitative record of shape but it is still difficult to compare such records from different samples. To carry out a meaningful comparison, it is necessary to find corresponding structures in images of different individuals and see how they have moved relative to each other. We developed methods for finding corresponding cells in different animals based not only on the locations of the cells but also on image-based measurements of gene expression within individual cells. We used this system to compare patterns of gene expression in developing fruit fly embryos by computationally aligning them with a common reference shape and measuring what differences exist between corresponding cells. This became a particularly useful approach to try and understand basic questions of how genes control differences in shape across closely related species. By aligning data from different species of fruit fly we found that while the shapes, individual gene expression levels and the DNA sequence that controls these levels varies across species, the joint expression patterns are highly preserved. Some aspects of biological shape have a much less consistent appearance and are better treated in a statistical way as a texture. For example, the precise shape of a pollen grain may vary quite a bit from one individual grain to the next, but the shape, size and density of individual sculptural estimates on the surface is consistent. By measuring and modeling these local shape characteristics we have been able to develop methods for automatically identify the species from which a particular pollen grain came from, often with accuracy that is nearly as good as a trained expert. A similar analysis was used to measure the distribution of structures inside individual cells. By analyzing the statistics of how individual fibers were oriented within heart muscle cells, we were able to quantify the extent to which these heart muscle cells were aligning and beating in synchrony when grown on a special substrate. These computational tools for shape analysis are thus proving to be useful in a variety of biological and bioengineering applications.
