Introduction A vast body of literature has been collected over the past decades attempting to extract blood flow rates from angiography.1 Most of these methods can be categorized as either transit-time2 or indicator-dilution3 techniques. These techniques have not been fully developed for regular clinical use, but the recent growth in endovascular treatment has rekindled interest in the use of angiography-based parametric maps4–6 for flow assessment in aneurysms, AVMs, and stenoses. In this preliminary retrospective study, we calculate mean arterial blood flow rates from angiography using transit-time methods and validate it against flow rates obtained from phase-contrast magnetic resonance (MR).
Methods Angiography (˜2 frames per second) and MR scans previously acquired in 6 patients were anonymously collected with IRB approval. Contrast concentration-time curves were recorded at proximal and distal locations on the right and left internal carotid arteries (figure 1). A lagnormal equation was fit to the curves and the transit-time between the two locations was calculated using three different methods (Mean Transit Time, Time to Peak, Cross Correlation). The volume of the arterial segment between the two locations was calculated. Then, arterial mean flow rate (ml/s)=arterial segment volume(ml)/transit-time(s). Mean flow rates in the same vessels obtained from phase-contrast MR (NOVA, VasSol Inc) were noted from patient records. Angiographic and NOVA flow rates were correlated to assess accuracy and statistical significance.
Results Statistically significant linear correlations were observed between angiographic and NOVA flow rates using the cross correlation (R^2=0.8) and time to peak (R^2=0.7) methods. When compared to MR flow, cross correlation had lower error (−9±23%) than time to peak (−24±30%) but this difference was not statistically significant (p=0.25).
Conclusions Determination of mean arterial blood flow from low frame rate patient angiography is possible and correlates well with NOVA values. The best method for calculating transit time may be cross correlation but a larger study is necessary to show this as well as to refine the accuracy of the methods.
Lieber BB, et al. Crit Rev Biomed Eng 2005;33(1):1–102.
Shpilfoygel SD, et al. Med Phys 2000;27(9):2008–23.
Hilal SK. Am J Roentgenol Radium Ther Nucl Med 1966 Apr;896–906.
Chien A, Viñuela F. J Neurointerv Surg 2013 Nov;5(Suppl 3):iii43–7.
Shakur SF, et al. J Neurointerv Surg 2018 Feb;10(2):156–161.
Ionita CN, et al. Proc SPIE Int Soc Opt Eng 2014 Mar 13;9038:90380L.
Disclosures M. Gross: None. H. Woo: 4; C; Vascular Simulations. 6; C; Cerenovus JnJ. D. Fiorella: 1; C; Penumbra, Microvention, Medtronic. 2; C; Penumbra, Microvention, Medtronic. 4; C; Vascular Simulations. 6; C; Cerenovus JnJ. C. Sadasivan: 4; C; Vascular Simulations.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.