Visualizing and Clustering Sub-Cellular Localization Images
Author Information
Author(s): Hamilton Nicholas A, Teasdale Rohan D
Primary Institution: The University of Queensland
Hypothesis
Can images of sub-cellular localization be arranged in two or three dimensions to allow easy comparison and differentiation?
Conclusion
The study demonstrates a novel methodology for visualizing and clustering high throughput sub-cellular localization images, allowing for effective comparison and discrimination.
Supporting Evidence
- The methodology was tested on 502 images with 10 known sub-cellular localizations.
- 3-neighbor classification accuracy was 83.2% in 3D and 68.9% in 2D.
- Images that are statistically similar are spatially close in the visualization.
Takeaway
This study shows a new way to look at lots of cell images so scientists can easily see which ones are similar and which are different.
Methodology
The iCluster methodology generates image statistics and maps them into 2 or 3 dimensions for visualization.
Limitations
The accuracy of clustering decreases when mapping to lower dimensions, and some classes of images may spatially overlap.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website