Improving Microarray Image Analysis with Rank Filters
Author Information
Author(s): Bengtsson Anders, Bengtsson Henrik
Primary Institution: Lund University
Hypothesis
Can rank and quantile filters provide better background estimation in microarray image analysis compared to traditional methods?
Conclusion
Advanced rank filters perform comparably to the best region-based methods, while morphological opening shows significant bias.
Supporting Evidence
- The rank filter showed a mean bias between 2 and -2, while morphological opening had a mean bias between -47 and -248.
- The variability of the morphological opening estimate was three times higher than for the rank filter.
- GenePix Pro's region-based estimate had variability more than ten times higher than the rank-filter estimate.
Takeaway
This study looks at how to better estimate the background noise in images from gene experiments, showing that some new methods work better than older ones.
Methodology
The study compares a new rank-filter algorithm with existing methods using microarray images from four slides.
Potential Biases
Morphological opening methods showed substantial spatial-dependent bias.
Limitations
The study assumes that background intensity is uniform across spots, which may not always be true.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website