Imaging of Intracranial Saccular Aneurysms

Abstract

Vascular imaging is an essential tool to appropriately diagnose and treat intracranial saccular aneurysms. There is extensive heterogeneity in aneurysm characteristics including location, size, shape, patient demographics, and clinical status that leads to a great diversity in both surgical and endovascular treatment options. This variability may elicit confusion when deciding the most appropriate imaging paradigm for an individual patient at particular time points. A collection of pre‐ and posttreatment scales and grades exist, but there is no current consensus on which one to implement. In this review, we discuss the key advantages and disadvantages of the available imaging modalities and how each can guide management. We also review novel imaging tools that are likely to alter the diagnostic landscape of intracranial aneurysms in the coming years.

Nonstandard Abbreviations and Acronyms

CE‐MRA

contrast‐enhanced magnetic resonance angiography

CTA

computed tomographic angiography

HR‐VWI

high‐resolution vessel wall imaging

ISAT

International Subarachnoid Aneurysm Trial

TOF‐MRA

time‐of‐flight magnetic resonance angiography

UIA

unruptured intracranial aneurysm

WSS

wall shear stress

Clinical Perspective

• Intracranial aneurysms are prevalent and generally have a low annual risk of devastating rupture.

• Each imaging modality has specific advantages and disadvantages in the evaluation and treatment of intracranial aneurysms.

• A collection of posttreatment scales has been developed, but there is no clear consensus on their utility.

• Ongoing and future clinical trials are needed to better understand the most appropriate imaging tool to use for each individual patient at each point in time.

The prevalence of incidentally found unruptured intracranial aneurysms (UIAs) is roughly 3.2% in a middle‐aged population without significant comorbidities, and the total number is increasing over time with the increased use of noninvasive imaging. ¹ , ² , ³ Many UIAs have a low rupture risk per year (0.1%–1.3%), ⁴ , ⁵ but aneurysmal subarachnoid hemorrhage (SAH) leads to mortality in 22% to 50% of patients and a high morbidity rate in those who do survive. ⁶ , ⁷ Vascular imaging is an essential step in the accurate diagnosis and treatment of intracranial aneurysms, but a fundamental challenge arises from the extensive heterogeneity of aneurysms. Cerebral aneurysms come in all shapes and sizes along the vascular tree. They are discovered in both the unruptured and ruptured state. Patients also present with a variety of demographic risk factors including (but not limited to) age, sex, family history, a history of hypertension, and a history of smoking exposure.

This clinical and morphological variability creates confusion regarding the optimal imaging approach for any individual patient. Observation and serial follow‐up are preferred for many patients, but the appropriate imaging frequency and modality is not clearly delineated. When procedural treatment for aneurysms is necessary, there is great diversity in surgical options such as clipping, trap‐and‐bypass, and parent‐vessel sacrifice, as well as a growing collection of endovascular options including primary coiling, coiling+stent or other neck remodeling device, flow diversion (±coils), flow disruption with intrasaccular devices (±stent), and parent‐vessel sacrifice. Within each category, there is additional variability in how the devices can be deployed (eg, device position and number of devices). Increasing therapeutic complexity has spawned a growing number of tailored imaging approaches to appropriately evaluate patient outcomes. Standard modalities include computed tomographic angiography (CTA), magnetic resonance angiography (MRA), and digital subtraction angiography (DSA), but a range of emerging tools, including high‐resolution vessel wall imaging (HR‐VWI), intraluminal imaging, and computational fluid dynamics (CFD), may transform aneurysm evaluation and treatment in the coming years.

In this review, we discuss the advantages and disadvantages of the available range of pre‐ and posttreatment imaging modalities and discuss the novel tools under development that have the potential to reshape the diagnostic landscape of intracranial aneurysms.

Standard Aneurysm Imaging Tools

Digital Subtraction Angiography

DSA is considered the gold standard for the evaluation of intracranial aneurysms because of excellent spatial and temporal resolution (Table ). The procedure is performed by a specialist, who obtains endovascular access with a catheter and injects contrast into the blood vessels of the neck and brain while simultaneously capturing images using ionizing radiation (ie, fluoroscopy). An add‐on method called 3‐dimensional rotational DSA is conducted by rotating the C‐arm around the patient while automatically injecting contrast. This method allows the operator to gain a more detailed and accurate view of the size, morphology, and neck of the aneurysm, as well as the parent‐vessel characteristics. This modality is also useful to obtain appropriate working angles for possible endovascular embolization and expected anatomy encountered during open surgical intervention. ⁸ , ⁹ The true sensitivity and specificity of DSA for detecting intracranial aneurysms is unknown given that it is the gold standard imaging modality. However, the false‐negative rate in patients with nontraumatic SAH was found to be 7.1% (2/28) in 1 retrospective study of 904 patients. ¹⁰ No vascular lesions were subsequently found in patients with cortical or perimesencephalic patterns of SAH. Similar results of 2 meta‐analyses demonstrated that ≈1 of 10 patients with aneurysmal‐pattern SAH with an initial negative DSA will detect an aneurysm on repeat DSA. ¹¹ , ¹² Notably, the missed lesions were mostly atypical such as blister, dissecting, and fusiform aneurysms. Repeat DSA is thus critical in patients with diffuse SAH with an initially negative DSA, yet the appropriate timing is not clearly established.

Table  .

