- Email: [email protected]
Quantitative 4D-Digital Subtraction Angiography to Assess Changes in Hepatic Arterial Flow during Transarterial Embolization: A Feasibility Study in a Swine Model
LABORATORY INVESTIGATION
Quantitative 4D-Digital Subtraction Angiography to Assess Changes in Hepatic Arterial Flow during Transarterial Embolization: A Feasibility Study in a Swine Model Ece Meram, MD, Colin Harari, BS, Gabe Shaughnessy, PhD, Martin Wagner, Dr Sc Hum, Chris L. Brace, PhD, Charles A. Mistretta, PhD, Michael A. Speidel, PhD, and Paul F. Laeseke, MD, PhD
ABSTRACT Purpose: To determine the feasibility of using time-resolved 3D-digital subtraction angiography (4D-DSA) for quantifying changes in hepatic arterial blood flow and velocity during transarterial embolization. Materials and Methods: Hepatic arteriography and selective transarterial embolization were performed in 4 female domestic swine (mean weight, 54 kg) using 100–300-μm microspheres. Conventional 2D and 4D-DSA were performed before, during, and after each embolization. From the 4D-DSA reconstructions, blood flow and velocity values were calculated for hepatic arterial branches using a pulsatility-based algorithm. 4D-DSA velocity values were compared to those measured using an intravascular Doppler wire with a linear regression analysis. Paired t-tests were used to compare data before and after embolization. Results: There was a weak-to-moderate but statistically significant correlation of flow velocities measured with 4D-DSA and the Doppler wire (r ¼ 0.35, n ¼ 39, P ¼ .012). For vessels with high pulsatility, the correlation was higher (r ¼ 0.64, n ¼ 11, P ¼ .034), and the relationship between 4D-DSA and the Doppler wire fit a linear model with a positive bias toward the Doppler wire (failed to reject at 95% confidence level, P ¼ .208). 4D-DSA performed after partial embolization showed a reduction in velocity in the embolized hepatic arteries compared to pre-embolization (mean, 3.96 ± 0.74 vs 11.8 2± 2.15 cm/s, P ¼ .006). Conclusion: Quantitative 4D-DSA can depict changes in hepatic arterial blood velocity during transarterial embolization in a swine model. Further work is needed to optimize 4D-DSA acquisitions and to investigate its applicability in humans.
ABBREVIATIONS 4D-DSA ¼ time-resolved 3D-digital subtraction angiography, APV ¼ averaged peak velocity, IPV ¼ instantaneous peak velocity, SACE ¼ subjective angiographic chemoembolization endpoint, SBR ¼ sideband ratio
Transarterial chemoembolization is the standard therapy for patients with intermediate-stage hepatocellular carcinoma (1). Efforts have been made to standardize the angiographic
endpoint for transarterial chemoembolization (eg, the subjective angiographic chemoembolization endpoint [SACE] rating scale) (2). However, currently available angiographic
From the Section of Interventional Radiology (E.M., P.F.L.), Department of Radiology, School of Medicine and Public Health (C.H.), Departments of Medical Physics (G.S., M.W., M.A.S.), and Radiology (C.L.B., C.A.M.), University of Wisconsin-Madison, 600 Highland Avenue, D4-352, Madison, WI 53792. Received May 7, 2018; final revision received January 10, 2019; accepted January 12, 2019. Address correspondence to E.M.; E-mail: [email protected]
from Siemens Healthineers, Germany. P.F.L. is a paid consultant for NeuWave Medical (Madison, WI), and Elucent Medical (Eden Praire, MN) and is a shareholder in Elucent Medical (Eden Praire, MN), Histosonics (Ann Arbor, MI), and McGinley Orthopedic Innovations (Casper, WY). None of the other authors have identified a conflict of interest.
M.W. is a co-owner of LiteRay Medical, LLC. (Minneapolis, MN). C.L.B. is a paid consultant for NeuWave Medical (Madison, WI), Ethicon (Somerville, NJ), Symple Surgical (Menlo Park, CA), and a shareholder in Elucent Medical (Eden Praire, MN), and Symple Surgical (Menlo Park, CA). M.A.S. receives grants
© SIR, 2019
From the SIR 2018 Annual Scientific Meeting.
J Vasc Interv Radiol 2019; ▪:1–7 https://doi.org/10.1016/j.jvir.2019.01.018
2 ▪ Quantitative 4D-DSA to Assess Changes in Hepatic Arterial Flow
endpoints for transarterial chemoembolization are subjective, variable, and with poor reproducibility among operators (2). A means to more accurately quantify the degree of stasis achieved during transarterial chemoembolization as a physiologic endpoint could increase not only the clinical efficacy of the procedure but also patient safety. Overembolization (ie, to complete stasis) can increase liver toxicity and accelerate liver failure (3). Complete stasis from a more proximal location has been shown to induce upregulation of angiogenic growth factors that can promote tumor growth (4–9). Therefore, an angiographic endpoint short of stasis (ie, substasis) is considered safer and more efficacious. In fact, embolization to substasis has been correlated with increased overall survival compared to embolization to complete stasis (10). Time-resolved 3D-digital subtraction angiography (4D-DSA) is a recently developed angiographic imaging modality capable of providing quantitative flow and velocity information within a 3D vascular volume of interest (11–15). To date, it has been almost exclusively investigated for brain imaging for obtaining detailed and time-resolved vascular information during interventional procedures (eg, for characterization of arteriovenous malformations) (16–20). Use of 4D-DSA within the thorax or abdomen has been limited, in part due to its susceptibility to motion artifacts. The purpose of this study was to determine the feasibility of using 4D-DSA to quantify changes in hepatic arterial flow and velocity during transarterial embolization in an in vivo porcine liver model.
MATERIALS AND METHODS All procedures were approved by the institutional research animal care and use committee and were compliant with regulatory guidelines. Transarterial embolization was performed in the livers of female domestic swine (n ¼ 4, mean weight, 54 kg). Subjects were sedated with an intramuscular administration of 7 mg/kg of tiletamine hydrochloride-zolazepam hydrochloride (Xyla-Ject; Phoenix Pharmaceutical, St. Joseph, Missouri), endotracheally intubated facilitated by 0.05 mg/kg atropine (Phoenix Pharmaceutical), and then underwent anesthesia induction and maintenance with 2% inhaled isoflurane (Halocarbon Laboratories, River Edge, New Jersey).
