The Importance of Computer Generated Data in Fluorescence Microscopy
| Authors | |
|---|---|
| Year of publication | 2014 |
| MU Faculty or unit | |
| Citation | |
| Description | In fluorescence microscopy, the proper evaluation of image segmentation and tracking algorithms, that are further used for observation of some particular processes or directly to diagnosis, is still an open task. The problem is based on the fact that currently there exists no exact knowledge, how the microscopic specimens look like if observed without any degradation introduced by the microscope setup. In the past, the only available quality measurement of the algorithms was an expert's knowledge. The expert either classified the results of selected algorithms or provided an annotation of some real image dataset that was further used for evaluation purposes. Both ways however suffer from two main issues. First, the expert's evaluation is nondeterministic. Second, for higher dimensional data (sequences of 2D or 3D images) the handmade annotation is impractical or even impossible. For this reason, the computer generated data, naturally accompanied by their ground truth, have started to appear. In this talk, a survey of the most important toolkits employed for generation of synthetic image data containing cells in fluorescence microscopy is given. We also mention the results achieved by the group CBIA. We will present a toolbox that can generate fully 3D digital phantoms of specific cellular components along with their corresponding images degraded by specific optics and electronics. The images can represent static scenes (fixed cells) as well as time-lapse sequences (living cells). Such synthetically generated images can serve as a benchmark dataset for measuring the quality of various segmentation and tracking algorithms. |
| Related projects: |