Image Segmentation

Image segmentation understood as the process of partitioning a digital image into meaningful regions is a problem that cannot have a universal general solution and for each application an optimal solution has to be searched. Our group has contributed with several segmentation techniques suitable for the segmentation of cellular objects such as cell nuclei, whole cells or their parts (e.g. chromosome territories or individual genes) as well as cells in tissues. The techniques has been designed for various imaging modalities (most notably fluorescence microscopy) and different dimensions including 2D, 3D, as well as 3D+time.

Historically, the major focus and know-how of the group lies in whole cell and cell nucleus segmentation in fluorescence microscopy (e.g., DAPI stained nuclei), from 2D in 1990s to 3D+time nowadays. The methods has been based at various principles:

Another major topic addressed in our group is small object (spot, dot, particle) detection and segmentation again from 2D to 3D+time. This comprises targets such as genes, telomeres, and protein sites. We have also published a study of methods for the 3D spot detection in confocal microscopy.

Besides the major modality addressed in our group, fluorescence microscopy, we have worked also on other modalities, such as phase-contrast images and transmission electron microscopy.

The research of segmentation techniques in our group is still active and is or has been supported by various research projects. The most important are: