Related Experiment Video
Updated: Aug 8, 2026
Label-Free Identification of Lymphocyte Subtypes Using Three-Dimensional Quantitative Phase Imaging and Machine Learning
Published on: November 19, 2018
Differentiation of low grade non-Hodgkin's lymphoma by digital image processing
Objective:
To identify different types of low grade B-cell non-Hodgkin's lymphoma (NHL) classified according to the Revised European American Lymphoma and Kiel classification systems by means of digital image processing.
Study Design:
Seventy-four touch imprints were scanned and analyzed. To compare common but intricate DNA stain with a routinely used panoptical dye, all lymphoma specimens had been stained by the Romanowsky-Giemsa method and 48 touch imprints redyed with Feulgen-Azure A. In both cases 30 features derived from size, and chromatin texture of each nucleus were evaluated.
Results:
Feulgen-stained touch imprints showed a 59% average probability of correct identification. The division of mantle cell lymphoma and chronic lymphocytic leukemia was difficult. In contrast, it was possible to distinguish all different types of lymphomas investigated if Romanowsky-Giemsa stain was used. Correct diagnoses were achieved for mantle cell lymphoma in 87.5%, follicle center cell lymphoma in 78%, chromic lymphocytic leukemia in 78%, immunocytoma in 75% and marginal zone B-cell lymphoma in 80%.
Conclusion:
The application of texture analysis is feasible in the classification of NHL.