Transarterial Embolization Technique All transarterial embolizations were performed in an angiography suite (Artis Zee; Siemens, Munich, Germany). Arterial access was obtained via a femoral arterial puncture. A vascular sheath was placed, and a 4-Fr angled glide catheter was used to select the common hepatic artery. Common hepatic digital subtraction arteriograms were performed by injecting iohexol 300 mgI/mL (Omnipaque 300; GE Healthcare, Waukesha, Wisconsin) at 2.5 mL/s for a volume of 10 mL and a max pressure of 800 psi to delineate the gastroduodenal artery and hepatic arterial branches.
Meram et al ▪ JVIR
Each subject underwent transarterial embolization of 2 liver lobes, with the remaining 2 lobes serving as controls. The embolizations were performed using either the 4-Fr angled glide catheter or a 2.8-Fr microcatheter (Progreat; Terumo Medical Corporation, Tokyo, Japan) depending on whether the individual subject had a right or left hepatic artery of sufficient diameter and length to accommodate a 4-Fr catheter. Once catheter position was confirmed, conventional bland embolization was performed under fluoroscopic guidance to complete stasis (ie, lack of antegrade flow) with microspheres (100–300-μm Embosphere Microsphere; Merit Medical Systems, South Jordan, Utah). 4D-DSA and intravascular Doppler wire measurements were acquired before embolization, after partial embolization, and after complete stasis had been reached (see below).
4D-DSA Technique The 4D-DSA technique requires 2 separate gantry rotations. The first rotation was performed prior to contrast injection to acquire native mask images. The second rotation commenced simultaneously with the contrast injection to capture temporal contrast kinetics. The 4D-DSA datasets were acquired using a 12-second acquisition time over an angular range of 260 degrees. The native and contrastenhanced images were subtracted, and a conventional 3D reconstruction was performed to generate a constraint volume. Time frames for each projection image were generated by performing a constrained back-projection and normalization (16). 4D-DSA acquisitions were performed with the catheter positioned in the common hepatic artery. Iohexol 300 mgI/mL was injected at 2.5 mL/s for a volume of 27.5 mL and a max pressure of 800 psi. 4D-DSA images were acquired with respiration suspended on end-inspiration to minimize the detrimental effect of motion on the reconstruction. No paralytics were administered. 3D-DSA images were acquired, reconstructed, and reviewed during the procedure. 4D-DSA image reconstruction (Fig 1) was performed retrospectively after completion of the procedures using a prototype provided by the manufacturer (Siemens Healthineers, Forchheim, Germany). The 4D-DSA reconstruction was used to determine average blood velocities (Fig 2) based on phase shifts in the cardiac pulsatility of the time-attenuation curves as described in Shaughnessy et al (14). Pulsatility refers to the temporal variation in iodine concentration at each point along a vessel and is signal that flow is derived from. This is in contrast to the typical use of the term referring to oscillating changes in fluid movement.
Intravascular Doppler Technique A 0.014-inch Doppler guidewire (FloWire; Philips Volcano, San Diego, California) was used to measure blood velocity in the common hepatic artery, gastroduodenal artery, left hepatic artery, right hepatic artery, and lobar arteries. Accurate placement of the wire within the target branches was confirmed fluoroscopically. Once in place, instantaneous
Volume ▪ ▪ Number ▪ ▪ Month ▪ 2019
3
Figure 1. Hepatic 4D-DSA. Upper row shows the hepatic vasculature from anterior angle with 3 different time frames; lower row shows the final time frame from 5 different angles. Similar images can be obtained for any time frame or angle. A ¼ anterior, R ¼ right, RA ¼ right anterior, LA ¼ left anterior, L ¼ left.
Figure 2. Quantitative 4D-DSA during embolization of the left hepatic artery (LHA) showing the pre-embolization (left) and postembolization (right) local velocity estimations in the hepatic vasculature with a reduction in the LHA and cutoffs of left medial (LM) and left lateral (LL) branches (arrows). Distal branches of the LM and LL arteries were embolized to complete stasis and are not visualized post-embolization. Accuracy decreases for distal branches or vessels with lower pulsatility.
peak velocities (IPVs) and temporally averaged peak velocities (APVs) were recorded (ComboMap Pressure and Flow System, Philips Volcano) continuously and simultaneously for 5–10 seconds. IPV was recorded every 5 ms, and APV was calculated from the average of the IPV every 2 seconds. APVs were also re-averaged over 2 cardiac cycles of systole and diastole to minimize beat-to-beat variations and get a more stable Doppler signal. Velocity measurements were used as a surrogate parameter of blood flow for comparison with 4D-DSA to avoid introducing errors that
would arise from manual measurements of cross-sectional areas of blood vessels.
Statistical Analysis A total of 61 paired 4D-DSA and Doppler wire data were acquired in hepatic arterial branches. Multiple analyses were done with 2 types of clustering of the data points. The first clustering was based on the strength of the pulsatility of the data because the 4D-DSA flow technique relies on the pulsatility to detect phase shifts. The data were divided into
4 ▪ Quantitative 4D-DSA to Assess Changes in Hepatic Arterial Flow
high and low pulsatility based on the sideband ratio (SBR), a measure of pulsatile strength. Better flow correlation has been previously demonstrated for 4D-DSA reconstructions with SBR >3. For this animal study, SBR >3 and SBR >5 were used to divide the data as moderate-to-high pulsatility and high pulsatility, respectively (14,15). Further analysis was done on the data points with high pulsatility. The second clustering was done based on the treatment with embolization. All pre-embolization measurements (n ¼ 23) were grouped with data from non-embolized vessels acquired during mid-to-post-embolization to form an untreated group (n ¼ 39). The change in velocity pre-to-partial embolization was analyzed in the treated group only (n ¼ 22). Correlations between DSA and Doppler measurements were evaluated using Pearson correlation coefficients and were reported for each clustering. To determine the agreement of the Doppler wire and 4D-DSA flow measurements, a chi-squared (c2 ) analysis was performed with linear regression. The chi-squared analysis was first used to assess the validity of a linear regression model to see if it provides a good fit to the data. Then the acceptable regions of the model were evaluated for whether there was a significant difference between the best model fit and another model fit point. To determine if the change in blood velocities during transarterial embolization can be reliably distinguished using 4D-DSA, paired t-tests were used to compare the velocity measurements of treated arteries obtained during pre- and partial embolization. The level of significance was chosen as P < .05.
