The application of matched filtering to contrast material dose reduction in digital subtraction angiography

Temporal filtering methods were applied to a 50% reduction of contrast material dose in intravenous digital subtraction angiography (DSA). The performance of matched filtering was reviewed, indicating a 50% reduction of noise through the integration of an entire (greater than 10 sec) DSA image sequence spanning the contrast material bolus arrival and washout. Used in conjunction with 50% contrast material dose reduction, matched filtering could provide final iodine signal-to-noise ratios (SNR) comparable to those of conventional DSA. Half-contrast-material-dose (20 ml) filtered DSA runs were compared with full-contrast-material-dose (40 ml) unfiltered runs for 5 carotid, 2 cerebral, and 4 peripheral studies. In all but 2 studies, the half-contrast-material-dose runs were judged to be diagnostic. In all 4 peripheral studies and in some of the 7 carotid and cerebral studies the 20-ml results were judged to be equivalent to the 40-ml results. It is concluded that the half-contrast-material-dose temporally filtered approach is appropriate for peripheral runoff studies. Matched filtering was experimentally compared with simple integrated remasking. If the identical image sequence was used for each, the two had virtually identical performance for integration times less than 10 sec, consistent with theoretical predictions. Two other applications of matched filtering, SNR enhancement and x-ray exposure reduction, are discussed schematically.