More information
Portrait Robert Haase

Life-sciences rely heavily on multi-dimensional, high spatio-temporal, multi-channel microscopical imaging of many samples under various conditions. Major technical challenges encompass manifold aspects of large-scale image data mining: economic image acquisition, well-organized image data storage, efficient image processing, reliable quantitative Bio-image analysis, reproducible post-processing workflows, and sustainably sharing experience in all these aspects. Analogously, similar challenges exist on the simulation side where scientists develop physical models mimicking biological observations. While the amount of available research image data is steadily increasing, also more and more new algorithms and methods emerge. The research field “Bio-image Analysis” is moving fast. New technologies like graphics processing units (GPUs), artificial intelligence and collaborative cloud-based image analysis software are staggeringly changing the way how we interact with image data. We approach these major challenges by building solid bridges between communities, especially image data scientists who develop new methods and experimental scientists who are eager to work with top-notch scientific Bio-image analysis software.

Our goal is to enable both scientific communities, algorithm developers and wet-lab experimentalists, to concentrate on their core-missions. Therefore, we automate post-acquisition image processing in the cloud, on local high-performance-computing infrastructure and on personal laptops. We combine state-of-the-art smart microscopy approaches, universally accessible large-scale image data storage, distributed computing, GPU-acceleration and machine learning algorithms in software libraries with handy user-interfaces that make state-of-the-art Bio-image analysis sustainable and accessible to everyone.

Our approach is based on three main concepts

  • Interdisciplinary collaboration. We are convinced that only diverse teams of scientists such as physicists, biologists, chemists, microscopists, computer scientists and image data scientists can solve the puzzles behind the Physics of Life together. We are strong team-players, and we grow by crossing borders between disciplines regularly. We customize image data analysis workflows together with our collaborators to get the most out of experimental data. We also optimize experimental setups to improve image quality with respect to best-possible scientific reasoning. Constant exchange, rapid prototyping, pair-programming and regular hackathons are our methods of choice for achieving scientific and technological goals in teams.
  • Knowledge exchange. By sharing all our knowledge and skills in image data science and data mining, we support the Physics of Life community to answer scientific questions efficiently. Therefore, we offer lectures, courses and ad-hoc consulting in image analysis and data science theory, and state-of-the-art software such as ImageJ/Fiji, QuPath, Icy, Scikit-Image, Python, NumPy, SciPy, Napari, OpenCL, Java, Maven, Omero, Knime, CellProfiler and more.
  • Method exploration, exploitation and tool development. We are well connected in the international developer community of image data analysis algorithms, methods and platforms. We support collaborators in open-source and open-science projects in order to gain early-access to new upcoming top-notch image data analysis tools. We assemble these tools together with our local collaborators into image data analysis workflows and tools optimally suited for untangling the Physics of Life.
Robert Haase Research: Figure Figure: In developmental biology, complex spatio-temporal patterns appear, such as during the onset of gastrulation in the red flour beetle (Tribolium castaneum). Workflows for processing of four-dimensional imaging data typically include operations such as (from left to right): 1) Background subtraction, 2) blob-detection, 3) Voronoi-tessellation, 4) neighbor mesh generation, 5) neighbor centroid distance quantification. Raw data source: Daniela Vorkel, Myers lab, MPI-CBG / CSBD

Future Projects and Goals

In developmental biology, studying tissue formation is common task which is increasingly targeted using reproducible, quantitative methods. When analyzing large amounts of imaging data, techniques for denoising, segmentation and feature extraction pose challenges as processing three-dimensional image data is computationally expensive. We face these challenges using modern computing technology such as graphics processing units (GPUs) [see Selected Publications, 1] and high-performance computing (HPC). As tissue formation is a highly complex process, new approaches to quantification are needed to differentiate stages of development in health and disease, as well as to compare tissue development and tissue regeneration. We combine graph theory in space (tissue neighborhood graphs) and time (lineage trees), with supervised and unsupervised machine learning (ML). Goal of our efforts are developing new user-friendly tools combining GPUs, HPC and ML so that biologists can have deeper insights into tissue formation of embryos and organoids. Therefore, we are establishing computational frameworks for processing grids of cells analogously to how image data scientists process grids of pixels for decades [2]. In close collaboration with biological and medical research groups on the biomedical research campus in Dresden-Johannstadt and international collaborators we ensure that our tools serve the research community broadly, and allow answering a wide range of scientific questions.

