Related Experiment Videos
Automated identification of diploid reference cells in cervical smears using image analysis
Jeroen A W M van der Laak1, Albertus G Siebers, Vincent M J I Cuijpers
1Department of Pathology, University Medical Center, Nijmegen, The Netherlands. J.vanderlaak@pathol.azn.nl
Cytometry
|April 5, 2002
Summary
Automated image analysis accurately identifies diploid epithelial cells in Papanicolaou (PAP) smears. This method for DNA ploidy analysis shows reliable results for premalignant cervical lesion evaluation.
Area of Science:
- Cytopathology
- Biomedical Image Analysis
- Quantitative Cytology
Background:
- DNA ploidy histograms from image analysis offer insights into premalignant cervical lesions.
- Accurate nuclear selection is crucial for reliable DNA measurement in cervical cytology.
- Manual selection of reference cells by cytotechnologists is the traditional method.
Purpose of the Study:
- To develop and describe a fully automated method for identifying diploid epithelial reference cell nuclei.
- To enable automated DNA ploidy analysis in Feulgen-stained Papanicolaou (PAP) smears.
Main Methods:
- A decision tree classifier with feature thresholds and discriminant functions was implemented.
- The system distinguishes between debris, inflammatory cells, and diploid/nondiploid epithelial cells.
- Automated diploid cell identification was validated against manual selection.
Main Results:
- The automated procedure successfully identified diploid epithelial reference cells.
- Automated DNA ploidy measurements showed an average deviation of 2.4% compared to manual measurements.
Conclusions:
- The automated method for identifying diploid epithelial cells in PAP smears is effective.
- This automated approach demonstrates potential utility in routine cytometric practice.
- The findings suggest a reliable alternative to manual cell selection for DNA ploidy analysis.