RESULTS Feasibility Analysis There was a weak-to-moderate but statistically significant correlation in peak blood flow velocities determined from 4D-DSA and Doppler wire measurements. Table shows the Pearson correlation coefficients of 4D-DSA calculations and Doppler wire measurements for all the different clusters. The untreated group (n ¼ 39) had a larger correlation coefficient (0.35, P ¼ .030) compared to the group containing all the data points (pre-, partial, and postembolization; n ¼ 61), which had a coefficient of 0.32 (P ¼ .012). When only moderate-to-high pulsatility data were considered, the sample size decreased (n ¼ 18), but the strength of the correlation increased to 0.56 (P ¼ .016). Similarly, data with only high pulsatility (n ¼ 11) had a coefficient of 0.64 (P ¼ .034). The strength of correlation between 2 methods was found to be higher for the data with higher pulsatility. For the data with high pulsatility, the linear model (Fig 3) between the 2 measurement methods could not be rejected at the 95% confidence level: c2(1, n¼11)¼13.3, m¼.55, b¼4.5 (P ¼ .208). Subsequently, the Dc2 analysis showed a confidence region within the linear regression parameters that excluded the no-linear-relation hypothesis (m ¼ 0)
Meram et al ▪ JVIR
Table. Pearson Correlation Coefficients and Corresponding P Values of the Experimental Data with Different Groupings Pearson coefficient (r)
P value
All Paired Data (n ¼ 61)
0.32
.012
Untreated Group (n ¼ 39)
0.35
.030
Untreated with SBR >3 (n ¼ 18)
0.56
.016
Untreated with SBR >5 (n ¼ 11)
0.64
.034
Group
Note–SBR reflects the strength of pulsatility, and higher values mean higher pulsatile strength. SBR ¼ sideband ratio.
Figure 3. Linear regression of 4D-DSA and Doppler wire for the data with high pulsatility; the slope shows the positive bias of Doppler wire compared to 4D-DSA measurements.
between these 2 measurement methods. Moreover, the equality between the Doppler wire and 4D-DSA method (m ¼ 1) was also excluded at the 95% confidence level. The Doppler wire measurements were also found to have a slight positive bias based on the regression analysis, and the R2 of the data with high pulsatility was 0.83 when the regression line passed through the origin (P < .001). The positive bias and lower slope shown in Figure 3 is reflective of the nature of the measurements with Doppler wire, which determines the temporally averaged maximum spatial velocity, whereas the 4D-DSA method measures the spatially and temporally averaged velocity.
Quantification of Change During Transarterial Embolization The effect of embolization on flow was analyzed in the treated vessels only (left hepatic, left lateral, and left medial). Figure 4 demonstrates 4D-DSA measurements performed pre- and partial embolization with the vesselby-vessel change for all the subjects. Paired t-test analysis
Volume ▪ ▪ Number ▪ ▪ Month ▪ 2019
5
Figure 4. Velocity estimates by quantitative 4D-DSA demonstrating the pig-by-pig and vessel-by-vessel change before and after partial embolization of left hepatic vasculature in 4 swine (the first animal did not have a left hepatic artery). P value of the paired t-test for preto-partial change was calculated as .006.
showed a reduction in blood flow velocity of the embolized left hepatic arteries with a mean velocity of 11.82 ± 2.15 versus 3.96 ± 0.74 cm/s for pre- and partial embolization, respectively (P ¼ .006). In Figure 5, the change in both 4DDSA and Doppler wire measurements taken during transarterial embolization is presented. 4D-DSA velocity measurements showed a higher decline in velocity compared to Doppler wire measurements taken during pre- and partial embolization (14.52 ± 2.53 vs 10.58 ± 1.60 cm/s, P ¼ .049). Complete stasis was achieved in all cases, at which point reliable measurements could not be made using 4D-DSA due to low signal and loss of pulsatility.
DISCUSSION This study aimed to assess the feasibility of using 4D-DSA to quantify hepatic arterial blood velocity and to investigate its ability to reflect changes in velocity during and after transarterial embolization. Moderate correlations in blood velocity were seen between 4D-DSA and an intravascular Doppler wire (r ¼ .32, P ¼ .012), and the correlations increased with increased signal strength, or pulsatility, of the 4D-DSA acquisitions (r ¼ 0.64 and P ¼ .034 for vessels with high pulsatility). A higher pulsatility is associated with larger temporal variation in image contrast, a higher signalto-noise ratio in the signal, and a more accurate velocity calculation (14,15). In addition, statistically significant reductions in blood velocity during transarterial embolization were demonstrated on 4D-DSA pre-to-partial embolization (mean velocity of 11.82 ± 2.15 vs 3.96 ± 0.74 cm/s for pre-
Figure 5. 4D-DSA and Doppler wire velocity (cm/s) estimates obtained during embolization of left hepatic vasculature in all treated vessels of 4 swine combined; dashed lines represent mean values. Pre- and partial mean velocity measured with Doppler wire and 4D-DSA were 14.52 ± 2.53 versus 10.58 ± 1.60 cm/s and 11.82 ± 2.15 versus 3.96 ± 0.74 cm/s, respectively. The range of values is related to different baseline velocities in different pigs and vessels; post-embolization estimates lacked good signal or high pulsatility with full occlusion of the vessels.
and partial embolization, respectively; P ¼ .006). Although velocity measurements could not be made after complete stasis was achieved secondary to low signal, non-quantitative angiographic methods can be used to determine complete stasis. Substasis (SACE II–III) is generally the desired endpoint during embolization (10), and further work is needed to optimize 4D-DSA for quantifying flow and velocity in low-flow states (ie, near substasis). More accurate and quantitative techniques for assessing the degree of stasis achieved during transarterial embolization as a
6 ▪ Quantitative 4D-DSA to Assess Changes in Hepatic Arterial Flow
physiologic endpoint may lead to higher success rates and safety profiles given that both outcomes and liver toxicity have been correlated with the degree of embolization (3,10). Quantitative flow imaging can be a means of achieving objective endpoints during interventional procedures, including liver embolization. 4D-DSA can give quantitative flow information in a 3D vasculature-of-interest. Because it relies on the subtraction of the mask images, 4D-DSA produces better results in stationary organs with minimal motion. As a relatively recent angiographic imaging modality, it has been mostly used in brain and peripheral arterial imaging rather than in the abdomen because of the motion artifacts caused by breathing. However, motion correction schemes are being developed that can enable the use of 4D-DSA in the imaging of the abdomen (21). With further development and optimization, quantitative 4D-DSA could be an integral part of transarterial procedures in the abdomen by facilitating angiographic endpoints more objectively and at the point of care. Given the relative paucity of data on the use of 4D-DSA in the abdomen, assessing the feasibility of the technique by correlating blood flow velocity with an intravascular Doppler probe was considered an important step. Despite an overall suboptimal correlation of 4D-DSA with the intravascular Doppler probe, the 2 techniques correlated reasonably well when the pulsatility on 4D-DSA was high. This suggests that additional optimization of acquisition parameters is necessary to maximize the pulsatility on 4D-DSA images. Alternatively, other flow quantification methods currently being developed for 4D-DSA may be evaluated for their feasibility in the abdomen (15). 4D-DSA depicted the changes in velocity during transarterial embolization, indicating that it may be used to quantify percent flow reduction during embolization and determine objective endpoints during procedures. This was a feasibility study that had several limitations. The study was performed in a non-tumor-bearing swine model without the use of chemotherapeutic agents. Therefore, the results may not directly translate to clinical patients with hepatocellular carcinoma undergoing transarterial chemoembolization. Although the total sample size provides sufficient power to the study, the power is lower for the subset analysis. Of 61 data points, 18 were more reliable in terms of their pulsatile strength, which is a vital component of the flow quantification technique used in this study. 4D-DSA is a developing modality, for which further optimization of acquisitions is needed. Further developments of 4D-DSA will involve motion correction schemes, advancement of quantification algorithms, as well as the optimization of injection parameters to prevent reflux and maintain pulsatility, which in turn may result in more reliable quantification of flow before and after treatments. An additional limitation is the inherent difference between the 4D-DSA technique and the Doppler wire, which was the method of validation
Meram et al ▪ JVIR
in this study. Although the velocity measurement from the Doppler wire gives an APV at a particular point or focus, 4D-DSA quantification produces a spatially and temporally averaged velocity. In conclusion, 4D-DSA can depict changes in blood flow velocity during transarterial embolization in an in vivo porcine liver model. Further work is needed to optimize 4DDSA acquisitions and to investigate its applicability for determining endpoints during clinical cases.
