Mathematical morphology operators for reading radioactivity DNA array images

We have recently proposed an automatic non-supervised set of algorithms for an accurate spot data extraction from Cy3/Cy5 cDNA microarrays using morphological operators. On the basis of an image signal intensity model, we explore in this study the application of advanced morphological filters for the spot determination and the spot signal quantification on radio-labelled cDNA arrays. The idea behind the approach here developed is to use the connected operators like area and contrast opening, adaptively applied according to the image features (size of spots, back.-round estimate, etc.), for defining a zone of the positive spots where computing effectively the spot intensity without the overlapping and back-round contributions. A gridding algorithm is only necessary to index the detected spots to be able to identify the corresponding biological probe.