Improved Cerebral Time-of-Flight Magnetic Resonance Angiography at 7 Tesla – Feasibility Study and Preliminary Results Using Optimized Venous Saturation Pulses

Purpose Conventional saturation pulses cannot be used for 7 Tesla ultra-high-resolution time-of-flight magnetic resonance angiography (TOF MRA) due to specific absorption rate (SAR) limitations. We overcome these limitations by utilizing low flip angle, variable rate selective excitation (VERSE) algorithm saturation pulses. Material and Methods Twenty-five neurosurgical patients (male n = 8, female n = 17; average age 49.64 years; range 26–70 years) with different intracranial vascular pathologies were enrolled in this trial. All patients were examined with a 7 Tesla (Magnetom 7 T, Siemens) whole body scanner system utilizing a dedicated 32-channel head coil. For venous saturation pulses a 35° flip angle was applied. Two neuroradiologists evaluated the delineation of arterial vessels in the Circle of Willis, delineation of vascular pathologies, presence of artifacts, vessel-tissue contrast and overall image quality of TOF MRA scans in consensus on a five-point scale. Normalized signal intensities in the confluence of venous sinuses, M1 segment of left middle cerebral artery and adjacent gray matter were measured and vessel-tissue contrasts were calculated. Results Ratings for the majority of patients ranged between good and excellent for most of the evaluated features. Venous saturation was sufficient for all cases with minor artifacts in arteriovenous malformations and arteriovenous fistulas. Quantitative signal intensity measurements showed high vessel-tissue contrast for confluence of venous sinuses, M1 segment of left middle cerebral artery and adjacent gray matter. Conclusion The use of novel low flip angle VERSE algorithm pulses for saturation of venous vessels can overcome SAR limitations in 7 Tesla ultra-high-resolution TOF MRA. Our protocol is suitable for clinical application with excellent image quality for delineation of various intracranial vascular pathologies.

Ultra-high-field magnetic resonance imaging (MRI) at 7 Tesla is currently being introduced in clinical application [19][20][21][22]. Many technical difficulties have been overcome already, but several restricting issues have still to be resolved. One major limitation is imposed by specific absorption rate (SAR) restriction. To unleash the full potential of 7 Tesla MRI, modified pulse sequences have to be employed. Such sequences with special spatial presaturation pulses were first described by FELMLEE [23], and their implementation, application, and clinical relevance have been demonstrated in several subsequent publications [1][2][3]24,25]. SAR limitations restrict conventional saturation pulses at 7 Tesla, due to the fact Figure 1. Shown is the localizer scan (a) of a 51-year-old male patient (subject 24) with a large arteriovenous malformation in the left hemisphere with thickness of TOF MRA imaging slab (IS) and saturation pulse (SP) illustrated. In a 29-year-old healthy subject a TOF MRA slab with identical coverage and spatial resolution was acquired with the non-optimized sequence in 17:48 min (b: sagittal view, d: transverse view). With the optimized sequence the identical slab was acquired in 6:22 min (c: sagittal view, e: transverse view). The same representative slice is shown for both sequences for direct comparison. In the optimized sequence TE could be reduced from 6.62 ms to 4.34 ms (due to a different asymmetric echo factor) and TR could be reduced from 56 ms to 20 ms (mainly due to reduced SAR) leading to better image contrast. doi: 10 that these limits lead to unreasonable extension of acquisition time or suboptimal image contrast based on lowered flip angles of imaging excitation pulses. This study presents a protocol for ultrahigh-resolution TOF MRA suitable for clinical application.

Ethics Statement
The study was conducted according to the principles expressed in the Declaration of Helsinki and was approved by the local university institutional review board (Ethik-Kommission der medizinischen Fakultä t der Universitä t Duisburg-Essen). Written informed consent was obtained before each examination.