Strengths and Weaknesses of the Standard and Emerging Imaging Tools Used to Assess Intracranial Saccular Aneurysms

Imaging modality	Strengths	Weaknesses
Standard imaging tools
MRA	Noninvasive No ionizing radiation	Lower spatial resolution No temporal resolution Long acquisition time Blood flow artifacts with TOF
CTA	Noninvasive Rapid acquisition Low cost	Ionizing radiation No temporal resolution Iodinated contrast side effects Artifact from metal and bone
DSA	Highest spatial resolution Highest temporal resolution	Invasive (risk of complications) Ionizing radiation High cost
Novel imaging tools
7 Tesla MRI	Noninvasive No ionizing radiation High spatial resolution	No temporal resolution Long acquisition time High cost
HR‐VWI	Potential to predict rupture	Longest acquisition times High artifact potential Unproven clinical utility
Intravascular imaging	Highest spatial resolution	Invasive (risk of complications) Stiff and large access catheters Unproven clinical utility
4‐dimensional imaging	Addition of temporal resolution Potential to predict rupture Noninvasive for MRA and CTA	Generally lower spatial resolution Unproven clinical utility

CTA indicates computed tomographic angiography; DSA, digital subtraction angiography; HR‐VWI, high‐resolution vessel wall imaging; MRA, magnetic resonance angiography; and TOF, time of flight.

Computed Tomographic Angiography

Several noninvasive methods can also be used to diagnose and follow‐up intracranial aneurysms. CTA is characterized as a volumetric helical computed tomographic study with a time‐optimized bolus of iodinated contrast to visualize arteries in the area of interest. In the emergency setting, CTA is conducted for a wide range of presenting symptoms including headache, neck pain, and weakness and incidentally detects cerebral aneurysms ≈3.3% of the time. ² CTA is relatively inexpensive, widely available, and has a fast acquisition time, but it is not routinely used to serially follow unruptured aneurysms because of the cumulative risk of ionizing radiation. Exceptions include patients unable to get alternate imaging studies, such as MRA. The sensitivity of CTA in detecting aneurysms is good, typically ranging from 96% to 98%. ¹³ , ¹⁴ However, compared with DSA, CTA may miss small aneurysms and falsely identify infundibula and venous structures as aneurysms. ¹⁵ , ¹⁶ , ¹⁷ Figure 1 demonstrates the increasing morphological detail that is appreciated with 3‐dimensional DSA compared with CTA; however, both methods may miss small irregular blebs that are directly visualized in the operating room.

Figure 1.

Multilobulated aneurysm visualized using 3 separate methods. A, CTA, B, 3‐dimensional DSA demonstrates increasing complexity to the aneurysm morphology. C, Direct visualization in the operating room. The white arrow indicates a small bleb that was not visualized using CTA and 3‐dimensional DSA imaging modalities. CTA indicates computed tomographic angiography; and DSA, digital subtraction angiography.

Magnetic Resonance Angiography

Time‐of‐flight magnetic resonance angiography (TOF‐MRA) is commonly used to detect and evaluate intracranial aneurysms because it can be conducted without the need for radiation or contrast. The sensitivity of MRA for detecting even small intracranial aneurysms (≤5 mm) has also been reported to be high, ranging from 98.2% to 98.7% in 1 study. ¹⁸ However, a recent large retrospective analysis comparing MRA with DSA found that the MRA sensitivity was only 50.9% (58/114) for detecting the irregular shape of small aneurysms < 5 mm, which is associated with rupture risk. ¹⁹

Contrast‐enhanced magnetic resonance angiography (CE‐MRA) is also effective in assessing intracranial aneurysms, ²⁰ , ²¹ with advantages including fast image acquisition, decreased motion artifact and larger field of view coverage at the cost of requiring gadolinium‐based contrast injection. One important application for CE‐MRA over TOF‐MRA is superior diagnostic accuracy in determination of residual aneurysms after treatment. This limitation of TOF‐MRA has been attributed to flow‐related disturbance in the presence of flow‐diverting stents or slow flow associated with coil embolization with resultant lower sensitivity in detecting aneurysm residual. ²² , ²³

Diagnostic Evaluation of Intracranial Aneurysms

The goal of the initial evaluation of a UIA is to determine whether the aneurysm should be observed over time or if it should be treated using one of the myriad therapeutic strategies (eg, coiling, flow diversion, surgery, or bypass). To answer this critical question, one must leverage the large pooled analyses of individual patient data and the resultant clinical scores created to predict the risk of UIA rupture. In 2014, the widely used PHASES risk score (population, hypertension, age, size of aneurysm, earlier aneurysmal SAH, and site of aneurysm) was developed on the basis of data from 8,382 patients and 10 272 UIAs and gives an estimate of the 5‐year rupture risk. ²⁴ A valid criticism of the PHASES score is that it does not include aneurysm morphology, an established risk factor for aneurysm rupture. ²⁵ The ELAPPS score (earlier aneurysmal SAH, location of aneurysm, age, population, size of aneurysm, and shape of aneurysm), developed in 2017, was derived from data including 1,507 patients with 1,909 UIAs and helps estimate the 3‐ and 5‐year aneurysm growth risks. ²⁶ From these large studies, it is clear that many growth/rupture risk factors are purely demographic, yet quality diagnostic imaging is required to obtain a full risk assessment for any individual aneurysm. The PHASES score was derived from MRA and CTA images, and the ELAPPS score was derived from MRA, CTA, and DSA images. In the future, as technology improves, clinical predictive models may rely more on novel imaging tools.

