Kainmüller Lab - Biomedical Image Analysis

Dagmar Kainmüller joined the BIH in May 2018 to establish the Junior Research Group 'Biomedical Image Analysis'.

Research focus

The Kainmueller Lab pursues theoretical advances in machine learning and combinatorial optimization to solve challenging image analysis problems in biology. Our aim is to facilitate scientific discovery via automated analysis of high-throughput microscopy data. We focus on capturing biological prior knowledge in machine learning models for accurate cell segmentation, annotation, and tracking, and develop computationally efficient solvers for the underlying optimization problems. This way we leverage biological priors, such as the stereotyped lineage of C. Elegans and known shape properties of Drosophila neurons, in a mathematically sound and effective way.

Research goals

 We aim at automated segmentation, annotation, matching, and tracking of cells in microscopy image data. We develop machine learning models that are able to exploit biological prior knowledge. Such knowledge is available at various levels of detail. For example, it is known that epithelial cells have approximately polygonal shapes, and form honeycomb-like grids. Our methods are able to exploit such knowledge for improved segmentation accuracy.

Another kind of biological prior our research is concerned with are stereotypies that have been identified in some model organisms, such as the nematode worm C. Elegans, where individual worms are exact copies of each other in terms of the number and function of their cell nuclei. This remarkable property allows for cell-level observations of gene expression made in different individuals to be merged into a common reference atlas. Establishing an exhaustive atlas of gene expression at the cell level will pave the way towards a fundamental understanding of how the genome of an organism encodes its development and function. However, to merge observations from different individuals into a reference atlas, cells have to be annotated, i.e. labelled with their unique biological names, in microscopy images. This is a hard and time-consuming task even for trained anatomists. In our C. Elegans project, we work on reducing the amount of required expert annotations to a viable minimum by not just automating the annotation task, but furthermore by training our underlying machine learning model in an unsupervised manner, i.e. without the need for manually annotated examples to learn from. We aim at phrasing such unsupervised training as a combinatorial optimization problem and develop a respective solver.

Similar to C. Elegans, the brain of the fruit fly Drosophila is stereotyped at the level of individual neurons. Our work on Drosophila aims at automated matching of in-vivo observations of neuron function in light microscopy to detailed observations of neuron connectivity in electron microscopy, and vice-versa.

Patient benefits

 Our generic automated methods for cell segmentation, classification, and tracking may help render novel analyses of patient-indivdual cell cultures viable in clinical practice. Furthermore, our work towards a fundamental understanding of basic scientific questions, such as how brains make decisions and form memories, and how a genome encodes organism development, may foster the development of novel therapies for developmental and neurological disorders.

About Dagmar Kainmüller

Prior to joining the BIH, Dagmar Kainmueller worked as an Independent Fellow at the HHMI Janelia Research Campus in Virginia, USA. From 2013 to 2016 she was an ELBE PostDoc in the lab of Gene Myers at the Max Planck Institute of Molecular Cell Biology and Genetics in Dresden, Germany. Dagmar pursued her PhD in medical image segmentation at the Zuse Institute Berlin and the University of Lübeck, for which she received the BVM Award in 2013.

Selected publications

Swoboda, P., Rother, C., Abu Alhaija, H., Kainmueller, D and Savchynskyy, B. (2017). A Study of Lagrangean Decompositions and Dual Ascent Solvers for Graph Matching. Proc. IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 1607-1616.

Dye, N.A., Popović, M., Spannl, S., Etournay, R., Kainmueller, D., Ghosh, S., Myers, E.W., Jülicher, F. and Eaton, S. (2017). Cell dynamics underlying oriented growth of the Drosophila wing imaginal disc. Development, doi:10.1242/dev.155069

Schlesinger, D., Jug, F., Myers, G., Rother, C., and Kainmueller, D. (2017). Crowd sourcing image segmentation with iaSTAPLE. Proc. IEEE International Symposium on Biomedical Imaging (ISBI), 401-405.

Royer, L., Richmond, D., Rother, C., Andres, B., and Kainmueller, D. (2016). Convexity Shape Constraints for Image Segmentation. Proc. CVPR, 402-410.

Richmond, D., Kainmueller*, D., Yang, M., Myers, G., and Rother, C. (2016). Mapping Auto-context Decision Forests to Deep ConvNets for Semantic Segmentation. Proc. British Machine Vision Conference (BMVC), 144.1-144.12. *shared first authors

Richmond, D., Kainmueller*, D., Glocker, B., Rother, C., and Myers, G. (2015). Uncertainty-driven Forest Predictors for Vertebra Localization and Segmentation. Proc. Medical Image Computing and Computer Assisted Intervention (MICCAI), Springer Lecture Notes in Computer Science (LNCS) 9349, 653–660. *shared first authors

Stapel, L.C., Lombardot, B., Broaddus, C., Kainmueller, D., Jug, F., E.W. Myers, and Vastenhouw, N. (2015). Automated detection and quantification of single RNAs at cellular resolution in zebrafish embryos. Development 143(3), 540-546.

Kainmueller, D., Jug, F., Rother, C., and Myers G. (2014). Active Graph Matching for Automatic Joint Segmentation and Annotation of C. Elegans. Proc. MICCAI, Springer LNCS 8673, 81-88.

Kainmueller, D., Lamecker, H., Weber, B., Heller, M., Hege, H.-C., and Zachow, S. (2013). Omnidirectional Displacements for Deformable Surfaces. Medical Image Analysis 17(4), 429-441.