[1589] Conversion of Traditional Immunohistochemistry into Virtual Multicolor Stains Using Feature Extraction and Pixel-Conversion Algorithms

Anne-Sophie K Meyer, Peter Moller, Jochen K Lennerz. Ulm University, Ulm, Germany

Background: Colocalization analysis is too time-consuming and cost-intensive for routine diagnostics. Therefore, immunophenotyping is -for the vast majority of clinical assessments- performed via mental combination of single stains. Here we present a set of pixel-conversion algorithms that generate a synthetic image for marker covisualization and/or subsequent quantification.
Design: Single immunohistochemical (IHC) stains were digitized and subsequently either re-stained and re-digitized or merged with separately digitized subsequent sections. The applied algorithms are composed of the following components: a user-defined feature extraction, a color conversion and an overlay component. To allow downstream analysis a customized link between ImageJ, Photoshop CS3 (Adobe Systems), and ImageScope (Aperio) or OlyVIA (Olympus) was generated using AutoIT (version 3.2.12.).
Results: The algorithms emulate multicolor staining without additional use of tissue or equipment and preserve the immunohistochemical stain for traditional assessment. The extracted and converted elements enable covisualization on digital merges and/or subsequent quantification of co-localized elements using established methods. The algorithms solve the common problem of species identity and/or cross-reactivity of primary antibodies because they allow co-visualization of markers without compromising staining specificity. When compared to investments in immunofluorescence microscopy equipment and validation of traditional multi-color stains, the effort to learn and perform the conversion algorithms is negligible.
Conclusions: The presented imaging tools emulate the principal advantages of multi-color fluorescence microscopy as well as multi-color immunohistochemistry without compromising traditional assessment. Thus, the algorithms expand the versatility of one of the most important molecular tools in diagnostic pathology (i.e. IHC) and allow visual integration that exemplifies the potential of digital pathology.
Category: Informatics

Monday, March 4, 2013 11:15 AM

Proffered Papers: Section H2, Monday Morning


Close Window