Study Design and Population
This prospective study evaluated the diagnostic ability of highresolution 7 Tesla TOF MRA for various intracranial vascular pathologies (aneurysms, AVM, AVF, Moya-Moya). The study group comprised of 25 neurosurgical patients (male n = 8, female n = 17; average age 49.64 years; range 26-70 years). Inclusion criteria were: 1) single or multiple intracranial vascular pathologies, 2) age 18-80 years, 3) ability to give informed consent and 4) legal competence. Exclusion criteria were: 1) monitoring on intensive care or intermediate care unit 2) cardiac pacemakers or any other electronic implants, 3) metallic implants, 4) pregnancy or breast feeding period, 5) claustrophobia and 6) chronic or episodic vertigo.

Scanner and Coil System
All images were acquired on a 7 Tesla whole-body MRI system (Magnetom 7 T, Siemens Healthcare, Erlangen, Germany) equipped with a 32-channel Rx/Tx head coil (Nova Medical, Wilmington, MA, USA). The scanner has a gradient system of 45 mT/m maximum amplitude and a slew rate of 200 mT/m/ ms.

Examination at 7 Tesla
Prior to sequence acquisition, B 0 shimming was performed using a vendor-provided gradient echo sequence and an algorithm based on the work of SCHAR [26]. A vendor-provided spin-echo sequence was used for B 1 field mapping and local flip angle optimization. Following slice-selective excitation, two refocusing pulses sequentially generate a spin-echo and a stimulated echo. Transmitter adjustment volume was placed in the center of the brain leading to an average flip angle variation of less than 20% in the superior sagittal sinus. Spatial coverage of the TOF MRA slab and the saturation pulse is demonstrated on a sagittal localizer scan of a 40-year-old male patient suffering from a large right parietal AVM (Figure 1). Time-averaged local SAR in Watts/kg was assessed for every scan. The values were measured by the SAR supervision system of the MR scanner and read out from the DICOM header of the acquired imaging data.

TOF MRA Sequence
The TOF MRA sequence is based on a 3D FLASH sequence with flow compensation and tilt-optimized non-saturated excitation (TONE) across the slab [27]. Datasets were acquired with an excitation flip angle of a = 18u, TE = 4.34 ms, TR = 20 ms, FOV 200 mm6169 mm646 mm, 112 slices per slab (oversampling 14%), GRAPPA acceleration factor R = 4 (phase direction), partial Fourier 6/8 in both slice and phase directions, matrix of 8966756 (non-interpolated), and a voxel size of 0.2260.2260.41 mm 3 in a total acquisition time of 6 min 22 s. For a shorter echo time, compared to the standard protocol, a different factor for the asymmetric echo was implemented in the modified protocol: 0.2 instead of 0.3 (ratio of columns acquired before and after echo). The variable-rate selective excitation (VERSE) algorithm [28] was used to reduce SAR contribution of excitation and venous saturation RF pulses [29]. Cutoff thresholds (percentage of maximum amplitude of the original pulse at which the pulse is cropped by the VERSE algorithm) of 50%/30% were chosen for excitation/saturation. The resulting duration was 0.6 ms/2 ms for excitation/saturation RF pulse (compared to 1 ms/3.8 ms in the unmodified standard protocol). A lower VERSE cutoff threshold for the saturation pulses than for the excitation pulses can be chosen because an exact saturation slice profile is not important as long as the saturation area is not overlapping with the imaging slice. The flip angle of the saturation RF pulses was additionally reduced (a SAT = 35u instead of 90u which is normally used) to face SAR constraints [27].

Venous Saturation Pulse
Modified saturation pulses were used to suppress the venous system within the slab and above including the superior sagittal sinus. To reduce total measurement time, the RF pulse and the following spoiler gradient were made shorter (2.3 ms [chosen duration before VERSE was applied]/4 ms instead of 4 ms/ 11 ms). To reduce SAR of the RF pulses, the VERSE algorithm was used on the saturation pulses with a 30% cut-off threshold. The flip angle, a SAT , of the saturation pulses was set to nominally 35u (variation of less than 20% in the superior sagittal sinus due to B 1 inhomogeneity) as suggested in our previously published paper [27] to allow high saturation without exceeding the SAR limitations.