ACKNOWLEDGMENTS This work was supported by a research contract from Siemens Healthineers and by an award from the University of Wisconsin School of Medicine and Public Health and the Herman and Gwendolyn Shapiro Foundation.
REFERENCES 1. Bruix J, Sherman M, American Association for the Study of Liver Diseases. Management of hepatocellular carcinoma: an update. Hepatology 2011; 53:1020–1022. 2. Lewandowski RJ, Wang D, Gehl J, et al. A comparison of chemoembolization endpoints using angiographic versus transcatheter intraarterial perfusion/MR imaging monitoring. J Vasc Interv Radiol 2007; 18: 1249–1257. 3. Geschwind JFH, Ramsey DE, Cleffken B, et al. Transcatheter arterial chemoembolization of liver tumors: effects of embolization protocol on injectable volume of chemotherapy and subsequent arterial patency. Cardiovasc Intervent Radiol 2003; 26:111–117. 4. Rhee TK, Young JY, Larson AC, et al. Effect of transcatheter arterial embolization on levels of hypoxia-inducible factor-1alpha in rabbit VX2 liver tumors. J Vasc Interv Radiol 2007; 18:639–645. 5. Liu L, Ren ZG, Shen Y, et al. Influence of hepatic artery occlusion on tumor growth and metastatic potential in a human orthotopic hepatoma nude mouse model: relevance of epithelial-mesenchymal transition. Cancer Sci 2010; 101:120–128. 6. Sergio A, Cristofori C, Cardin R, et al. Transcatheter arterial chemoembolization (TACE) in hepatocellular carcinoma (HCC): the role of angiogenesis and invasiveness. Am J Gastroenterol 2008; 103: 914–921. 7. Delis SG, Bakoyiannis A, Biliatis I, Athanassiou K, Tassopoulos N, Dervenis C. Model for end-stage liver disease (MELD) score, as a prognostic factor for post-operative morbidity and mortality in cirrhotic patients, undergoing hepatectomy for hepatocellular carcinoma. HPB (Oxford) 2009; 11:351–357. 8. Li X, Feng GS, Zheng CS, Zhuo C-K, Liu X. Expression of plasma vascular endothelial growth factor in patients with hepatocellular carcinoma and effect of transcatheter arterial chemoembolization therapy on plasma vascular endothelial growth factor level. World J Gastroenterol 2004; 10: 2878–2882. 9. Shim JH, Park JW, Kim JH, et al. Association between increment of serum VEGF level and prognosis after transcatheter arterial chemoembolization in hepatocellular carcinoma patients. Cancer Sci 2008; 99: 2037–2044. 10. Jin B, Wang D, Lewandowski RJ, et al. Chemoembolization endpoints: effect on survival among patients with hepatocellular carcinoma. AJR Am J Roentgenol 2011; 196:919–928. 11. Grist TM, Mistretta CA, Strother CM, Turski PA. Time-resolved angiography: past, present, and future. J Magn Reson Imaging 2012; 36: 1273–1286. 12. Mistretta CA. Sub-Nyquist acquisition and constrained reconstruction in time resolved angiography. Med Phys 2011; 38: 2975–2985. €litz P, Struffert T, et al. 4D DSA for dynamic visualization of 13. Lang S, Go cerebral vasculature: a single-center experience in 26 cases. AJNR Am J Neuroradiol 2017; 38:1169–1176.
Volume ▪ ▪ Number ▪ ▪ Month ▪ 2019
14. Shaughnessy G, Schafer S, Speidel MA, Strother CM, Mistretta CA. Measuring blood velocity using 4D-DSA: a feasibility study. Med Phys 2018; 31:919. 15. Wu Y, Shaughnessy G, Hoffman CA, et al. Quantification of blood velocity with 4D digital subtraction angiography using the shifted least-squares method. AJNR Am J Neuroradiol 2018; 3:541. 16. Davis B, Royalty K, Kowarschik M, et al. 4D digital subtraction angiography: implementation and demonstration of feasibility. AJNR Am J Neuroradiol 2013; 34:1914–1921. 17. Sandoval-Garcia C, Royalty K, Yang P, et al. 4D DSA a new technique for arteriovenous malformation evaluation: a feasibility study. J Neurointerv Surg 2016; 8:300–304. 18. Lescher S, Gehrisch S, Klein S, Berkefeld J. Time-resolved 3D rotational angiography: display of detailed neurovascular anatomy in
7
patients with intracranial vascular malformations. J Neurointerv Surg 2017; 9:887–894. 19. Chen KK, Guo WY, Yang HC, et al. Application of time-resolved 3D digital subtraction angiography to plan cerebral arteriovenous malformation radiosurgery. AJNR Am J Neuroradiol 2017; 38:740–746. 20. Sandoval-Garcia C, Royalty K, Aagaard-Kienitz B, Schafer S, Yang P, Strother C. A comparison of 4D DSA with 2D and 3D DSA in the analysis of normal vascular structures in a canine model. AJNR Am J Neuroradiol 2015; 36:1959–1963. 21. Wagner M, Laeseke P, Harari C, Schafer S, Speidel M, Mistretta C. Feasibility of intra-acquisition motion correction for 4D DSA reconstruction for applications in the thorax and abdomen. Proc SPIE 10574, Medical Imaging 2018: Image Processing, 1057415 (March 2, 2018). https://doi. org/10.1117/12.2293812.
