Differentiation of low-grade non-Hodgkin's lymphomas using paraffin sections by image processing

In a previous study, we were able to demonstrate that the differentiation of low-grade non-Hodgkin's lymphomas (NHLs) using digital image analysis of touch imprints obtained from native tumor tissue is feasible. The availability of touch imprints in routine diagnostics, however, is restricted. Therefore, we extended our studies toward paraffin sections being used as routine material for histological diagnoses. To identify five types of NHL classified according to the Revised European American Lymphoma classification, paraffin sections (n=53) of NHL and 9 reactive lymphoid tissues (RLTs) were scanned with a color-video-based microscope system and analyzed by digital image processing. A reliable division between benign and neoplastic lymphoproliferations was achieved. We were able to identify 78% of RLTs as benign and 94% of NHLs as neoplastic. The average probability of correct identification into the six subgroups was 66%. In detail, 78% of RLTs, 50% of chronic lymphocytic lymphomas and MALT-type lymphomas, 72% of mantle cell lymphomas, and 67% of follicle center cell lymphoma were classified correctly. Although the method of subclassifying or identifying NHLs on the basis of a computer-mediated assay is still not usable in daily practice, we show that a reliable differentiation between reactive and neoplastic lymphoproliferative lesions can be achieved by analysis of paraffin sections with high-resolution image analysis and that it is possible to define nuclear structures by identifying subtypes of NHL.