Image Evaluation
Image evaluation was performed in consensus by two experienced neuroradiologists on standard post-processing Picture Archiving and Communcation system workstations (Centricity RIS 4.0i, GE Healthcare, USA). Visual evaluation was performed on 3D image reconstructions and on the source images in axial orientation.
For qualitative analysis the following features were evaluated, utilizing a five-point scale (5 = excellent, 4 = good, 3 = moderate, 2 = poor, and 1 = non-diagnostic): 1. Delineation arterial vessels in Circle of Willis 2. Delineation of vascular pathologies 4. Presence of artifacts 5. Vessel tissue contrast 6. Overall image quality Prior to quantitative measurements, signal intensities of all scans were normalized. The image histogram of each subject was stretched, and shifted to match the baseline histogram of subject 1. Although this method does not correct for intra-subject B 1 inhomogeneities, inter-scan variability of signal intensities caused by e.g. different transmitter voltage, head size and other factors is successfully minimized [30]. The histogram normalization was performed using the inormalize tool by ZIJDENBOS (version 1.5.1 for OS X as part of Medical Imaging NetCDF toolbox -MINC). MINC is a medical imaging data format and associated set of tools and libraries developed at the Montreal Neurological Institute and freely available online (http://www.bic.mni.mcgill.ca/ServicesSoftware).
After preprocessing quantitative signal intensity measurements were then performed by region-of-interest (ROI) analysis using the interactive display tool ''fslview'' (version 3. ð Þas well as adjacent gray matter Signal GM ð Þ . Diameters for ROIs slightly varied between subjects to account for anatomical variations and to maximize ROI size without including voxels from other tissue classes. The average diameter for the ROI of the CoS was 8-12 mm; the ROI for MCA was 3-5 mm in diameter; the ROI for brain parenchyma amounted to approximately 10 mm. In surgically treated Moyamoya patients ROIs were placed in the superficial temporal artery to middle cerebral artery bypass instead of the proximal M1 segment. Sufficient venous saturation in the confluence of sinuses was achieved by definition, if signal intensities in CoS were lower than in GM (VTCR_CoS.0). Arterial vessel tissue contrast was rated subjectively by both raters with focus on local angioarchitecture (VTCR_MCA.0.2 was defined as sufficient).
Differences of continuous scaled variables were tested by Student's t-test. For every mean value the standard error of mean (SE, s = ffiffi ffi n p ) was calculated as an estimate of the population mean.
Statistical analysis was carried out with the STATA software Delineation of Circle of Willis was rated excellent except for Moyamoya angiopathy, mainly due to the disease specific early ramifications of the middle cerebral artery branches. Delineation of vascular pathologies was rated excellent except for AVF. The complex angioarchitecture and close relationship to the skull bone complicate depiction in this pathology. Vessel tissue contrast was excellent in aneurysms and Moyamoya angiopathy, but poor to excellent in AVM and AVF. Arterialized hyperintense venous blood alters vessel tissue contrast in these patients. Presence of artifacts was rated mainly good to excellent for all pathologies and only pulsation artifacts led to lower ratings. VERSE specific artifacts did not significantly alter the images. Overall image quality was good to excellent for all pathologies. doi:10.1371/journal.pone.0106697.g002 Table 3. Quantitative normalized signal intensity measurements.

Results
All 25 neurosurgical patients tolerated the examination well and were successfully examined without exceeding SAR limitations (time averaged applied RF power in Watts/kg 8.493, range 6.077-9.988, SE 0.241) using 35u flip angles for saturation pulses. Table 1 gives an overview of basic demographic data and timeaveraged local SAR in Watts/kg for every scan. Table 2 lists detailed information about consensus ratings for vessel and pathology delineation, vessel-tissue contrast, artifacts and overall image quality. A lower rating for presence of artifacts in some of the patients was solely due to pulsation artifacts. Slight aliasing artifacts were observable only in the uppermost/lowermost 2-3 slices in the axial images, which did not impair evaluation by the raters. Further VERSE specific artifacts (e.g. slice profile distortions or slice shifts especially for off-resonant spins) were not observed. Ratings were good to excellent for the majority of patients and pathologies. Figure 2 presents the ratings in boxplots for all pathologies and all rated categories. Table 3 shows detailed information on ROI mean signal intensity in confluence of sinuses, middle cerebral artery M1 segment (superficial temporal artery to middle cerebral artery bypass in surgically treated patients) and adjacent gray matter. Calculations of vessel-tissue contrast ratios show sufficient vessel-tissue contrast for venous vessels, arterial vessels and adjacent gray matter for the majority of patients and pathologies. The applied venous saturation pulses resulted in vessel-tissue contrast ratios for all patients as follows: mean VTCR_CoS    Figure 4 shows a scan of a 42 years-old female patient (subject 20) after bilateral superficial temporal artery to middle cerebral artery bypass surgery.