Quantitative 4D-Digital Subtraction Angiography to Assess Changes in Hepatic Arterial Flow during Transarterial Embolization: A Feasibility Study in a Swine Model Ece Meram, MD, Colin Harari, BS, Gabe Shaughnessy, PhD, Martin Wagner, Dr Sc Hum, Chris L. Brace, PhD, Charles A. Mistretta, PhD, Michael A. Speidel, PhD, and Paul F. Laeseke, MD, PhD
ABSTRACT Purpose: To determine the feasibility of using time-resolved 3D-digital subtraction angiography (4D-DSA) for quantifying changes in hepatic arterial blood flow and velocity during transarterial embolization. Materials and Methods: Hepatic arteriography and selective transarterial embolization were performed in 4 female domestic swine (mean weight, 54 kg) using 100–300-μm microspheres. Conventional 2D and 4D-DSA were performed before, during, and after each embolization. From the 4D-DSA reconstructions, blood flow and velocity values were calculated for hepatic arterial branches using a pulsatility-based algorithm. 4D-DSA velocity values were compared to those measured using an intravascular Doppler wire with a linear regression analysis. Paired t-tests were used to compare data before and after embolization. Results: There was a weak-to-moderate but statistically significant correlation of flow velocities measured with 4D-DSA and the Doppler wire (r ¼ 0.35, n ¼ 39, P ¼ .012). For vessels with high pulsatility, the correlation was higher (r ¼ 0.64, n ¼ 11, P ¼ .034), and the relationship between 4D-DSA and the Doppler wire fit a linear model with a positive bias toward the Doppler wire (failed to reject at 95% confidence level, P ¼ .208). 4D-DSA performed after partial embolization showed a reduction in velocity in the embolized hepatic arteries compared to pre-embolization (mean, 3.96 ± 0.74 vs 11.8 2± 2.15 cm/s, P ¼ .006). Conclusion: Quantitative 4D-DSA can depict changes in hepatic arterial blood velocity during transarterial embolization in a swine model. Further work is needed to optimize 4D-DSA acquisitions and to investigate its applicability in humans.
ABBREVIATIONS 4D-DSA ¼ time-resolved 3D-digital subtraction angiography, APV ¼ averaged peak velocity, IPV ¼ instantaneous peak velocity, SACE ¼ subjective angiographic chemoembolization endpoint, SBR ¼ sideband ratio
Transarterial chemoembolization is the standard therapy for patients with intermediate-stage hepatocellular carcinoma (1). Efforts have been made to standardize the angiographic
endpoint for transarterial chemoembolization (eg, the subjective angiographic chemoembolization endpoint [SACE] rating scale) (2). However, currently available angiographic
From the Section of Interventional Radiology (E.M., P.F.L.), Department of Radiology, School of Medicine and Public Health (C.H.), Departments of Medical Physics (G.S., M.W., M.A.S.), and Radiology (C.L.B., C.A.M.), University of Wisconsin-Madison, 600 Highland Avenue, D4-352, Madison, WI 53792. Received May 7, 2018; final revision received January 10, 2019; accepted January 12, 2019. Address correspondence to E.M.; E-mail: [email protected]
from Siemens Healthineers, Germany. P.F.L. is a paid consultant for NeuWave Medical (Madison, WI), and Elucent Medical (Eden Praire, MN) and is a shareholder in Elucent Medical (Eden Praire, MN), Histosonics (Ann Arbor, MI), and McGinley Orthopedic Innovations (Casper, WY). None of the other authors have identified a conflict of interest.
M.W. is a co-owner of LiteRay Medical, LLC. (Minneapolis, MN). C.L.B. is a paid consultant for NeuWave Medical (Madison, WI), Ethicon (Somerville, NJ), Symple Surgical (Menlo Park, CA), and a shareholder in Elucent Medical (Eden Praire, MN), and Symple Surgical (Menlo Park, CA). M.A.S. receives grants
© SIR, 2019
From the SIR 2018 Annual Scientific Meeting.
J Vasc Interv Radiol 2019; ▪:1–7 https://doi.org/10.1016/j.jvir.2019.01.018
2 ▪ Quantitative 4D-DSA to Assess Changes in Hepatic Arterial Flow
endpoints for transarterial chemoembolization are subjective, variable, and with poor reproducibility among operators (2). A means to more accurately quantify the degree of stasis achieved during transarterial chemoembolization as a physiologic endpoint could increase not only the clinical efficacy of the procedure but also patient safety. Overembolization (ie, to complete stasis) can increase liver toxicity and accelerate liver failure (3). Complete stasis from a more proximal location has been shown to induce upregulation of angiogenic growth factors that can promote tumor growth (4–9). Therefore, an angiographic endpoint short of stasis (ie, substasis) is considered safer and more efficacious. In fact, embolization to substasis has been correlated with increased overall survival compared to embolization to complete stasis (10). Time-resolved 3D-digital subtraction angiography (4D-DSA) is a recently developed angiographic imaging modality capable of providing quantitative flow and velocity information within a 3D vascular volume of interest (11–15). To date, it has been almost exclusively investigated for brain imaging for obtaining detailed and time-resolved vascular information during interventional procedures (eg, for characterization of arteriovenous malformations) (16–20). Use of 4D-DSA within the thorax or abdomen has been limited, in part due to its susceptibility to motion artifacts. The purpose of this study was to determine the feasibility of using 4D-DSA to quantify changes in hepatic arterial flow and velocity during transarterial embolization in an in vivo porcine liver model.
MATERIALS AND METHODS All procedures were approved by the institutional research animal care and use committee and were compliant with regulatory guidelines. Transarterial embolization was performed in the livers of female domestic swine (n ¼ 4, mean weight, 54 kg). Subjects were sedated with an intramuscular administration of 7 mg/kg of tiletamine hydrochloride-zolazepam hydrochloride (Xyla-Ject; Phoenix Pharmaceutical, St. Joseph, Missouri), endotracheally intubated facilitated by 0.05 mg/kg atropine (Phoenix Pharmaceutical), and then underwent anesthesia induction and maintenance with 2% inhaled isoflurane (Halocarbon Laboratories, River Edge, New Jersey).
Transarterial Embolization Technique All transarterial embolizations were performed in an angiography suite (Artis Zee; Siemens, Munich, Germany). Arterial access was obtained via a femoral arterial puncture. A vascular sheath was placed, and a 4-Fr angled glide catheter was used to select the common hepatic artery. Common hepatic digital subtraction arteriograms were performed by injecting iohexol 300 mgI/mL (Omnipaque 300; GE Healthcare, Waukesha, Wisconsin) at 2.5 mL/s for a volume of 10 mL and a max pressure of 800 psi to delineate the gastroduodenal artery and hepatic arterial branches.