Methodological and Technical Expertise

Our group focuses on bio-image analysis and image data science for gaining insights from microscopy imaging data. We routinely work with live-imaging fluorescence microscopy data and support scientists at the institute and on campus with expertise in extracting quantitative measurements. We also contribute to post-processing of measurement data, for example to identify relationships between these measurements, phenotypes, genotypes, and internal and external forces. We are experts in classical image processing and make use of machine learning [3] and deep Learning when necessary. While we are used to a wide range of image analysis tools and methods [4], our technical core expertise resides in image data analysis using Python programming and Jupyter notebooks. In our projects we extend the image visualization and analysis software Napari. We practice open science, interdisciplinary knowledge exchange and are embedded in the Network of European BioImage Analysts (NEUBIAS) to organize our projects, symposia and teaching activities in the international context [5].


since 2020
Bio-image Analysis Technology Development group leader, Cluster of Excellence Physics of Life, TU Dresden

Postdoctoral researcher, Computational Microscopist, Myers lab, Max Planck Institute for Molecular Cell Biology and Genetics, Dresden

Bio-image Analyst, Scientific Software Engineer, Scientific Computing Facility, Max Planck Institute for Molecular Cell Biology and Genetics, Dresden

Doctoral degree in Medical Image Processing (Dr. rer. medic.), Medical Faculty Carl Gustav Carus, TU Dresden

Research assistant, Image Analyst, High-Precision Radiotherapy Group (Dr. R. Perrin / Dr. C. Richter), OncoRay, Medical Faculty Carl Gustav Carus, TU Dresden

Research assistant, PhD student, Biological and Molecular Imaging Group (Prof. N. Abolmaali), OncoRay, Medical Faculty Carl Gustav Carus, TU Dresden

Dipl.-Inf. (FH) Computer Science, HTW Dresden

Selected Publications

[1] Robert Haase, Loic Alain Royer, Peter Steinbach, Deborah Schmidt, Alexandr Dibrov, Uwe Schmidt, Martin Weigert, Nicola Maghelli, Pavel Tomancak, Florian Jug, Eugene W Myers
CLIJ: GPU-accelerated image processing for everyone
Nat Methods 17, 5–6 (2020)

[2] Robert Haase
Image Processing Filters for Grids of Cells Analogous to Filters Processing Grids of Pixels
Frontiers in Computer Science (2021)

[3] Matthias Arzt, Joran Deschamps, Christopher Schmied, Tobias Pietzsch, Deborah Schmidt, Pavel Tomancak, Robert Haase, Florian Jug
LABKIT: Labeling and Segmentation Toolkit for Big Image Data
Frontiers in Computer Science (2022)

[4] Robert Haase, Elnaz Fazeli, David Legland, Michael Doube, Siân Culley, Ilya Belevich, Eija Jokitalo, Martin Schorb, Anna Klemm, Christian Tischer
A Hitchhiker‘s guide through the bio-image analysis software universe
FEBS Letters (2022)

[5] Gabriel G Martins, Fabrice P Cordelières, Julien Colombelli, Rocco D'Antuono, Ofra Golani, Romain Guiet, Robert Haase, Anna H Klemm, Marion Louveaux, Perrine Paul-Gilloteaux, Jean-Yves Tinevez, Kota Miura
Highlights from the 2016-2020 NEUBIAS training schools for Bioimage Analysts: a success story and key asset for analysts and life scientists
F1000Research, 10(334) (2021)