Dural arteriovenous fistula (AVF)
Two patients with arteriovenous fistulas were enrolled in the study (male n = 2, female n = 0; average age 59.5 years; range 58-61 years). Consensus ratings were excellent for delineation of Circle of Willis (mean 5.0, range 5-5, SE 0.0), moderate and good for vascular pathology delineation (mean 3.5, range 3-4, SE 0.5), moderate for vessel-tissue contrast (mean 3.0, range 3-3, SE 0.0), good for presence of artifacts (mean 4.0, range 4-4, SE 0.0) and good for overall image quality (mean 4.0, range 4-4, SE 0.0). The applied venous saturation pulses resulted in vessel-tissue contrast ratios as follows: mean VTCR_CoS = 0.131 (range 0.121-0.141, SE 0.01); VTCR_MCA = 0.678 (range 0.628-0.728, SE 0.05). As in AVM patients venous saturation was lower compared to other pathologies due to inherent features of dural AVFs, in particular direct shunting of arterialized blood into the venous sinuses. High signal intensity of arterialized blood is similar to signal intensity of the left middle cerebral artery M1 segment. Signal intensity of venous sinuses decays the further away from the fistula point it is measured. The special features of dural AVF in 7 Tesla TOF MRA scans are exemplarily illustrated ( Figure 5) in a 58-year-old male patient (subject 11) with a complex occipital dural AVF (Cognard [32] type IIb) fed by branches of both external carotid and vertebral arteries.