Meram et al ▪ JVIR
Each subject underwent transarterial embolization of 2 liver lobes, with the remaining 2 lobes serving as controls. The embolizations were performed using either the 4-Fr angled glide catheter or a 2.8-Fr microcatheter (Progreat; Terumo Medical Corporation, Tokyo, Japan) depending on whether the individual subject had a right or left hepatic artery of sufficient diameter and length to accommodate a 4-Fr catheter. Once catheter position was confirmed, conventional bland embolization was performed under fluoroscopic guidance to complete stasis (ie, lack of antegrade flow) with microspheres (100–300-μm Embosphere Microsphere; Merit Medical Systems, South Jordan, Utah). 4D-DSA and intravascular Doppler wire measurements were acquired before embolization, after partial embolization, and after complete stasis had been reached (see below).
4D-DSA Technique The 4D-DSA technique requires 2 separate gantry rotations. The first rotation was performed prior to contrast injection to acquire native mask images. The second rotation commenced simultaneously with the contrast injection to capture temporal contrast kinetics. The 4D-DSA datasets were acquired using a 12-second acquisition time over an angular range of 260 degrees. The native and contrastenhanced images were subtracted, and a conventional 3D reconstruction was performed to generate a constraint volume. Time frames for each projection image were generated by performing a constrained back-projection and normalization (16). 4D-DSA acquisitions were performed with the catheter positioned in the common hepatic artery. Iohexol 300 mgI/mL was injected at 2.5 mL/s for a volume of 27.5 mL and a max pressure of 800 psi. 4D-DSA images were acquired with respiration suspended on end-inspiration to minimize the detrimental effect of motion on the reconstruction. No paralytics were administered. 3D-DSA images were acquired, reconstructed, and reviewed during the procedure. 4D-DSA image reconstruction (Fig 1) was performed retrospectively after completion of the procedures using a prototype provided by the manufacturer (Siemens Healthineers, Forchheim, Germany). The 4D-DSA reconstruction was used to determine average blood velocities (Fig 2) based on phase shifts in the cardiac pulsatility of the time-attenuation curves as described in Shaughnessy et al (14). Pulsatility refers to the temporal variation in iodine concentration at each point along a vessel and is signal that flow is derived from. This is in contrast to the typical use of the term referring to oscillating changes in fluid movement.
Intravascular Doppler Technique A 0.014-inch Doppler guidewire (FloWire; Philips Volcano, San Diego, California) was used to measure blood velocity in the common hepatic artery, gastroduodenal artery, left hepatic artery, right hepatic artery, and lobar arteries. Accurate placement of the wire within the target branches was confirmed fluoroscopically. Once in place, instantaneous
Volume ▪ ▪ Number ▪ ▪ Month ▪ 2019
3
Figure 1. Hepatic 4D-DSA. Upper row shows the hepatic vasculature from anterior angle with 3 different time frames; lower row shows the final time frame from 5 different angles. Similar images can be obtained for any time frame or angle. A ¼ anterior, R ¼ right, RA ¼ right anterior, LA ¼ left anterior, L ¼ left.
Figure 2. Quantitative 4D-DSA during embolization of the left hepatic artery (LHA) showing the pre-embolization (left) and postembolization (right) local velocity estimations in the hepatic vasculature with a reduction in the LHA and cutoffs of left medial (LM) and left lateral (LL) branches (arrows). Distal branches of the LM and LL arteries were embolized to complete stasis and are not visualized post-embolization. Accuracy decreases for distal branches or vessels with lower pulsatility.
peak velocities (IPVs) and temporally averaged peak velocities (APVs) were recorded (ComboMap Pressure and Flow System, Philips Volcano) continuously and simultaneously for 5–10 seconds. IPV was recorded every 5 ms, and APV was calculated from the average of the IPV every 2 seconds. APVs were also re-averaged over 2 cardiac cycles of systole and diastole to minimize beat-to-beat variations and get a more stable Doppler signal. Velocity measurements were used as a surrogate parameter of blood flow for comparison with 4D-DSA to avoid introducing errors that
would arise from manual measurements of cross-sectional areas of blood vessels.
Statistical Analysis A total of 61 paired 4D-DSA and Doppler wire data were acquired in hepatic arterial branches. Multiple analyses were done with 2 types of clustering of the data points. The first clustering was based on the strength of the pulsatility of the data because the 4D-DSA flow technique relies on the pulsatility to detect phase shifts. The data were divided into
4 ▪ Quantitative 4D-DSA to Assess Changes in Hepatic Arterial Flow
high and low pulsatility based on the sideband ratio (SBR), a measure of pulsatile strength. Better flow correlation has been previously demonstrated for 4D-DSA reconstructions with SBR >3. For this animal study, SBR >3 and SBR >5 were used to divide the data as moderate-to-high pulsatility and high pulsatility, respectively (14,15). Further analysis was done on the data points with high pulsatility. The second clustering was done based on the treatment with embolization. All pre-embolization measurements (n ¼ 23) were grouped with data from non-embolized vessels acquired during mid-to-post-embolization to form an untreated group (n ¼ 39). The change in velocity pre-to-partial embolization was analyzed in the treated group only (n ¼ 22). Correlations between DSA and Doppler measurements were evaluated using Pearson correlation coefficients and were reported for each clustering. To determine the agreement of the Doppler wire and 4D-DSA flow measurements, a chi-squared (c2 ) analysis was performed with linear regression. The chi-squared analysis was first used to assess the validity of a linear regression model to see if it provides a good fit to the data. Then the acceptable regions of the model were evaluated for whether there was a significant difference between the best model fit and another model fit point. To determine if the change in blood velocities during transarterial embolization can be reliably distinguished using 4D-DSA, paired t-tests were used to compare the velocity measurements of treated arteries obtained during pre- and partial embolization. The level of significance was chosen as P < .05.