Aneurysm Location

In 2003, the International Study of Unruptured Intracranial Aneurysms included data from 1692 patients and 2686 untreated aneurysms and demonstrated that, compared with internal carotid artery aneurysms, tip of the basilar and posterior communicating artery aneurysms had a significantly higher risk of rupture (relative risk, 2.3 [95% CI, 1.1–4.8] and 2.1 [95% CI, 1.1–4.2], respectively). ²⁷ In 2012, the Natural Course of Unruptured Cerebral Aneurysm study in Japan included data from 5720 patients with 6697 UIAs and found that compared with aneurysms in the middle cerebral arteries, aneurysms in the anterior communicating arteries, and posterior communicating arteries were more likely to rupture (hazard ratio [HR], 2.02 [95% CI, 1.13–3.58] and 1.90 [95% CI, 1.12–3.21], respectively). ²⁸ The PHASES score used data from both International Study of Unruptured Intracranial Aneurysms and Unruptured Cerebral Aneurysm study, but also 4 other smaller cohorts with individual patient data and identified posterior cerebral artery/posterior communicating artery and anterior cerebral artery (ACA)/anterior communicating artery aneurysms as the highest risk and also identified middle cerebral artery aneurysms as having a slightly higher risk of rupture compared with internal carotid artery aneurysms. ²⁴ The PHASES score, while widely used, has its limitations, as it does not include a comprehensive assessment of aneurysm risks factors, and it is subject to the underlying biases of the studies utilized to generate the scoring system. The importance of aneurysm location may be highlighted by considering cavernous internal carotid artery aneurysms, which have a negligible risk of rupture compared with aneurysms beyond the distal dural ring. ²⁷

Aneurysm Size

Aneurysm size is a strong imaging predictor of aneurysm rupture and should be routinely assessed on all imaging studies. The measure is defined as the maximum aneurysm diameter and can be assessed on MRA and CTA but is most accurately represented by DSA because of higher spatial resolution. Spatial resolution varies by scanner and imaging protocol but typically, CE‐MRA ranges between 0.8 and 1.0 mm, TOF‐MRA between 0.5 and 0.6 mm, CTA between 0.35 and 0.6 mm (depending on detector row number), and DSA between 0.1 and 0.2 mm. ²⁹ , ³⁰ The resolution of DSA allows reliable identification of aneurysms sized as small as 1.5 mm, but CTA and MRA struggle to identify aneurysms smaller than 3 mm. Aneurysm size can be classified as small, medium, large, giant, and super giant, but these classifications are arbitrary and heterogeneous and best practice is to simply report the measurements. Using a reference size of <5 mm, PHASES identified increased risk of rupture for aneurysms of size 7–9.9 mm (HR, 2.4 [95% CI, 1.6–3.6]), 10–19.9 mm (HR, 5.7 [95% CI, 3.9–8.3]), and ≥20 mm (HR, 21.3 [95% CI, 13.5–33.8]) on multivariate analysis. ²⁴ Other recent studies and meta‐analyses have confirmed aneurysm size as a reliable risk factor for aneurysm growth and rupture. ²⁶ , ³¹ , ³² However, it is widely known that given the high prevalence of small aneurysms in the population (ie, small aneurysms are more common than large aneurysms), many will indeed rupture. This creates the paradox that despite a lower risk of rupture for any individual small aneurysm, small aneurysms still cause a significant proportion of all SAH. ³³ , ³⁴ Other risk factors such as demographics, aneurysm morphology (discussed below), and growth in size likely play a significant role in rupture of these small aneurysms. Annual aneurysm growth rates have been estimated to be ≈3% of aneurysms/year, and data suggest that growing intracranial aneurysms are 30 times more likely to rupture compared with stable intracranial aneurysms. ³¹ , ³⁵

Aneurysm Morphology

There is a growing consensus that morphological factors beyond maximum aneurysm diameter play a role in rupture potential. The pooled analysis of 1507 patients in the ELAPPS study found that “irregular aneurysm shape”—defined as the presence of blebs, aneurysm wall protrusions, or multiple lobes—led to a significant risk of aneurysm growth (multivariate HR, 1.45 [95% CI, 1.07–1.97]). ²⁶ A more recent cohort from China including 1216 patients and 1514 aneurysms found that aneurysm irregularity (lobulated or with blebs, and higher aspect ratio) was independently associated with rupture status (odds ratio [OR] for irregularity, 2.88 [95% CI, 2.20–3.77]; P<0.001; OR for aspect ratio, 1.12 [95% CI 1.01–1.24]; P=0.036). ³⁶

There are numerous techniques to assess the 2‐dimensional and 3‐dimensional morphology of aneurysms including (but not limited to) aneurysm width, perpendicular height, maximum height (from midpoint of neck), neck width, aspect ratio (height divided by neck width), parent‐vessel diameter, size ratio (ratio of maximum diameter to the parent‐vessel diameter), multiple lobes, blebs, aneurysm angle, and aneurysm volume (Figure 2). Two‐dimensional measurements can be achieved with MRA, CTA, and DSA; and again, DSA provides the most accurate assessment because of its high spatial resolution. Linear dimensions such as depth, maximum width, and neck width are typically measured using the digital imaging platform caliper tool. The tool is precise but not always accurate such that aneurysms <2 mm may vary in linear dimension by ±10%, and aneurysms 2 to 10 mm may vary in linear dimension by ±5% and still be considered stable on CTA. ³⁷ Manual 3‐dimensional measurements can be obtained to crudely estimate aneurysm volume using the ABC/2 method, ³⁸ but the field is clearly moving toward computer‐aided imaging assessments of complex aneurysm morphology. Other volumetric indices including undulation index, nonsphericity index, and ellipticity index have all been correlated to aneurysm instability. ³⁹ , ⁴⁰ One recent study implemented artificial intelligence to measure aneurysm volume in 5 patients who were managed conservatively but then went on to rupture and found that 2 of the 5 aneurysms had growth that was not initially appreciated using standard measurements in a high‐volume center. ⁴¹ The remaining morphological features presented in Figure 2 have been found to be significantly associated with aneurysm growth or rupture in small studies, yet large‐scale prospective aneurysm morphology studies are generally lacking. One general limitation of these features is that they are static, yet aneurysm morphology is inherently dynamic, changing even over the course of 1 cardiac cycle. Four‐dimensional imaging (discussed below) may be necessary to fully understand the morphological risk profile of any individual aneurysm. ⁴²

Figure 2.

