Informatics colloquium 20. 3. Image analysis and synthesis in the context of cell tracking
Informatics colloquium 20. 3. 2018, 14:00 lecture hall D2
RNDr. Vladimír Ulman, Ph.D., FI MU
Abstract: The present biomedical research increasingly relies on automated
processing and analysis of large amount of image data. Two fundamental tasks in
this area are automated cell segmentation and tracking. The task of cell
segmentation is to split the image into several disjoint regions, each
accurately outlining every single cell. The task of cell tracking is to link
together the regions that represent the same cell in a time-lapse image
sequence. This allows for time-resolved quantitative description of events and
changes in displayed cell behaviour. Needless to say, the accuracy of the two
tasks influences accuracy of the consequent biological analysis.
This talk will be focused on ways to evaluate performance quality of
segmentation and tracking algorithms. In particular, I will review the
evaluation protocol implemented in the Cell Tracking Challenge, an international
benchmarking initiative that FI MU is part of, and identify current challenges
in the field. Last but not least, I will give an outlook on a artificial image
synthesis, which I see as a prospective way to foster the future development in
the field.