Renogram and deconvolution parameters in diagnosis of renal artery stenosis

Abstract

Aim: Multivariatestatisticalmethods can be usedfor objective analysis. The emphasisisonanalysingrenalfunction parameters together, notone at atime.The aim is to identify curve parameters usefulinmakingpredictionsin kidneys with andwithout renalarterystenosis (RAS). Patients, methods: 68 patients with resistant hypertension weresubjected to captoprilrenographywith 99m Tc-DTPA. Variants of backgroundareas andbackgroundsubtraction methodswereemployed.Acorrection wasapplied forloss of renalparenchyma. Parameters fromtime-activitycurves andretention curvesfrom deconvolution werecalculated. Renal angiographyestablishedthe presenceorabsenceof RAS.Logistic regression analysis, usingage-and kidney size-adjustedmodels, wasperformedtoassess thecapability of renographyand deconvolution to differentiate between kidneys with andwithout RAS. Results: Discrimination between normal kidneys andRAS wasachievedby deconvolution andbyrenography. Deconvolution wasthe method of first rankwithasensitivity of 87% andaspecificity of 98%. For separation of RAS andkidneys with parenchymal insufficiency deconvolutionwas themethodoffirst rankwithasensitivity of 80 %and specificity of 89 %, whereasrenographyproducedpoor results. Conclusion: The best performance with 99m Tc-DTPAwas basedonnormalisedbackgroundsubtraction usingarectangular area between the kidneys. Deconvolution producedthe most favourableresults in the separation of kidneys with andwithout RAS.For separation of RAS andkidneys with parenchymal insufficiency conventional renographyproducedpoor results. Conceptually, the resultso falogisticr egression analysis of renalfunction parameters may raise possibilitiesinthe fieldofcomputer-aideddiagnosis.