Morphometric analysis for early detection of changes in cellular structure in a toxicological experiment.
This paper presents a method for semi-automatic analysis of morphometry in image cytometry studies. We examine morphological changes in neuroblastoma cells caused by a toxin. Nuclei are automatically recognized from images using a constrained seeded region growing algorithm. The shape of the nuclei is quantified by fitting ellipses to the identified regions. Features such as centroids, elongation, orientation and size are extracted from this fitting process. Under certain assumptions, it is demonstrated that the log elongation measurement has an asymptotically normal distribution whose variance is dependent only on object size. This allows weighted linear models to be fitted to elongation measurements. Hypothesis tests show that over time, there is a significant elongation due to application of the toxin. The other variables are not significant, but there is a significant 'image' effect. The scientific implications of these findings are considered.
Duke Scholars
Published In
DOI
ISSN
Publication Date
Volume
Issue
Start / End Page
Location
Related Subject Headings
- Subtraction Technique
- Statistics & Probability
- Pattern Recognition, Automated
- Neurotoxins
- Neuroblastoma
- Linear Models
- Laser Scanning Cytometry
- Image Enhancement
- Humans
- Cell Nucleus Size
Citation
Published In
DOI
ISSN
Publication Date
Volume
Issue
Start / End Page
Location
Related Subject Headings
- Subtraction Technique
- Statistics & Probability
- Pattern Recognition, Automated
- Neurotoxins
- Neuroblastoma
- Linear Models
- Laser Scanning Cytometry
- Image Enhancement
- Humans
- Cell Nucleus Size