Related Experiment Videos
Malignancy-associated changes in monocytes and lymphocytes in acute leukemias measured by high-resolution image
1Department of Hematology and Internal Medicine, University of Würzburg, Germany.
Insights
This study used image processing and statistical analysis to classify leukemia subtypes. The method successfully differentiated acute myeloid leukemias and identified distinct subtypes of monocytes and lymphocytes.
Area of Science:
- Hematology
- Medical imaging
- Computational biology
Background:
- Morphologic analysis is crucial for classifying lymphoid and myeloid leukemias.
- Existing methods for leukemia classification can be complemented by advanced techniques.
- Understanding cellular morphology is key to accurate diagnosis.
Purpose of the Study:
- To develop and validate an image processing method for leukemia classification.
- To investigate the utility of color and texture algorithms in analyzing blood cell morphology.
- To identify distinct mathematical subtypes of lymphocytes and monocytes in leukemic samples.
Main Methods:
- Analysis of approximately 23,000 Romanowsky-Giemsa-stained peripheral blood smear cells using a high-resolution color TV/microscope system.
- Application of color and texture algorithms for cell analysis.
- Multivariate statistical analysis to identify subtypes of lymphocytes and monocytes.
Main Results:
- Lymphocytes and monocytes showed leukemia-associated morphologic changes.
- Seven mathematical subtypes of lymphocytes and five of monocytes were identified.
- Acute myeloid leukemias (AML, AMMOL, AMOL) were differentiated by monocyte subtype distribution.
- Acute lymphoblastic leukemias (B-ALL, T-ALL) were discernible using lymphocyte subtypes.
- Distinction between acute myeloid conditions and viral infections (e.g., Epstein-Barr virus) was achieved.
Conclusions:
- Image processing significantly enhances the clinical diagnosis of acute leukemias.
- The study revealed that "normal" cell populations in malignant leukemias exhibit abnormalities.
- The identified subtypes provide a more refined classification of leukemic cells.
Abstract:
A number of methods are available for classifying lymphoid and myeloid leukemias in peripheral blood and bone marrow. However, in clinical diagnosis an initial and particularly important step is morphologic analysis. All the cells in this investigation were classified by two hematologic experts. In most cases, immunophenotyping and immunocytochemical analyses were performed. Routinely prepared Romanowsky-Giemsa-stained peripheral blood smears (approximately 23,000 cells) were scanned by a high-resolution color TV/microscope system and analyzed by color and texture algorithms. In addition to blast cells, lymphocytes and monocytes exhibited a leukemia-associated change in morphology. The calculated texture and color features were most significant for the subtyping performed by the statistical program. With multivariate statistical analysis, seven mathematical subtypes of lymphocytes and five of monocytes could be found over all the specimens. Acute myeloblastic leukemia (AML, M1-M2), acute myelomonocytic leukemia (AMMOL, M4) and acute monocytic leukemia (AMOL, M5) could be differentiated by their distributions of monocyte subtypes. However, this was impossible for the lymphocyte subtypes. Acute lymphoblastic leukemias (B-ALL and T-ALL) were discernible with the aid of lymphocyte subtypes and acute myeloid conditions from viral infections, such as with the Epstein-Barr virus. The method increased the relevance of image processing in clinical diagnosis of acute leukemias and showed that the "normal" cell populations were not really normal in malignant leukemias.