Differentiation of low-grade non-Hodgkin's lymphomas using paraffin sections by image processing
1Institute of Virology and Immunology, Laboratory for Image Processing, University of Würzburg, Germany. susanne@megabase.aecom.yu.edu
Cytometry
|May 14, 1998
Summary
Digital image analysis of paraffin sections can reliably differentiate benign lymphoid tissues from non-Hodgkin's lymphomas (NHLs). This method shows potential for identifying NHL subtypes, advancing diagnostic capabilities.
Area of Science:
- Digital pathology
- Hematopathology
- Computational imaging
Background:
- Previous studies demonstrated digital image analysis of touch imprints for non-Hodgkin's lymphoma (NHL) differentiation.
- Touch imprints are not routinely available in diagnostics, necessitating alternative methods.
Purpose of the Study:
- To evaluate the feasibility of digital image analysis of paraffin-embedded sections for NHL classification.
- To differentiate between reactive lymphoid tissues (RLTs) and various NHL subtypes using histological images.
Main Methods:
- Paraffin-embedded sections from 53 NHL cases and 9 RLTs were analyzed.
- A color-video-based microscope system and digital image processing were employed.
- The Revised European American Lymphoma (REAL) classification was used for NHL subtyping.
Main Results:
- The analysis achieved reliable differentiation between benign RLTs (78% correctly identified) and neoplastic NHLs (94% correctly identified).
- Average correct identification across six subgroups was 66%, with specific accuracies for RLTs (78%), chronic lymphocytic lymphomas (50%), MALT-type lymphomas (50%), mantle cell lymphomas (72%), and follicle center cell lymphoma (67%).
Conclusions:
- High-resolution image analysis of paraffin sections enables reliable differentiation of reactive versus neoplastic lymphoproliferative lesions.
- The study demonstrates the potential for defining nuclear structures to identify NHL subtypes, although not yet suitable for daily practice.