A collection of metrics used to measure aneurysm morphology.

Novel Imaging Methods for Intracranial Aneurysms

7 Tesla MRI

Most aneurysms can be reliably assessed with 1.5 or 3 Tesla MRI systems. However, detection of smaller aneurysms (<5 mm) and distinguishing aneurysms from vascular anatomic variants such as infundibula can be difficult. ⁴³ , ⁴⁴ In recent years, technological advances have led to the development of 7 Tesla MRI systems that offer higher signal‐to‐noise ratio compared with 1.5 or 3 Tesla systems. ⁴⁵ One retrospective study performed 7 Tesla MRI on 30 patients with suspected aneurysms seen on 3 Tesla MRI and found that differentiation between aneurysms and vascular variants could be accomplished in every single case. ⁴⁶ Notably, in 18 of the 30 cases, the diagnosis was changed from intracranial aneurysm to normal vessel (infundibulum or perforating artery), which terminated unnecessary and costly follow‐up. 7 Tesla also has the potential to reduce reliance on invasive DSA for detailed imaging. One study compared 7 Tesla 3‐dimensional TOF and 7 Tesla magnetization‐prepared rapid acquisition gradient echo MRI to DSA in a series of 64 unruptured aneurysms and found excellent delineation of the parent vessel, aneurysm dome, and aneurysm neck using 7 Tesla MRI that compared well with DSA (kappa, 0.81–0.98). ⁴⁷ Routine clinical access to 7 Tesla MRI is still limited by cost and availability but this may change in the coming years. Further limitations of widespread adoption of 7 Tesla MRI include current implants for aneurysm treatment (eg, coils, stents, and clips) have not been proven compatible with 7 Tesla MRI, and many patients report transient subjective symptoms (eg, dizziness, nausea, and headache) associated with the increased magnetic flux density. ⁴⁸

Radiomics

Radiomics is a relatively new technique that implements postprocessing algorithms to extract potentially useful quantitative features from imaging data, and recently a few studies have demonstrated its potential utility for the analysis of intracranial aneurysms. Liu et al ⁴⁹ found that flatness (a measure that shows the relationship between the largest and smallest principal components of the shape) was significantly associated with aneurysm stability (OR, 0.584 [95% CI, 0.374–0.894]; P=0.015). Zhang et al ⁵⁰ implemented radiomics to assess features that may be related to incomplete aneurysm occlusion after flow diversion and found that higher elongation value (defined as the square root of the quotient of length of the largest and second‐largest principal component axes of the aneurysm) was independently associated with incomplete occlusion of aneurysms (OR, 0.03 [95% CI, 0.01–0.17]; P<0.001). However, a more recent study found that while flatness and elongation were relatively good predictors of rupture status (area under the curve values of 0.72 and 0.71, respectively), neither performed as well as more conventional morphological parameters such as aspect ratio (area under the curve, 0.75) and size ratio (area under the curve, 0.73). ⁵¹ More research is needed to determine the full clinical utility of radiomics in aneurysm characterization.

Aneurysm Wall Enhancement

Long‐term activation of the inflammatory cascade—including the presence of intraluminal macrophages, T lymphocytes, monocytes, mast cells, and complement activation—is a key pathological feature of intracranial aneurysms. ⁵² Chronic inflammation causes the death of smooth muscle cells, growth in myointimal fibrosis, and, ultimately, vessel wall thinning that leads to rupture. ⁵³ Histopathological evidence suggests underlying inflammation as a potential cause for aneurysm wall degeneration.

Aneurysm wall enhancement can be depicted by magnetic resonance HR‐VWI (Figure 3) and has been used as a surrogate of this degenerative process in unstable intracranial aneurysms. HR‐VWI correlates with macrophage infiltration, intraluminal thrombi, and vasa vasorum proliferation on histology. ⁵⁴ , ⁵⁵ , ⁵⁶ , ⁵⁷ High contrast‐to‐noise ratio and spatial resolution are required to successfully image the relatively thin intracranial vessel wall. This is achieved through “black blood” techniques (ie, intravoxel dephasing of flowing blood or preparation pulses) that suppress luminal blood. ⁵⁸ Adequate suppression of the surrounding cerebrospinal fluid is also critical and can be achieved through various optimized magnetic resonance (MR) sequences (eg, delay alternating with nutation for tailored excitation). ⁵⁸ , ⁵⁹ , ⁶⁰ The presence of aneurysm wall enhancement is usually judged by coregistration and comparison of HR‐VWI images performed before and after gadolinium contrast administration. Although no widely accepted scoring system has been developed, a common semiquantitative scoring system has been adopted by many investigators, including ours. ⁶¹ Using this 3‐scale scoring system, aneurysm wall enhancement is assessed as follows: 0=no wall enhancement or similar degree of enhancement to the normal arterial wall; 1=partial enhancement that may involve any areas of aneurysm such as dome, the intermediate portion, the neck, or a bleb; 2=circumferential enhancement, encompassing the entire wall of aneurysm. A logistic challenge for HR‐VWI is that long acquisition times on the order of 10 to 12 minutes are typically required, and the patient must remain still for the duration of acquisition to obtain diagnostic image quality for interpretation.

