Christian Held

Suche


A4 Automated and interactive learning image-analysis for fluorescence-microscopy by the example of microbial effectors in host cells

Principal investigator
Thomas Wittenberg

Mentor
Heinrich Sticht

PhD exam: 21.01.2014

Automated and interactive learning image-analysis for fluorescence-microscopy by the example of microbial effectors in host cells

The interpretation of micrographs plays an important role when analysing microbial effectors in host cells. In some cases this is a very repetitive and time consuming task. Due to a possible biased interpretation, results are sometimes hard to reproduce between different laboratories and institutes. To increase reproducibility and effectiveness, novel automated and interactive image processing methods are required.
One common task is counting the average number of effectors, e.g. Salmonella cells, which are located in the cytoplasm of a host cell. To resolve this task, special software is required to perform a correct segmentation of each individual effector cell. This procedure can be complicated by cell aggregates or variations within the specimens and should also allow segmentation of each cell’s cytoplasm. Therefore, the selection of the optimal method and its parameterization are crucial to obtain accurate and reproducible segmentation results. To date the selection of the optimal segmentation routine for a certain application requires expert knowledge about the characteristics of the segmentation task. This turns out to be an obstacle for researchers from biology or medicine. Therefore, the first part of this project is to incorporate novel, self-learning components that will be able to select the best segmentation method automatically from a pool of different available procedures. This selection will be based on a set of user defined, annotated ground-truth data, which can be generated by interactive tools, e.g. “live-wire”-segmentation or by marking (drawing) the object of interest. Then, the best available segmentation method should be selected based on a performance index, and can then be applied to perform a segmentation of the remaining objects in the micrographs.
In the second part of this research project, the semantic gap between scientist and machine should be bridged. As final aim this should allow the user to define parameters and tasks by using an intuitive user interface. Including this step, typical tasks from biology or medicine should be solved without requiring expert knowledge on image processing or the development of scripts.

Figure: Example segmentation of Salmonella cells performed with maximum intensity linking.

 

Publications

Held, C., Wiesmann, V., Palmisano, R., Ensser, A. and Wittenberg, T. (2012). Automated high throughput image analysis of the ND10 complex of KSHV cells. Conf Proc DGBMT accepted.

Held, C., Wenzel, J., Lang, R., Palmisano, R. and Wittenberg, T. (2012). Comparison of Methods for Splitting of Touching and Overlapping Macrophages in Fluorescent Micrographs. Conf Proc ICIAR, accepted.

Webel, R., Solbak, S. M., Held, C., Milbradt, J., Gross, A., Eichler, J., Wittenberg, T., Jardin, C., Sticht, H., Fossen, T. and Marschall, M. (2012). The nuclear import of isoforms of the cytomegalovirus kinase pUL97 is mediated by differential activity of NLS1 and NLS2 both acting through classical importin alpha binding. J Gen Viorol, doi: 10.1099/vir.0.040592-0.

Held, C., Wenzel, J., Webel, R., Marschall, M., Lang, R., Palmisano, R. and Wittenberg, T. (2011). Using Multimodal Information for the Segmentation of Fluorescent Micrographs with Application to Virology and Microbiology. Conf Proc IEEE Eng Med Biol Soc, pp. 6487-6490.

Wenzel, J., Held, C., Palmisano, R., Teufel, S., David, J.P., Wittenberg, T. and Lang, R. (2011) Measurement of TLR-Induced Macrophage Spreading by Automated Image Analysis: Differential Role of Myd88 and MAPK in Early and Late Responses. Front Physiol, 2:71.

Held, C., Palmisano, R., Häberle, L, Hensel, M. and Wittenberg, T. (2011) Comparison of parameter-adapted segmentation methods for fluorescence micrographs. Cytometry A, 79 (11), 933-945.

Webel, R., Milbradt, J., Auerochs, S., Schregel, V., Held, C., Nöbauer, K., Razzazi-Fazeli, E., Jardin, C., Wittenberg, T., Sticht, H. and Marschall, M. (2011) Two isoforms of the protein kinase pUL97 of human cytomegalovirus are differentially regulated in their nuclear translocation. J Gen Virol , 92, 638-649.

 

Presentations

July 2012 4th Annual Retreat, Erlangen School of Molecular Communication, Kloster Banz, Bad Staffelstein, Germany
Automated Analysis of Fluorescent Micrographs (With Minimum User Interaction)
Talk
     
October 2011 First International SFB 796 Conference: Mechanisms of viral host cell manipulations: from plants to humans, Bamberg, Germany
(Semi-)Automated image analysis with application to fluorescence microscopy
Poster
     
July 2011 3rd Annual Retreat, Erlangen School of Molecular Communication, Kloster Banz, Bad Staffelstein, Germany
Towards the Automated Analysis of Multimodal Micrographs
Talk and Poster
     
September 2010 2nd Annual Retreat, Erlangen School of Molecular Communication, Kloster Banz, Bad Staffelstein, Germany
Automated vs. interactive learning image-analysis
Talk
     
September 2009 First Annual Retreat, Erlangen School of Molecular Communication, Schloss Atzelsberg, Atzelsberg, Germany
Automated and interactive learning image-analysis
Talk