Related Experiment Videos
Reproducibility and efficiency testing of two sampling methods for image analysis in cytology
F Altenstrasser1, A R Weger, J Lindholm
1Department of Pathology, University of Innsbruck, Austria.
Summary
Comparing consecutive and random selection methods for nuclear DNA analysis, this study found both offer high reproducibility. Random selection requires fewer nuclei measurements for accurate results in nuclear area and DNA histogram entropy assessments.
Area of Science:
- Pathology
- Computational Biology
- Oncology
Background:
- Quantitative nuclear DNA analyses and nuclear morphometry are vital in pathology.
- Variability in results among observers using identical methods is a common issue.
- Differences in sampling methods may contribute to this observer variability.
Purpose of the Study:
- To evaluate the reproducibility and efficiency of two image analysis selection methods: consecutive and random.
- To compare the number of nuclei required for reproducible results using each method.
Main Methods:
- An interactive imaging analysis system was used to measure 10 cancer and hyperplasia cases twice.
- Two sampling methods (consecutive and random systematic selection) were applied.
- Median nuclear area and DNA histogram entropy were calculated parameters.
Main Results:
- Consecutive selection required 150 nuclei for reproducible nuclear area and 110 for entropy.
- Random systematic selection required 80 nuclei for nuclear area and 100 for entropy.
- Both methods demonstrated high reproducibility and good correlation.
Conclusions:
- Both consecutive and random selection methods yield reproducible results in nuclear DNA analysis.
- Random systematic selection is more efficient, requiring fewer measurements.
- The concept of "plateau values" aids in determining optimal measurement numbers for efficiency.