Author/Authors :
Hsaio, Ming-Chi Mechanical Engineering - National Taiwan University of Science and Technology, Taiwan , Chang, Peter I. Mechanical Engineering - National Taiwan University of Science and Technology, Taiwan
Abstract :
Thisresearchpresentedanaccurateandefficientcontourlengthestimationmethoddevelopedfor DNA digitalcurvesacquiredfromAtomicForceMicroscopy(AFM)images. ThisautomationmethodiscalibratedagainstdifferentAFMresolutionsandidealtobeextendedtoalldifferentkindsofbiopolymersamples,encompassingalldifferentsamplestiffnesses. Themethodologyconsidersthedigitalcurvelocalgeometricrelationship,asthesedigitalshapesegmentsandpixelconnectionsrepresenttheactualmorphologyofthebiopolymersampleasitisbeingimagedfromtheAFMscanning. Inordertoincorporatethetruelocalgeometryrelationshipthatisembeddedinthecontinuousformoftheoriginalsample,oneneedstofindthisgeometrycounterpartinthedigitizedimage. Thiscounterpartisrealizedbytakingtheskeletonbackboneofthesamplecontourandbyusingthesedigitizedpixels’connectionrelationshiptofinditslocalshaperepresentation.Inthisresearch,oneusesthe8-connectFreemanChainCode(CC) todescribethedirectionalconnectionbetween DNA imagepixels,inordertoaccountforthelocalshapesoffourconnectedpixels.Theresultisanovelshapenumber(SN)systemderivedfromCC,whichisafullyautomatedalgorithmthatcanbeappliedtoDNAsamplesofanylengthforaccurateestimation,withefficientcomputationalcost. Thisshape-wiseconsiderationisweightedtomodifythelocallengthwithgreatprecision,accountingforallthedifferentmorphologiesofthebiopolymersample, andresultedwithaccuratelengthestimation,astheerrorfallsbelow0.07%, anorderofmagnitudeimprovementcomparedtopreviousfindings.
Keywords :
Resolution-Free Accurate DNA Contour Length Estimation , Atomic Force Microscopy Images , TheAtomicForceMicroscopy(AFM)