Discussion
Time-of-flight MRA is a non-invasive imaging technique using FLASH gradient-echo magnetic resonance sequences to visualize the contrast between the high signal intensity of inflowing blood and saturated stationary tissue. TOF MRA utilizing 1.5 Tesla has been established as an excellent clinical diagnostic method for preoperative and pre-interventional planning as well as for follow-up controls in cerebro-vascular pathologies, and has gained ground on the accepted gold standard, invasive digital subtraction angiography.
The increase of the magnetic field strength from 1.5 Tesla to 3 Tesla has been proven beneficial for clinical neuroimaging and feasible for ultra-high-field 7 Tesla MRI, facilitating a significant increase in spatial resolution [4,8,9,[15][16][17]33,34]. MOENNINGHOFF et al. presented a high-resolution 7 Tesla TOF MRA protocol for clinical application including venous saturation pulses. To maintain SAR limitations, time-of-repetition (TR) had to be increased up to 57 ms, resulting in an increase in acquisition time to almost 15 minutes for comparable spatial coverage as the presented protocol. The previously suggested solution [27] to overcome SAR limitations within a patient tolerable scanning time has not yet been applied in a clinical setting for various intracerebral vascular pathologies.
The impairment of venous saturation at ultra-high field strength is mainly due to SAR limitations that prohibit the usage of venous saturation pulses within reasonable acquisition times.
One proposed solution to overcome SAR limitations is to play out a regular 90u saturation pulse only every n-th readout of the kspace [8]. The drawback of this approach would be that either measurement time is not used efficiently or excitation RF pulses experience different effective TRs which could lead to an artifact appearing similar to aliasing [35] caused by periodic signal variation in k-space lines. The exact effect of different effective TRs would have to be assessed prior to routine clinical application. It is also difficult to achieve a good balance between venous saturation and optimal SAR utilization, as this method is more rigid than mere flip angle variations. Our approach to overcome SAR limitations uses customized low flip angle VERSE algorithm [28,29] saturation pulses to significantly reduce the SAR of each individual saturation pulse rather than reducing the number of saturation pulses. The VERSE algorithm was used to decrease SAR contribution of excitation and saturation RF pulses. Based on this reduction, flip angles of saturation pulses were optimized for adequate venous saturation while remaining within SAR restrictions. Without optimization of flip angles and VERSE algorithm acquisition time was almost 3 times longer (6:22 min versus 17:48 min) in a 29-year-old healthy volunteer and TE had to be increased from 4.34 ms to 6.62 ms. Also the TR had to be increased from 20 ms to 56 ms to stay within SAR limits. Especially the prolonged TR leads to lower image contrast. A detailed description of the applied pulse sequence and considerations on ideal flip angles for saturation pulses have previously been published [27]. Differences in image contrast between nonoptimized and optimized TOF MRA sequences are illustrated in Figure 1.
Only in patients with AVMs and AVF, venous saturation was reduced compared to other pathologies. This was presumably due to the inherent features of both pathologies, specifically the fast blood flow of venous drainage. In addition, only part of the AVMs were within the field of saturation pulses and consequently unsaturated blood was drained over the veins. Even though the venous saturation was reduced in these patients, the application of saturation pulses could still partially suppress the signal of draining veins with an overall good image quality.
The high diagnostic relevance of saturation pulses in combination with ultra-high-resolution TOF MRA is exemplarily shown in the case of Subject 12 ( Figure 5). The complex dural AVF (COGNARD [32] type IIb) could be clearly visualized and the fistulous point with the hyperintense artery entering the hypointense venous sinus was clearly depicted by mixed unsaturated arterial and saturated venous blood. This effect can presumably only be visualized with effective venous saturation, and might not be clearly visible in TOF MRA imaging without venous saturation pulses or in contrast-enhanced T 1 -weighted images. Future patients suffering of AVF will be examined with and without venous saturation pulses to further investigate the diagnostic potential of this effect.
Consensus ratings of two experienced readers ranged between good and excellent for the majority of patients and most of the evaluated features. In combination with short acquisition time and possibility to visualize the complete intracranial arterial vasculature with ultra-high resolution the presented TOF MRA protocol is suitable for clinical application.
Clearly, this study is not free of limitations. The study group comprised of patients suffering from various intracranial vascular pathologies and therefore patient numbers for individual pathologies are small. Nevertheless, to the authors knowledge this is one of the largest neurosurgical patients cohorts scanned with TOF MRA at 7 Tesla, published in literature. Further investigations with larger patient cohorts of single pathologies should be the focus of future studies. Ultra-high-field imaging in patients who underwent treatment has been restricted so far, as Guglielmi detachable coils, aneurysm clips, most cranial fixation plates and other metallic implants are not certified for 7 Tesla MR imaging. Although preliminary results on implant safety in cerebral 7 Tesla MRI have been recently demonstrated [36][37][38], up to now avoiding metallic implants during surgery is the only feasible way for postoperative scans. The excellent image quality is exemplarily shown in a 42-year-old female patient (subject 25) after bilateral superficial temporal artery to middle cerebral artery bypass surgery (Figure 4). Future studies on the diagnostic potential of 7 Tesla MRI for follow-up of treated aneurysms, AVM, AVF and patients with Moyamoya angiopathy would be of high scientific and clinical interest. Another limitation is posed by the lack of a comparison to digital subtraction angiography, the diagnostic gold standard. However, with MRA offering equivalent non-invasive vessel diagnostics to DSA, the main focus of this trial was set on testing the feasibility and diagnostic potential of high-resolution TOF MRA at 7 Tesla utilizing novel low flip angle venous saturation pulses in a variety of intracranial vascular pathologies.
In summary, this study presents the clinical application of the previously reported practical solution for saturation pulses in 7 Tesla ultra-high-resolution TOF MRA that maintains SAR levels within the accepted safety standards combined with short acquisition times [27]. Excellent imaging quality was shown for the majority of patients in a larger cohort suffering from diverse cerebro-vascular pathologies.