Fluorescence Microscopy / Life Science Research

Problem: manually counting cells that express multiple fluorescent markers is tedious work and prone to errors and intra/inter observer variance            

Solution: automatically analyze scans and create Excel file

Current Status:

• Successfully employed in practice
• Unterer B, et al. IFN--response mediator GBP-1 represses human cell proliferation by inhibiting the Hippo signaling transcription factor TEAD. Biochem J. 2018 Aug 17
• Prototypical standalone software available
• C/C++ API in development
  

Solution:

• Large modular toolbox with ~50 image processing building blocks available. We combine and tune these building blocks in order to create powerful image processing algorithms that can be stored as presets/templates
  
• The solution is free of AI and so does not require a large annotated ground truth database
• Optionally, an image processing pipeline can be automatically optimized by providing ground truth annotations for one or a few sample images

Problem: manually identifying and counting cells in close proximity to each other is tedious work and prone to errors and intra/inter observer variance                                                        

Solution: automatically analyze scans and create Excel file                                                 

Current Status:

• Successfully employed in practice
• C. Gross, et al. The Tax-Inducible Actin-Bundling Protein Fascin Is Crucial for Release and Cell-to-Cell Transmission of Human T-Cell Leukemia Virus Type 1 (HTLV-1). PLoS Pathog. 2016 Oct 24;12(10):e1005916
• C/C++ API in development

Solution:

• Large modular toolbox with ~50 image processing building blocks available. We combine and tune these building blocks in order to create powerful image processing algorithms that can be stored as presets/templates
• The solution is free of AI and so does not require a large annotated ground truth database
• Optionally, an image processing pipeline can be automatically optimized by providing ground truth annotations for one or a few sample images

Problem: manually taking measurements on a cell’s shape, size and area or its nucleus-cytoplasm ratio is tedious work and prone to errors and intra/inter observer variance

Solution: automatically analyze scans and create Excel file with various statistics on all cells

Current Status:

• Successfully employed in practice
  Wiesmann V, et al. Automated high-throughput analysis of B cell spreading on immobilized antibodies with whole slide imaging. Current Directions in Biomedical Engineering. 1.1 (2015): 224-227
• C/C++ API in development
  

Solution:

• Large modular toolbox with ~50 image processing building blocks available. We combine and tune these building blocks in order to create powerful image processing algorithms that can be stored as presets/templates
  
• The solution is free of AI and so does not require a large annotated ground truth database
• Optionally, an image processing pipeline can be automatically optimized by providing ground truth annotations for one or a few sample images

Goal: Automatic detection and quantification of microorganisms on FISH samples

Partners: joint effort with the Biofilm Center Charité / MoKi Analytics GmbH, Chili GmbH, HB Technologies                                                          

Current Status:

• Segmentation Tissue vs Background
• Segmentation of defects (holes) in tissue
• Spot detection of in 40x scans – spots are likely bacteria
• Find Hotspots
  

Database:

• 30 annotated FISH whole slide scans (40x WSIs and single 100x field of views) of natural and synthetic heart valves
  

Solution:

• Detect potential bacteria in 40x scans and create 100x scans at hotspots

More information about iSOLID
iSOLID is supported by the program „Photonics in the LifeSciences“ (13N14917) by the German Ministry for Education and Research (BMBF).