RESULTS Feasibility Analysis There was a weak-to-moderate but statistically significant correlation in peak blood flow velocities determined from 4D-DSA and Doppler wire measurements. Table shows the Pearson correlation coefficients of 4D-DSA calculations and Doppler wire measurements for all the different clusters. The untreated group (n ¼ 39) had a larger correlation coefficient (0.35, P ¼ .030) compared to the group containing all the data points (pre-, partial, and postembolization; n ¼ 61), which had a coefficient of 0.32 (P ¼ .012). When only moderate-to-high pulsatility data were considered, the sample size decreased (n ¼ 18), but the strength of the correlation increased to 0.56 (P ¼ .016). Similarly, data with only high pulsatility (n ¼ 11) had a coefficient of 0.64 (P ¼ .034). The strength of correlation between 2 methods was found to be higher for the data with higher pulsatility. For the data with high pulsatility, the linear model (Fig 3) between the 2 measurement methods could not be rejected at the 95% confidence level: c2(1, n¼11)¼13.3, m¼.55, b¼4.5 (P ¼ .208). Subsequently, the Dc2 analysis showed a confidence region within the linear regression parameters that excluded the no-linear-relation hypothesis (m ¼ 0)
Meram et al ▪ JVIR
Table. Pearson Correlation Coefficients and Corresponding P Values of the Experimental Data with Different Groupings Pearson coefficient (r)
P value
All Paired Data (n ¼ 61)
0.32
.012
Untreated Group (n ¼ 39)
0.35
.030
Untreated with SBR >3 (n ¼ 18)
0.56
.016
Untreated with SBR >5 (n ¼ 11)
0.64
.034
Group
Note–SBR reflects the strength of pulsatility, and higher values mean higher pulsatile strength. SBR ¼ sideband ratio.
Figure 3. Linear regression of 4D-DSA and Doppler wire for the data with high pulsatility; the slope shows the positive bias of Doppler wire compared to 4D-DSA measurements.
between these 2 measurement methods. Moreover, the equality between the Doppler wire and 4D-DSA method (m ¼ 1) was also excluded at the 95% confidence level. The Doppler wire measurements were also found to have a slight positive bias based on the regression analysis, and the R2 of the data with high pulsatility was 0.83 when the regression line passed through the origin (P < .001). The positive bias and lower slope shown in Figure 3 is reflective of the nature of the measurements with Doppler wire, which determines the temporally averaged maximum spatial velocity, whereas the 4D-DSA method measures the spatially and temporally averaged velocity.
Quantification of Change During Transarterial Embolization The effect of embolization on flow was analyzed in the treated vessels only (left hepatic, left lateral, and left medial). Figure 4 demonstrates 4D-DSA measurements performed pre- and partial embolization with the vesselby-vessel change for all the subjects. Paired t-test analysis
Volume ▪ ▪ Number ▪ ▪ Month ▪ 2019
5
Figure 4. Velocity estimates by quantitative 4D-DSA demonstrating the pig-by-pig and vessel-by-vessel change before and after partial embolization of left hepatic vasculature in 4 swine (the first animal did not have a left hepatic artery). P value of the paired t-test for preto-partial change was calculated as .006.
showed a reduction in blood flow velocity of the embolized left hepatic arteries with a mean velocity of 11.82 ± 2.15 versus 3.96 ± 0.74 cm/s for pre- and partial embolization, respectively (P ¼ .006). In Figure 5, the change in both 4DDSA and Doppler wire measurements taken during transarterial embolization is presented. 4D-DSA velocity measurements showed a higher decline in velocity compared to Doppler wire measurements taken during pre- and partial embolization (14.52 ± 2.53 vs 10.58 ± 1.60 cm/s, P ¼ .049). Complete stasis was achieved in all cases, at which point reliable measurements could not be made using 4D-DSA due to low signal and loss of pulsatility.
DISCUSSION This study aimed to assess the feasibility of using 4D-DSA to quantify hepatic arterial blood velocity and to investigate its ability to reflect changes in velocity during and after transarterial embolization. Moderate correlations in blood velocity were seen between 4D-DSA and an intravascular Doppler wire (r ¼ .32, P ¼ .012), and the correlations increased with increased signal strength, or pulsatility, of the 4D-DSA acquisitions (r ¼ 0.64 and P ¼ .034 for vessels with high pulsatility). A higher pulsatility is associated with larger temporal variation in image contrast, a higher signalto-noise ratio in the signal, and a more accurate velocity calculation (14,15). In addition, statistically significant reductions in blood velocity during transarterial embolization were demonstrated on 4D-DSA pre-to-partial embolization (mean velocity of 11.82 ± 2.15 vs 3.96 ± 0.74 cm/s for pre-
Figure 5. 4D-DSA and Doppler wire velocity (cm/s) estimates obtained during embolization of left hepatic vasculature in all treated vessels of 4 swine combined; dashed lines represent mean values. Pre- and partial mean velocity measured with Doppler wire and 4D-DSA were 14.52 ± 2.53 versus 10.58 ± 1.60 cm/s and 11.82 ± 2.15 versus 3.96 ± 0.74 cm/s, respectively. The range of values is related to different baseline velocities in different pigs and vessels; post-embolization estimates lacked good signal or high pulsatility with full occlusion of the vessels.
and partial embolization, respectively; P ¼ .006). Although velocity measurements could not be made after complete stasis was achieved secondary to low signal, non-quantitative angiographic methods can be used to determine complete stasis. Substasis (SACE II–III) is generally the desired endpoint during embolization (10), and further work is needed to optimize 4D-DSA for quantifying flow and velocity in low-flow states (ie, near substasis). More accurate and quantitative techniques for assessing the degree of stasis achieved during transarterial embolization as a
6 ▪ Quantitative 4D-DSA to Assess Changes in Hepatic Arterial Flow
physiologic endpoint may lead to higher success rates and safety profiles given that both outcomes and liver toxicity have been correlated with the degree of embolization (3,10). Quantitative flow imaging can be a means of achieving objective endpoints during interventional procedures, including liver embolization. 4D-DSA can give quantitative flow information in a 3D vasculature-of-interest. Because it relies on the subtraction of the mask images, 4D-DSA produces better results in stationary organs with minimal motion. As a relatively recent angiographic imaging modality, it has been mostly used in brain and peripheral arterial imaging rather than in the abdomen because of the motion artifacts caused by breathing. However, motion correction schemes are being developed that can enable the use of 4D-DSA in the imaging of the abdomen (21). With further development and optimization, quantitative 4D-DSA could be an integral part of transarterial procedures in the abdomen by facilitating angiographic endpoints more objectively and at the point of care. Given the relative paucity of data on the use of 4D-DSA in the abdomen, assessing the feasibility of the technique by correlating blood flow velocity with an intravascular Doppler probe was considered an important step. Despite an overall suboptimal correlation of 4D-DSA with the intravascular Doppler probe, the 2 techniques correlated reasonably well when the pulsatility on 4D-DSA was high. This suggests that additional optimization of acquisition parameters is necessary to maximize the pulsatility on 4D-DSA images. Alternatively, other flow quantification methods currently being developed for 4D-DSA may be evaluated for their feasibility in the abdomen (15). 4D-DSA depicted the changes in velocity during transarterial embolization, indicating that it may be used to quantify percent flow reduction during embolization and determine objective endpoints during procedures. This was a feasibility study that had several limitations. The study was performed in a non-tumor-bearing swine model without the use of chemotherapeutic agents. Therefore, the results may not directly translate to clinical patients with hepatocellular carcinoma undergoing transarterial chemoembolization. Although the total sample size provides sufficient power to the study, the power is lower for the subset analysis. Of 61 data points, 18 were more reliable in terms of their pulsatile strength, which is a vital component of the flow quantification technique used in this study. 4D-DSA is a developing modality, for which further optimization of acquisitions is needed. Further developments of 4D-DSA will involve motion correction schemes, advancement of quantification algorithms, as well as the optimization of injection parameters to prevent reflux and maintain pulsatility, which in turn may result in more reliable quantification of flow before and after treatments. An additional limitation is the inherent difference between the 4D-DSA technique and the Doppler wire, which was the method of validation
Meram et al ▪ JVIR
in this study. Although the velocity measurement from the Doppler wire gives an APV at a particular point or focus, 4D-DSA quantification produces a spatially and temporally averaged velocity. In conclusion, 4D-DSA can depict changes in blood flow velocity during transarterial embolization in an in vivo porcine liver model. Further work is needed to optimize 4DDSA acquisitions and to investigate its applicability for determining endpoints during clinical cases.