Figure 3.

Example of vessel wall enhancement for a 4‐mm left MCA bifurcation aneurysm in a 62‐year‐old man. A, Three‐dimensional volume rendered image (posteroanterior) shows a superiorly projecting left MCA aneurysm (white arrow). B, Precontrast HR‐VWI of the aforementioned left MCA aneurysm (blue arrow). C, Postcontrast HR‐VWI shows subtle enhancement of the aneurysm wall (yellow arrow). 3D indicates 3‐dimensional; HR‐VWI, high resolution vessel wall imaging; and MCA, middle cerebral artery.

One large meta‐analysis (including 1761 total aneurysms) assessed the utility of aneurysm wall enhancement for predicting aneurysm instability (growth or rupture) and found that aneurysm wall enhancement had a high sensitivity and only mixed specificity, ⁶² indicating that the lack of vessel wall enhancement is useful to rule out aneurysm instability; however, its presence did not necessarily indicate aneurysm instability. Therefore, patients without aneurysmal vessel wall enhancement may be able to avoid the inherent risk of treatment. Aneurysm wall enhancement was found to develop at a later time point in 1.5% of aneurysms in 1 separate prospective study, suggesting the continued need for follow‐up HR‐VWI. ⁶³ The low positive predictive value of HR‐VWI may be partially explained by the fact that slow intraaneurysmal flow can lead to pseudo vessel wall enhancement attributable to incomplete signal suppression. ⁶⁴ Enhanced signal suppression methods and higher‐resolution imaging may improve the predictive performance in the future. One recent study of 29 patients assessed aneurysm wall thickness and vessel wall enhancement with 7 Tesla TOF‐MRA and pre‐ and postcontrast black blood SPACE (fast spin echo with variable flip angle trains) MRI and found a significant correlation between the 2 measures (thicker aneurysms were associated with more enhancement), but the clinical significance of this finding still requires more investigation. ⁶⁵

Recently, there have been promising efforts to limit aneurysm growth and rupture by targeting the inflammatory cascade. Aspirin use, specifically, has been correlated with decreased aneurysm growth and rupture rates in several observational studies. ⁶⁶ , ⁶⁷ Roa et al also demonstrated that aspirin use was significantly associated with decreased aneurysm wall enhancement on HR‐VWI (OR, 0.22 [95% CI, 0.06–0.83]; P=0.026), ⁶⁸ further suggesting that HR‐VWI may be used as a noninvasive imaging biomarker for aneurysm instability.

A separate MRI technique called quantitative susceptibility mapping has been used in patients with sentinel headaches to detect microbleeds in the aneurysm wall, which may represent a biomarker of later presentation of SAH. ⁶⁹ However, because of artifacts, the technique is currently inadequate for visualization of aneurysms that are close to the skull base and air cells. A recent study assessed both quantitative susceptibility mapping and HR‐VWI in 40 subjects with 51 unruptured intracranial aneurysms and demonstrated that when using positive quantitative susceptibility mapping as a surrogate for microbleeds, HR‐VWI had a high 96.6% negative predictive value, ⁷⁰ thus supporting the previously discussed potential for HR‐VWI to rule out aneurysm instability. ⁶²

Current research efforts include 7 Tesla HR‐VWI and 3‐dimensional aneurysm wall enhancement mapping, which may allow for more detailed assessments of aneurysm wall blebs that are prone to rupture. ⁷¹ Additionally, ongoing research into functional imaging that targets molecular markers of inflammation in aneurysms has shown promise in initial pilot studies, but further safety and efficacy studies are needed. ⁷² , ⁷³

Intravascular Imaging

Intravascular ultrasound (IVUS) is a catheter‐based imaging procedure that uses an endovascular piezoelectric transducer to send ultrasound images into the adjacent tissue and detect reflected echoes to generate tomographic images of the vessel lumen and wall. ⁷⁴ The images are obtained by passing the IVUS catheter distal to the region of interest and slowly withdrawing the catheter at approximately 1 mm/s. ⁷⁵ IVUS was originally developed to determine the most appropriate treatment for indeterminate coronary atherosclerotic lesions and to assess for apposition after stent placement. ⁷⁶ IVUS has been used in a few small studies to assess the parent vessel during flow diversion cases. One study examined 5 patients undergoing flow diversion and the vessel wall and Doppler flow measurements with IVUS influenced the choice of stent size and number of stents placed in 3 of 5 patients. ⁷⁷ Another report detailed the use of IVUS to assess stent‐assisted coil embolization of a giant cavernous aneurysm in which the angiographic images were limited because of superimposition of the parent vessel in all projections. ⁷⁸ Overall, IVUS has been limited in the assessment of intracranial pathology because of poor navigability and large size of early IVUS catheters (available devices have approximate outer diameters of 3.5F; eg, https://www.bostonscientific.com/en‐US/products/ffr‐ivus‐systems/opticross‐18‐peripheral‐imaging‐catheter.html). In addition, IVUS is known to overestimate the size of the true lumen, which may be attributable to poor visualization of the lumen–intima interface. ⁷⁹ This could lead to incorrect sizing of flow‐diverting stents.

