Automated Cell Segmentation, Tracking and Quantitative Analysis: Mathematical Methods and Algorithms
Abstract
Recent developments in time-lapse microscopy enable the observation and quantification of cell-cycle progression, cell migration, and growth control of live cells. Imaging and quantitative analysis techniques are being systematically integrated in biological and medical studies that investigate cellular immune response, embryonic development, tumorigenesis, and drug effects. The tasks of detecting and tracking individual cells in a time series of images are critical components of quantitative analyses. The large volume of data produced by fluorescence microscopy and imaging modalities further emphasizes the need for automated and robust techniques that can address the challenges in accurate detection and segmentation as well as tracking. The topic of this proposal is the development of techniques for fully automated cell segmentation, tracking, event handling, and quantification. We aim to develop automated methods that successfully (i) detect and (ii) track cells enabling the analysis of their static and dynamic behavior including cell morphology, cell migration, and changes in cell states (mitosis and apoptosis, for example). Another objective of our study is to conduct performance evaluation experiments for the main stages of cell segmentation and tracking. We will pursue a solution of the cell segmentation problem in the joint spatio-temporal domain to overcome weaknesses of previous works that operate only on the spatial domain of each frame. We will develop PDE-based spatio-temporal motion diffusion techniques to detect the cell motion. To refine cell delineation accuracy produced by motion diffusion-based segmentation, we propose to use energy minimizing geometric active contours that assume a piece-wise constant image region model as a special case of the Mumford-Shah segmentation framework. We will introduce temporal linking of the region-based level sets to allow for faster convergence and to resolve non-convexity that affects energy-based minimization that is typical in image analysis inverse problems. In the cell tracking part of this work we will adopt variational methods for joint local-global optical flow computation to estimate the cell motion. We will utilize the predicted cell motion along with cell characteristics in a probabilistic Maximum Likelihood decision strategy assuming Markov dependency to find cell correspondences between consecutive frames. To perform track linking and to identify the cell states in the time-lapse sequence we propose to find the solution that minimizes a global cost function defined over the set of all cell tracks by a heuristic approach. We will compute morphological, motility, diffusivity, and velocity measures using the time-lapse images, the cell label maps, and the tracking data. Finally, we will explore the application of these descriptors to tissue characterization. We will evaluate the performance of our proposed system on several datasets of fluorescence microscopy images with varying levels of difficulty. We will validate the cell segmentation and tracking stages both individually and as a joint system against reference standards that were manually generated. We plan to calculate measures of region overlap between the proposed and reference delineations to measure segmentation accuracy. Furthermore we will evaluate cell tracking accuracy by calculating the similarity of acyclic graphs constructed from the cell tracks produced by our system versus the reference standard.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 09, 2020
- Source ID
- W911NF2010095
Entities
People
- Sokratis Makrogiannis
Organizations
- Army Contracting Command
- Delaware State University
- United States Army