Many biological cells generate forces that allow them to migrate and change morphology. Cell migration plays an important role in many biological processes such as embryonic development, the inflammatory immune response system, wound repair, and tumor metastasis, and is therefore the object of a lot of biomedical research. Time-lapse microscopy, in which a microscope obtains images of cells at fixed intervals, is arguably the most important tool available to in detail study and characterize cell behavior such as proliferation, mitosis (cell division), apoptosis (cell death), migration, and morphology. Automation of and advances in microscopy hardware, and the introduction high-resolution digital image sensors, have enabled the acquisition of large amounts of time-lapse image data from biomedical experiments: The key open problem is no longer to acquire the data but instead to efficiently and reliably analyze it.
This talk will give an overview of current efforts, within the Department of Signal Processing at KTH, to develop automated methods applicable in the study of cell migration and morphology in time-lapse microscopy. The topic of cell tracking broadly falls within the area of multiple target tracking and computer vision, but key differences between tracking cells and tracking many other types of objects requires substantial renewed efforts. Apart from migrating, cells frequently adhere to each other and are segmented jointly, significantly change their morphology, divide, die, and are often imaged using imaging modalities in which they are hard to segment. In our work we aim to address the fundamental questions in multiple target tracking introduced by the cell tracking application. We will also in this talk present some of our efforts to encapsulate our results in software that is directly applicable as a tool for future biomedical research.