Optical coherence tomography (OCT) is a different intravascular imaging method that uses interferometry and light backscatter to visualize vessel wall and lumen morphology with a spatial resolution 10 times higher than IVUS or DSA. ⁸⁰ Furthermore, interobserver reliability in measurements with OCT is significantly greater when compared with IVUS. ⁸¹ , ⁸² In brief, the procedure requires clearance of red blood cells from the field of view (using saline or contrast flush), and afterwards, the OCT catheter equipped with an optical fiber and focusing lens is rotated and pulled back quickly through the lumen of the artery of interest. The technique is widely used in cardiac procedures; however, early neuroendovascular efforts frequently failed because of large (eg, 2.7F) and stiff catheters. ⁸³ , ⁸⁴ Recently, high‐frequency OCT has been developed with more navigable 1.2F catheters compatible with 0.017‐inch microcatheters specifically tailored for neuroendovascular purposes. ⁸⁵ Given axial resolution ranging from 10 to 20 μm, OCT is capable of visualizing lumen size, intimal flaps, patency of perforators and small branches, stent apposition, thrombus development, endothelialization, and all layers of the vessel wall, which may be helpful in predicting aneurysm rupture. ¹ , ⁸⁶ , ⁸⁷ Some notable limitations include a limited depth of view for OCT at ≈2 mm, compared with ≈5 mm for IVUS. ⁸¹ OCT imaging is also blunted by highly attenuating tissue such as lipid‐containing plaques. ⁸⁸

A third alternative to IVUS and OCT is direct visualization of the aneurysm and parent vessel with a fiber optic microangioscope. ⁸⁹ One recent device (Vena Medical, Kitchener, Ontario, Canada) was successfully navigated to the middle cerebral artery bifurcation in a cadaver model and could visualize the aneurysm neck and stent apposition in a rabbit model as well as aneurysm coiling and intrasaccular flow disruption in a porcine model. ⁹⁰ A notable limitation includes the necessity for flow arrest with a balloon guide and flushing with saline at ≈50 cc/min to obtain adequate visualization. In vivo experiments are currently needed to fully understand the feasibility and clinical utility in human patients.

Hemodynamic Imaging and Analysis

Computational Fluid Dynamics

CFD is an analytical tool derived from Digital Imaging and Communications in Medicine data sets of 3‐dimensional CTA or 3‐dimensional rotational angiography to create patient‐specific, dynamic flow models that can potentially be used to better understand aneurysm initiation, growth, and potential for rupture. ⁹¹ Several small studies have analyzed CFD for both unruptured and ruptured aneurysms and found that ruptured aneurysms have notable differentiating characteristics such as complex and unstable flow patterns, either low or high wall shear stress (WSS), high oscillatory shear index, low aneurysm formation indicator, prolonged relative residence time, and high oscillatory velocity index (Figure 4). ⁹² , ⁹³ , ⁹⁴ , ⁹⁵ WSS is a dynamic feature such that high WSS may be prominent at the initiation of bleb but then change to low WSS as the morphology changes. ⁹⁶ CFD can also be used to predict hyperplastic remodeling of aneurysms, which may help avoid intraoperative surgical risks such as inadequate temporary clipping and obstruction of small perforators. ⁹⁷ Several known risk factors for aneurysm recanalization include large size, a wide neck, and lower coil packing density. Using CFD, 1 small study also identified high WSS and small vortex formation near the aneurysm neck remnant of partially coiled aneurysms as imaging biomarkers significantly associated with aneurysm instability and recanalization. ⁹⁸ While promising, CFD is not commonly used in the clinical setting because of the complexity of the analytical pipelines, long processing times, limited resources, and still unestablished clinical utility.

Figure 4.

Example of rWSS created from 4‐dimensional flow MRI from the same 4‐mm left MCA bifurcation aneurysm observed from Figure 3 . A, There is a heterogeneous distribution of WSS with high values near the dome and low values near the neck (calculated rWSS‐heterogeneity indices: large SD, 0.60; and a dispersed IQR, 0.30–0.81). B, Zoomed‐in and rotated view of the same aneurysm. Red color indicates higher WSS, and blue indicates lower WSS. 4D indicates 4‐dimensional; IQR, interquartile range; MCA, middle cerebral artery; rWSS, relative wall shear stress; SD, standard deviation; and WSS, wall shear stress.

Four‐Dimensional CTA

Four‐dimensional CTA is an ECG‐gated, time‐resolved, volumetric computed tomographic imaging sequence that produces angiographic images from patient‐specific average flow rates with relatively good spatial resolution (≈0.5 mm³). ⁹⁹ The technique is increasingly being used to improve CFD hemodynamic analyses for cerebral aneurysms. Several studies have demonstrated that 4‐dimensional CTA can identify focal areas of irregular aneurysm wall motion, which may represent increased risk of rupture. ⁴² , ¹⁰⁰ Cancelliere et al ¹⁰¹ reported that geometric measurements in 4‐dimensional CTA and 3‐dimensional rotational angiography modalities were comparable, with slight differences seen in parent artery lumen diameter and curvature. They also found that 4‐dimensional CTA slightly underestimates aneurysmal sac volumes and overestimates aneurysm neck size, particularly for smaller aneurysms. Despite inferior spatial resolution, 4‐dimensional CTA advantages include its inherent noninvasiveness (ie, safe), cost effectiveness, and its ability to be implemented in community hospitals that do not have current neuroendovascular capabilities.

Four‐Dimensional MR

Typical MRI uses TOF sequences or gadolinium contrast to visualize intracranial vessels and their aneurysms. Four‐dimensional MR implements ECG‐synchronized 3‐dimensional phase‐contrast MRI and postprocessing techniques to noninvasively evaluate 3‐dimensional intracranial blood flow. ¹⁰² Four‐dimensional MR can assess blood flow volumes, flow velocities, and WSS in particular areas of interest. This may have clinical implications, as several small studies have demonstrated that abnormal WSS may be associated with thin aneurysm walls and aneurysm growth. ¹⁰³ , ¹⁰⁴ Other potentially useful 4‐dimensional MR metrics include inflow velocities and intra‐aneurysmal flow profiles, but further research is needed to clarify their clinical utility. Pereira et al ¹⁰⁵ assessed 10 patients with aneurysms treated by flow diversion and demonstrated that 4D‐MR could be used to analyze in‐vivo treatment‐related blood flow reduction (35%–71% in their cohort).

