Ultrasound Elastography Using EMD Analysis

Abstract

Ultrasoundelastographyisanon-invasivemethodwhichimagestheelasticityofsoft-tissues. Tomakeanimage, preandafterasmallcompression, ultrasoundradio frequency(RF)signalsareacquiredandthetimedelaysbetweenthemareestimated. Thefirstdifferentiationofdisplacementestimationsiscalledelastogram. Inthisstudy, wearegoingtomakeanelastogramusingtheprocessingmethodnamed empirical mode decomposition (EMD). EMDisananalytictechniquewhichdecomposesacomplicatedsignaltoacollectionofsimplesignalscalledintrinsicmodefunctions(IMFs). TheideaofpaperisusingtheseIMFsinsteadofprimaryRFsignals. Toimplementthealgorithmstwodifferentdatasetsselected. Thefirstonewasdatafromasandwichstructureofnormalandcookedtissue. The second datasetconsistedofaround 180 framesacquiredfromamalignantbreasttumor. Fordisplacementestimating, twodifferentmethods, cross-correlationandwavelettransform, wereusedtooandforevaluatingthequality, twoconventionalparameters, signal-to-noiseratio(SNR)andcontrast-to-noiseratio(CNR)calculatedforeachimage. ResultsshowthatinbothmethodsafterusingEMDthequalityimproves. InfirstdatasetandcrosscorrelationtechniqueCNRandSNRimproveabout 16dBand 9dBrespectively. Insamedatasetbyusingwavelettechnique, theparametersshow14dBand10dBimprovementrespectively.In second dataset(breasttumordata)CNRandSNRincrosscorrelationmethodimprove 18dBand 7dBandinwavelettechniqueimprove 17dBand 6dBrespectively.