Image Analysis


In the Image Analysis research area, we develop and apply methods for computational processing and analysis of digital microscopy images, especially for solving problems in biology and medicine. A wide range of biomedical research projects rely on microscopy data which require specifically designed image analysis workflows to support the identification of complex biological processes. In close collaboration with our partners from research and industry, we develop algorithms and software that analyze microscopy images, for example fluorescence microscopy images of cells or other kinds of organic material. Our focus in algorithm development is mainly on methods of signal processing, machine learning, and computer vision for information extraction and interpretation. Besides filtering and denoising for pre-processing and feature extraction methods, we use machine learning approaches such as deep learning for object detection, segmentation, and object tracking. In addition to workflows and algorithms, we implement image analysis software (e.g. Spotty) for biological research to automatically detect, track, and analyze cells and particles to support fundamental research and drug development. For image analysis, we build on tools and frameworks such as MATLAB, OpenCV, and TensorFlow.



Selected Publications
Selected Projects
Screening 2.0
In Austria there are currently around 573.000 to 645.000 people affected by Diabetes mellitus. This is about 8 to 9 percent of all Austrians. About 143.000 to 215.000 people do not know about their diabetes yet (2 to 3%). They have a higher risk for complications and long-term effects. Because of demographic developments and the rising prevalence of lifestyle related risk factors (overweight/obesity, physical inactivity, unfavourable nutrition, smoking, etc.) the number of diabetics and diabetes related diseases will rise further.
To counter these developments the consortium of Screening 2.0 aims at the development of a comprehensive concept of non-invasive diagnostic tools for individual, exhaustive, and user-friendly diabetes screening (diabetes as a first step) in combination with e-health applications. It involves a service innovation combined with a product innovation:
* Development of a printed qualitative diabetes-screening strip used in a trend analysis for early detection of diabetes
* Development of a comprehensive communication concept
Test strips and screening methods already exist in various ways. However, these are not printed but mostly dipped or coated. Through the development of the printing process in combination with the development of the suitable print varnish production cost and used resources should be reduced significantly. Furthermore the strip must be designed in a way to be easily included in a postal mailing which is the prerequisite for the service innovation.
To implement early detection successfully into the existing health system an innovative service must be developed. Current communication- and supply channels need to be considered while it needs to be carried out nationwide. Screening enables an addition to current supply and communication. Therefore the communication cycle can be closed for the first time during the screening process and not – as before – after the analysis of prescribed medication (therapy) through health insurances. The figure depicts how the screening can extend the existing supply and communication chain.
Head
DI(FH) Viktoria Dorfer MSc
Researchers
DI(FH) Viktoria Dorfer MSc
Susanne Schaller MMSc
Prof. (FH) DI Dr. Stephan Winkler
Duration
2015 - present
Research Areas
ImageAnalysis
DataScience
Research Institutions
Research Center Hagenberg
University of Applied Sciences, Upper Austria, Hagenberg Campus
Research focus
Software technology and application