Four‐Dimensional DSA

Four‐dimensional DSA is a recently developed technique that allows for the reconstruction of 3‐dimensional DSA acquisitions into series of fully time‐resolved 3‐dimensional DSA volumes at frame rates of up to 30/s. ¹⁰⁶ , ¹⁰⁷ For aneurysm analysis, 4‐dimensional DSA enables 3‐dimensional viewing from any desired spatial projection at any time during the passage of the contrast bolus, thereby eliminating the problem of vascular overlap. ¹⁰⁸ Intra‐aneurysmal flow patterns created by 4‐dimensional DSA may also help neurointerventionalists focus coil embolization efforts toward areas of aneurysmal blood inflow more aggressively than aneurysmal blood outflow, thereby potentially reducing rupture potential. ¹⁰⁹

Posttreatment Imaging

There is no standard guideline for posttreatment imaging of intracranial aneurysms. This is because of the tremendous heterogeneity in both aneurysm characteristics (eg, location, size, and morphology) and treatment types (eg, microsurgical clipping, endovascular coiling, flow diversion, and flow disrupting saccular devices). The treatment methods have varying recanalization rates, and any individual aneurysm may not fall neatly into the patterns based on large trials. This creates uncertainty for individual clinicians and clinical trialists. How often should patients be serially imaged? For what duration? Which imaging modality is preferred?

Duration and Timing of Follow‐Up Imaging

Given that the majority of aneurysm recurrences occur in the first year following treatment, ¹¹⁰ the initial interval image is typically obtained within 3 to 6 months of treatment. If stable at that time, the next image is obtained around 6 months to 1 year later. If the aneurysm continues to demonstrate stability going forward, imaging may be spaced out to 1‐ to 5‐year intervals. Based on feasibility, clinical trials have typically limited imaging follow‐up to ≤5 years; thus, historically, it was common practice to stop imaging stable aneurysms after 5 years. However, 1 study demonstrated that recurrence happens in up to 12.4% of aneurysms between 5 and 10 years, particularly in patients with an initial treatment Raymond grade 2 (relative risk, 7.08 [99% CI, 1.24–40.37]) or aneurysms sized >10 mm (relative risk, 4.37 [99% CI, 1.83–10.44]). ¹¹¹ This indicates that longer (potentially lifelong) follow‐up duration is required for higher‐risk aneurysms.

Follow‐Up Imaging Modalities

DSA is the gold standard vascular imaging method and is typically preferred for the initial follow‐up imaging after aneurysm treatment. DSA has the highest spatial and temporal resolution and is ideal for patients who received any of the various treatments given the ability to limit imaging artifact from metal implants. The most notable downside of DSA includes the possibility of neurological complications in ≈2.6% of patients and permanent neurological disability in ≈0.14% of patients. ¹¹² These risks are untenable for long‐term monitoring, and noninvasive tools are thus recommended beyond the initial, high‐risk time period.

MRA is the noninvasive imaging modality of choice for long‐term monitoring. MRA can be conducted both as contrast enhanced or without gadolinium contrast as a time‐of‐flight study. CE‐MRA is generally preferred given that gadolinium contrast is relatively low risk, ¹¹³ the acquisition time is faster than TOF‐MRA, and TOF‐MRA has a lower sensitivity for slowly perfusing aneurysms. ¹¹⁴ , ¹¹⁵ One meta‐analysis found that in comparison to DSA for coiled aneurysms, TOF‐MRA had a sensitivity of 86% (95% CI, 82%–89%) and specificity of 84% (95% CI, 81%–88%) while CE‐MRA had a sensitivity of 86% (95% CI, 82%–89%) and specificity of 89% (95% CI, 85%–92%). ¹¹⁶ MRA follow‐up of stent‐assisted coiling, flow diversion, and endosaccular flow disruption is, unfortunately, less adequate. Endoluminal stents and flow disruptors create magnetic susceptibility artifacts and Faraday cage effects that can create the false appearance of parent artery stenosis or occlusion. ¹¹⁷ , ¹¹⁸ One small study of 22 patients found that CE‐MRA performed better than TOF‐MRA for aneurysm remnant detection (sensitivity 83% versus 50% and specificity 100% versus 100%, respectively); however, both demonstrated poor specificity for parent‐vessel stenosis/occlusion detection (sensitivity 100% versus 100% and specificity 63% versus 32%, respectively). ²³ For follow‐up of flow disruption with the Woven Endobridge device (Sequent Medical, Aliso Viejo, CA), 2 studies have demonstrated poor sensitivity with MRA. One demonstrated that only 3 of 5 inadequately occluded aneurysms were detected by MRA, ¹¹⁹ and the other study demonstrated a sensitivity of 25% for aneurysm remnant detection. ¹²⁰ Based on these trials, DSA is the preferred imaging modality for flow diversion and flow disruption, but the frequency and follow‐up duration is not clearly delineated.

