摘要
AsthetopologicalpropertiesofeachspotinDNAmicroarrayimagesmayvaryfromoneanother,weemployedgranulometriestounderstandtheshape-sizecontentcontributedduetoasignificantintensityvaluewithinaspot.Analysiswasperformedonthemicroarrayimagethatconsistedof240spotsbyusingconceptsfrommathematicalmorphology.Inordertofindoutindicesforeachspotandtofurtherclassifythem,weadoptedmorphologicalmultiscaleopenings,whichprovidedmicroarraysatmultiplescales.Successiveopenedmicroarraysweresubtractedtoidentifytheprotrusionsthatweresmallerthanthesizeofstructuringelement.Spot-wisedetails,intermsofprobabilityoftheseobservedprotrusions,werecomputedbyplacingaregularlyspacedgridonmicroarraysuchthateachspotwascenteredineachgrid.Basedontheprobabilityofsizedistributionfunctionsoftheseprotrusionsisolatedateachlevel,weestimatedthemeansizeandtextureindexforeachspot.Withthesecharacteristics,weclassifiedthespotsinamicroarrayimageintobrightanddullcategoriesthroughpatternspectrumandshape-sizecomplexitymeasures.Thesesegregatedspotscanbecomparedwiththoseofhybridizationlevels.
出版日期
2004年04月14日（中国期刊网平台首次上网日期，不代表论文的发表时间）