ACKNOWLEDGMENTS This work was supported by a research contract from Siemens Healthineers and by an award from the University of Wisconsin School of Medicine and Public Health and the Herman and Gwendolyn Shapiro Foundation.
REFERENCES 1. Bruix J, Sherman M, American Association for the Study of Liver Diseases. Management of hepatocellular carcinoma: an update. Hepatology 2011; 53:1020–1022. 2. Lewandowski RJ, Wang D, Gehl J, et al. A comparison of chemoembolization endpoints using angiographic versus transcatheter intraarterial perfusion/MR imaging monitoring. J Vasc Interv Radiol 2007; 18: 1249–1257. 3. Geschwind JFH, Ramsey DE, Cleffken B, et al. Transcatheter arterial chemoembolization of liver tumors: effects of embolization protocol on injectable volume of chemotherapy and subsequent arterial patency. Cardiovasc Intervent Radiol 2003; 26:111–117. 4. Rhee TK, Young JY, Larson AC, et al. Effect of transcatheter arterial embolization on levels of hypoxia-inducible factor-1alpha in rabbit VX2 liver tumors. J Vasc Interv Radiol 2007; 18:639–645. 5. Liu L, Ren ZG, Shen Y, et al. Influence of hepatic artery occlusion on tumor growth and metastatic potential in a human orthotopic hepatoma nude mouse model: relevance of epithelial-mesenchymal transition. Cancer Sci 2010; 101:120–128. 6. Sergio A, Cristofori C, Cardin R, et al. Transcatheter arterial chemoembolization (TACE) in hepatocellular carcinoma (HCC): the role of angiogenesis and invasiveness. Am J Gastroenterol 2008; 103: 914–921. 7. Delis SG, Bakoyiannis A, Biliatis I, Athanassiou K, Tassopoulos N, Dervenis C. Model for end-stage liver disease (MELD) score, as a prognostic factor for post-operative morbidity and mortality in cirrhotic patients, undergoing hepatectomy for hepatocellular carcinoma. HPB (Oxford) 2009; 11:351–357. 8. Li X, Feng GS, Zheng CS, Zhuo C-K, Liu X. Expression of plasma vascular endothelial growth factor in patients with hepatocellular carcinoma and effect of transcatheter arterial chemoembolization therapy on plasma vascular endothelial growth factor level. World J Gastroenterol 2004; 10: 2878–2882. 9. Shim JH, Park JW, Kim JH, et al. Association between increment of serum VEGF level and prognosis after transcatheter arterial chemoembolization in hepatocellular carcinoma patients. Cancer Sci 2008; 99: 2037–2044. 10. Jin B, Wang D, Lewandowski RJ, et al. Chemoembolization endpoints: effect on survival among patients with hepatocellular carcinoma. AJR Am J Roentgenol 2011; 196:919–928. 11. Grist TM, Mistretta CA, Strother CM, Turski PA. Time-resolved angiography: past, present, and future. J Magn Reson Imaging 2012; 36: 1273–1286. 12. Mistretta CA. Sub-Nyquist acquisition and constrained reconstruction in time resolved angiography. Med Phys 2011; 38: 2975–2985. €litz P, Struffert T, et al. 4D DSA for dynamic visualization of 13. Lang S, Go cerebral vasculature: a single-center experience in 26 cases. AJNR Am J Neuroradiol 2017; 38:1169–1176.
Volume ▪ ▪ Number ▪ ▪ Month ▪ 2019
14. Shaughnessy G, Schafer S, Speidel MA, Strother CM, Mistretta CA. Measuring blood velocity using 4D-DSA: a feasibility study. Med Phys 2018; 31:919. 15. Wu Y, Shaughnessy G, Hoffman CA, et al. Quantification of blood velocity with 4D digital subtraction angiography using the shifted least-squares method. AJNR Am J Neuroradiol 2018; 3:541. 16. Davis B, Royalty K, Kowarschik M, et al. 4D digital subtraction angiography: implementation and demonstration of feasibility. AJNR Am J Neuroradiol 2013; 34:1914–1921. 17. Sandoval-Garcia C, Royalty K, Yang P, et al. 4D DSA a new technique for arteriovenous malformation evaluation: a feasibility study. J Neurointerv Surg 2016; 8:300–304. 18. Lescher S, Gehrisch S, Klein S, Berkefeld J. Time-resolved 3D rotational angiography: display of detailed neurovascular anatomy in
7
patients with intracranial vascular malformations. J Neurointerv Surg 2017; 9:887–894. 19. Chen KK, Guo WY, Yang HC, et al. Application of time-resolved 3D digital subtraction angiography to plan cerebral arteriovenous malformation radiosurgery. AJNR Am J Neuroradiol 2017; 38:740–746. 20. Sandoval-Garcia C, Royalty K, Aagaard-Kienitz B, Schafer S, Yang P, Strother C. A comparison of 4D DSA with 2D and 3D DSA in the analysis of normal vascular structures in a canine model. AJNR Am J Neuroradiol 2015; 36:1959–1963. 21. Wagner M, Laeseke P, Harari C, Schafer S, Speidel M, Mistretta C. Feasibility of intra-acquisition motion correction for 4D DSA reconstruction for applications in the thorax and abdomen. Proc SPIE 10574, Medical Imaging 2018: Image Processing, 1057415 (March 2, 2018). https://doi. org/10.1117/12.2293812.