CTA is a low‐cost, noninvasive alternative to MRI, yet there are notable short‐term side effects from iodinated contrast and long‐term side effects from cumulative radiation. In addition, image quality is severely hampered by hardening artifacts created by clips, coils, and stents. ¹²¹ Novel methods to reduce artifacts include metal artifact reduction algorithms and high‐spatial‐resolution CT scanners, ¹²² , ¹²³ but studies assessing the long‐term efficacy of CTA for aneurysm follow‐up are currently lacking.

Grading of Results

The formal assessment of aneurysm treatment is not straightforward. Over the years, a collection of grades and scales have been put forth that are difficult to apply and do not clearly predict clinical outcome. On one end of the spectrum, scales may be as simple as: were you able to deliver the coil (yes or no)? However, in most cases, a binary yes/no is simply not good enough to risk stratify patients going forward. Patient outcome (eg, modified Rankin Scale) could be considered as a clear end point, but this can be confounded by medical comorbidities that lead to decline from causes distinct from aneurysm treatment. In the ISAT (International Subarachnoid Aneurysm Trial) follow‐up trial, the occlusion grade after endovascular therapy and after neurosurgical treatment were introduced. ¹¹⁰ Occlusion grade after endovascular therapy had 3 grades: 1 for total occlusion, 2 for subtotal occlusion (minor residual sac filling or neck remnant), and 3 for incomplete filling or substantial residual sac filling. Occlusion grade after neurosurgical treatment had just 2 grades: 1 for total occlusion and 2 for incomplete occlusion.

The Raymond scale, developed in 2001 for endovascular therapy, is more widely used and is defined as class I, complete aneurysm occlusion; class II, residual aneurysm neck; and class III, residual aneurysm. ¹²⁴ In 2015, an update to the Raymond scale (the modified Raymond–Roy classification) was developed on the basis of retrospective data from 370 patients with 390 aneurysms. ¹²⁵ The study introduced 2 subclassifications of residual aneurysms: class IIIa, residual aneurysm with contrast within coil interstices; and class IIIb, residual aneurysm with contrast along aneurysm wall. The authors demonstrated that compared with class IIIa aneurysms, class IIIb aneurysms were less likely to improve over time (14.89% versus 83.34%; P<0.001), were more likely to remain incompletely occluded (85.11% versus 16.67%; P<0.001), and had a trend toward higher subsequent rupture rate (3.23% versus 0.00%; P=0.068), results that have been validated in an external cohort. ¹²⁶

Unfortunately, the modified Raymond–Roy classification is inherently inadequate for the evaluation of flow‐diverted aneurysms. In 2010, the O'Kelly–Marotta grading scale was developed to address this insufficiency. In this scale, both the volume‐of‐contrast and degree‐of‐contrast stasis are assessed from the arterial to the venous phase of the angiogram run. Each aneurysm is given an aneurysm filling grade A, complete (>95%); B, incomplete (5%–95%); C, neck remnant (<5%); or D, no filling (0%) as well as a stasis grade of 1, no stasis (clearance in arterial phase); 2, moderate stasis (clearance in capillary phase); or 3, significant stasis (contrast persists into venous phase). ¹²⁷ The scale has demonstrated good interobserver and excellent intraobserver variability. ¹²⁸ Two separate scales were also developed in 2011 that graded both aneurysm occlusion and parent‐vessel hemodynamics. ¹²⁹ , ¹³⁰ A principal feature of these scales is the assessment of the degree of contrast filling just after treatment, which has not clearly demonstrated predictive value in posttreatment aneurysm recurrence or rupture risk. The Flow‐Diverting Stent Score, created in 2017, implemented multivariate analysis of factors predictive of occlusion, but was based on the Raymond scoring system, which may be more appropriate for coiled aneurysms. More recently, in 2021, a new score—the 4F‐flow diversion predictive score—was developed. ¹³¹ The score implements 4 independent predictors of aneurysm occlusion: fusiform shape, flow jet, filling, and final stasis. In the study's validation group, excellent aneurysm occlusion discrimination was demonstrated with a receiver operating characteristic area of 0.894 (0.843–0.933). ¹³¹ Despite this collection of new flow diversion–specific scales, any new treatment‐specific scale will require prospective validation to gain widespread acceptance. Indeed, recent large flow diversion clinical trials still largely use the original 2001 Raymond–Roy classification. ¹³² , ¹³³ , ¹³⁴

Conclusions

The current standard‐of‐care in aneurysm imaging is to conduct DSA in the initial diagnostic evaluation and in the short‐term follow‐up period after treatment. For longer‐term follow‐up, transition to CE‐MRA is typically adequate, yet flow‐diverting and flow‐disrupting implants may require a longer duration of DSA follow‐up before transition to noninvasive imaging. Given the relative novelty of these devices, it is currently unclear how to assess their performance in the long term. Novel tools such as 7 Tesla MRI, inflammatory imaging, endovascular imaging, and hemodynamic analysis have shown promise in preliminary trials, but larger validation studies and increased resources are needed to expand their use. In the future, one or several of these tools may be incorporated into clinical predictive models to give a more accurate estimate of rupture potential. Finally, there is clearly a need for more uniform and translatable imaging metrics to evaluate patients with intracranial aneurysms. As the complexity of diagnostic and interventional options grows, large‐scale prospective collaborations are needed to gain consensus on the most appropriate metrics and scales to guide the clinical management of intracranial aneurysms.

Sources of Funding

None.

Disclosures

Dr Nael is a medical advisory board member for Olea Medical and a consultant for Brainomix. Dr Colby reports consultancy for Stryker, MicroVention, Cerenovus, and Medtronic. Dr Liebeskind reports consultancy as an imaging core lab for Cerenovus, Genentech, Medtronic, Stryker, and Rapid Medical. The remaining authors have nothing to disclose.