Assessing the reliability of pulsatility in four-dimensional digital subtraction angiography time concentration curves

Objective. The time concentration curves (TCCs) from a Four-Dimensional Digital Subtraction Angiography (4D DSA) reconstruction contain cardiac derived pulsatility waveforms. These can be used to quantify blood flow. This work\u27s purpose was to evaluate the ability to visualize and characterize these waveforms in 4D DSA acquisitions done using a variety of injection protocols. Approach. Forty-two 4D DSA acquisitions were obtained in 6 canines under a variety of injection protocols. Heart rate was recorded during the acquisitions. Two blinded reviewers qualitatively scored the pulsatility waveforms in TCCs obtained from 147 ROIs. A Fourier space based method for automated pulsatility detection was developed to determine the correlation of the Fundamental Frequency (FF) in the TCCs with heart rate. The sensitivity of pulsatility detection using this quantitative method was compared to the observer\u27s visual detection. Main results. Recorded heart rate and Fourier space calculated heart rates were highly similar (107.0 versus 106.7 p = 0.42). Reviewer scores of 0 (n = 15, pulsatility poorly defined and random), 0.5 (n = 21, pulsatility ambiguous) or 1 (n = 111, pulsatility clear and regular) correlated with the signal strength of the FF. The quantitative algorithm allowed pulsatility detection in 10 of 36 acquisitions where visual assessment was inconclusive. There was no association between injections done with a variety of injection parameters and either the recognition of pulsatility or its link with heart rate. Significance. Cardiac derived pulsatility in the TCCs of a 4D DSA reconstruction may be detected visually as well as automatically using a Fourier space method. FF correlates highly with heart rate. In TCCs where pulsatility was difficult or impossible to recognize visually, this method improved the ability to differentiate the FF from noise. In this study there was no correlation between injection protocols and the quality of the pulsatility waveforms