Automated microscopic image analysis of immunofluorescence-stained targets on tissue sections is challenged by autofluorescent erythrocytes, which interfere with target segmentation and quantification. To resolve this, authors of a recent Cytometry Part A article developed ARETE, a system for in silico recognition and removal of erythrocytes. To avoid blocking a fluorescence channel, they chose to recognize erythrocytes in the transmission channel only, which also allows adherence to any existing staining protocol. Rather than implement a classical image analysis system by hand, they utilized established machine learning techniques that are used, for example, to recognize smiling faces in state-of-the-art digital cameras. In effect, biologists just had to mark up samples of erythrocytes as well as areas without erythrocytes and the image analysis system was created automatically. By this approach biologists' implicit knowledge - what they simply know but cannot explain - can also be captured. The authors believe that such systems are the future of image analysis.
Read more about it in the article:
Automated REcognition of Tissue-associated Erythrocytes (ARETE)—a new tool in tissue cytometry
Andreas Heindl, Alexander K. Seewald, Theresia Thalhammer, Giovanna Bises, Martin Schepelmann, Hana Uhrova, Sabine Dekan, Ildiko Mesteri, Radu Rogojanu and Isabella Ellinger
Article first published online: 11 FEB 2013 | DOI: 10.1002/cyto.a.22258 and in print in the April 2013 issue of Cytometry Part A.
Link to article: http://onlinelibrary.wiley.com/doi/10.1002/cyto.a.22258/full