4D DSA for Dynamic Visualization of Cerebral Vasculature: A Single-Center Experience in 26 Cases

Discussion

Currently, 2D DSA (in combination with 3D DSA) is regarded as the criterion standard in the diagnostic work-up of cerebrovascular disease.^2,13 However, existing 2D and 3D techniques are limited in different ways. Although 2D DSA offers high spatial and temporal resolution, it is limited by the C-arm system in special projections (eg, with extreme angulation). Despite 3D DSA offering no limitation concerning the choice of projections, this method is limited because of lack of temporal information and consecutive vascular overlap. This limitation is especially true in the case of AVMs and AVFs. Therefore, 4D DSA might be helpful in such tasks as evaluation of the AVM angioarchitecture and details of dAVF filling and drainage because it combines advantages of both techniques.

In our case series, 4D DSA, as a novel imaging technique, was performed successfully in all patients and was suitable for imaging of vascular malformations such as AVMs, cerebral aneurysms, and dAVFs. We describe a possible postprocessing method that offers conventional and time-resolved 3D DSA images with high resolution.

In line with recent literature, qualitative parameters obtained from 4D DSA also demonstrated excellent agreement with 2D DSA in our series regarding image quality and arteriovenous visualization of the angioarchitecture of the pathology. Sandoval-Garcia et al⁸ and Lescher et al⁹ observed a superior visualization of intranidal structures and the fistulous point of a dAVF, respectively, with 4D DSA because time-resolved 3D imaging enables easy differentiation of arterial and venous structures. We fully agree with the hypothesis of Sandoval-Garcia et al that 4D DSA might simplify the understanding of complex vascular malformations and provides valuable support in risk assessment and therapeutic strategies.⁸ Especially after publication of A Randomized Trial of Unruptured Brain Arteriovenous Malformations (ARUBA),¹⁴ an individual assessment of the risk factors regarding hemorrhage has become even more important for generating individual therapeutic strategies.

Previous publications have exclusively analyzed 4D DSA in regard to qualitative parameters.^6⇓–8 To evaluate the accuracy of 4D DSA compared with the current criterion standard (2D DSA), we measured, on the one hand, the maximum diameter of the injection vessel and, on the other hand, specific architectural features of vascular pathologies (eg, nidus size in AVMs, aneurysm size, diameter of the main feeder of a dAVF). Our quantitative evaluation of 4D DSA shows excellent agreement with 2D DSA and demonstrates the reliability of this new technique. However, due to the wide range of possible windowing and missing standardization of the postprocessing, over- and undersizing of vascular structures might be a potential source of error. Moreover, due to the maximum 12-second scan time, 4D DSA could be limited in the visualization of very slow vascular malformations, (eg, low-flow dAVFs, with a bolus passage of >12 seconds). Therefore, 4D DSA might not be considered a substitution for the conventional 2D DSA series, but it has the potential to replace 3D DSA in complex vascular malformations and can help decrease the amount of 2D projections in many cases. Hence, a significant reduction of radiation dose and contrast medium, especially in patients with complex vascular pathologies,⁶ is possible. Consequently, this new technique might offer obvious advantages, in particular for young patients or those with renal dysfunction, and might lead to a reduction of the procedural time, which is seen as a protective factor in periprocedural complications.¹⁵

In contrast to Lescher et al,⁹ who observed a limited visualization of small meningeal dAVF feeders, our readers did not feel restricted in identifying even the smallest vessels (eg, lenticulostriate or thalamoperforating arteries). Especially, the use of MIP images can be helpful in differentiating the smallest vessels. Only in cases of heavy steal phenomena (n = 2) or movement artifacts (n = 1), was visualization of the smallest vessels limited in our series. In our experience, these different findings might refer to different injection protocols (8 seconds in our series versus 7 seconds in the series of Lescher et al) because the duration of arterial contrast could have a significant influence on the IQ and visualization of vessels. Because prolonged injection potentially generates, as also observed by other authors,^8,9 overlapping angiographic phases, further investigation of contrast application to establish optimized injection protocols seems obligatory.

The effective patient dose was measured as 1.2 mSv, which is slightly higher than that of a conventional 3D rotational angiography (0.9 mSv, 5s DSA) and a standard 2D angiographic series (1 mSv).¹¹ Similar to 3D rotational angiography, the effective patient dose can be additionally decreased by using collimation, which has not been performed during the dose measurements with the anthropomorphic phantom.¹¹

Several noninvasive approaches have tried to generate time-resolved 3D images of the cerebral vasculature. Several authors applied time-resolved MR angiography (4D MRA) with spin-labeling or contrast-enhanced sequences on typical cerebrovascular pathologies and demonstrated high agreement with DSA in hemodynamic information¹⁶ in the assessment of AVMs^17⇓–19 and fistulas.^20⇓–22 4D MRA seems to be a promising alternative for DSA. In fact, 4D MRA avoids the use of radiation and, in using arterial spin-labeling techniques, contrast agents. However, these techniques are limited in various ways. Compared with DSA, 4D MRA provides reduced spatial resolution^18,22,23 with resulting difficulty in differentiating small arterial vessels^18,21,22 or draining veins.^19,20 Furthermore, ferromagnetic artifacts (eg, retainers, dental implants)¹⁸ and a long acquisition time²⁰ might contribute to reduced diagnostic quality. Still, recent literature advises performing DSA for definite evaluation of the preferred therapy strategy¹⁷ or the posttherapeutic follow-up of AVMs.²⁴

Another encouraging approach is time-resolved CT angiography (4D CTA), which has already found multifunctional applications in cerebrovascular diagnostics (eg, in the evaluation of dAVFs, AVMs, and cerebral collaterals in cases of vessel occlusion).²⁵ Several publications demonstrated the clinical feasibility²⁶ and a high agreement between 4D CTA and DSA.^27⇓–29 Although the spatial resolution of 4D CTA is rated higher than that of 4D MRA,³⁰ both modalities are struggling with the same limitations. Small vessels (eg, tiny feeders of an AVM²⁷) and specific angioarchitectural details of dAVFs²⁹ do not have the same visual quality as in DSA. Willems et al³¹ analyzed the value of 4D CTA in the diagnostic work-up of dAVFs in 11 cases. Although both readers fully agreed in 11 cases on an established classification system, there was 1 low-flow fistula that was not detected on 4D CTA at all. Therefore, relying on 4D CTA alone might lead to loss of detailed information (eg, on the exact number of feeders or the exact angioarchitecture of the nidus).

Although data acquisition was successful in all 26 cases, our analysis has several limitations. First, it was limited by the small amount of any pathology. Moreover, any dataset in our series was acquired by selective injections either via the ICA or VA. Therefore, our series does not cover datasets acquired by injection from the subclavian, external, or common carotid artery. Missing standardization for contrast application and postprocessing is the most important limitation of our analysis.