Both digital subtraction and recursive filtering schemes have been employed successfully for intravenous and intraarterial arteriography. Either processing method results in an image(s), S, which is a linear combination of discrete images Ij acquired during the flow of iodinated contrast material, i.e., S = 1; ??Nj=0kjlj, j = 0 where kj are the weighting coefficients for the N+1 samples. It is shown that for a given set of images {Ij} there exists a set of weighting coefficients {kj} which maximizes the iodine signal to noise ratio and simultaneously removes stationary background anatomy. The kj are related to the contrast dilution curve measured over an artery of interest, kj = s[j]-s??, where {s[j]} is the set of measured image variations due to the flow of contrast material, and s??, is the mean value of the s[j]. This choice of kj defines a matched filter. Compared to subtraction angiography, matched filtering is 4-6 times more dose efficient.