Quantitative Time-of-Flight Head Magnetic Resonance Angiography of Cerebrovascular Disease.

BACKGROUND: Standard Cartesian time-of-flight (TOF) head magnetic resonance angiography (MRA) is routinely used to evaluate the intracranial arteries, but does not provide quantitative hemodynamic information that is useful for patient risk stratification as well as for monitoring treatment and tracking changes in blood flow over time. Quantitative TOF (qTOF) MRA represents a new and efficient method for simultaneous evaluating the intracranial arteries and quantifying blood flow velocity, but it has not yet been evaluated in patients with cerebrovascular disease. PURPOSE: To evaluate qTOF for simultaneously evaluating the intracranial arteries and quantifying intracranial blood flow velocity in patients with cerebrovascular disease, without the need for a phase contrast (PC) scan. STUDY TYPE: Prospective. SUBJECTS: Twenty-four patients (18 female, 6 male) with cerebrovascular disease. FIELD STRENGTH/SEQUENCES: Head MRA at 3 T using gradient-echo 3D qTOF, standard Cartesian TOF, and PC protocols. ASSESSMENT: Three independent readers assessed arterial image quality using a 4-point scale (1: non-diagnostic, 4: excellent) and artifact presence. Total and component flow velocities obtained with qTOF and PC were measured. STATISTICAL TESTS: Wilcoxon signed-rank tests, Gwet's AC2, intraclass correlation coefficients (ICC) for absolute agreement, Bland-Altman analyses, tests of equal proportions. P values <0.05 were considered statistically significant. RESULTS: Averaged across readers and compared to standard Cartesian TOF, qTOF significantly improved overall arterial image quality (3.8 ± 0.2 vs. 3.6 ± 0.5), image quality at locations of pathology (3.7 ± 0.5 vs. 3.4 ± 0.7), and increased the proportion of evaluations rated without artifacts (63.9% [46/72] vs. 37.5% [27/72]). qTOF significantly agreed with PC for total flow velocity (ICC = 0.71) and component flow velocity (ICC = 0.89). DATA CONCLUSION: qTOF angiography of the head matched or improved upon the image quality of standard Cartesian TOF, reduced image artifacts, and provided quantitative hemodynamic data, without the need for a PC scan. EVIDENCE LEVEL: 2 TECHNICAL EFFICACY: Stage 2.

Dark Blood Contrast-Enhanced Brain MRI Using Echo-uT1 RESS.

BACKGROUND: The widely used magnetization-prepared rapid gradient-echo (MPRAGE) sequence makes enhancing lesions and blood vessels appear bright after gadolinium administration. However, dark blood imaging using T1-weighted Sampling Perfection with Application optimized Contrast using different flip angle Evolution (T1 SPACE) can be advantageous since it improves the conspicuity of small metastases and leptomeningeal disease. As a potential alternative to T1 SPACE, we evaluated a new dark blood sequence called echo-uT1 RESS (unbalanced T1 Relaxation-Enhanced Steady-State). PURPOSE: We compared the performance of echo-uT1 RESS with Dixon fid-uT1 RESS, MPRAGE, and T1 SPACE. STUDY TYPE: Retrospective, IRB approved. SUBJECTS/PHANTOM: Phantom to assess flow properties of echo-uT1 RESS. Twenty-one patients (14 female, age range 35-82 years) with primary and secondary brain tumors. FIELD STRENGTH/SEQUENCES: 3 Tesla/MPRAGE, T1 SPACE, Dixon fid-uT1 RESS, echo-uT1 RESS. ASSESSMENT: Flow phantom signal vs. velocity as a function of flip angle and sequence. Qualitative image assessment on 4-point scale. Quantitative evaluation of tumor-to-brain contrast, apparent contrast-to-noise ratio (aCNR), and vessel-to-brain aCNR. STATISTICAL TESTS: Friedman and Mann-Whitney U tests. A P value <0.05 was considered statistically significant. RESULTS: In the phantom, echo-uT1 RESS showed greater flow-dependent signal loss than fid-uT1 RESS. In patients, blood vessels appeared bright with MPRAGE, gray with fid-uT1 RESS, and dark with T1 SPACE and echo-uT1 RESS. For MPRAGE, Dixon fid-uT1 RESS, echo-uT1 RESS, and T1 SPACE, respective tumor-to-brain contrast values were 0.6 ± 0.3, 1.3 ± 0.5, 1.0 ± 0.4, and 0.6 ± 0.4, while normalized aCNR values were 68.9 ± 50.9, 128.4 ± 59.2, 74.2 ± 42.1, and 99.4 ± 73.9. DATA CONCLUSION: Volumetric dark blood contrast-enhanced brain MRI is feasible using echo-uT1 RESS. The dark blood effect was improved vs. fid-uT1 RESS, while both uT1 RESS versions provided better tumor-to-brain contrast than MPRAGE. Whereas T1 SPACE provided better tumor aSNR, echo-uT1 RESS provided better Weber contrast, lesion sharpness and a more consistent dark blood effect. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 1.

Improved Brain Tumor Conspicuity at 3 T Using Dark Blood, Fat-Suppressed, Dixon Unbalanced T1 Relaxation-Enhanced Steady-State MRI.

OBJECTIVES: Contrast-enhanced magnetic resonance imaging (MRI) is the cornerstone for brain tumor diagnosis and treatment planning. We have developed a novel dual-echo volumetric dark blood pulse sequence called Dixon unbalanced T1 relaxation-enhanced steady-state (uT 1 RESS) that improves the visibility of contrast-enhancing lesions while suppressing the tissue signals from blood vessels and fat. The purpose of this study was to test the hypothesis that Dixon uT 1 RESS would significantly improve the conspicuity of brain tumors compared with magnetization-prepared rapid gradient echo (MPRAGE), as well as to determine potential limitations of the technique. MATERIALS AND METHODS: This retrospective study was approved by the hospital institutional review board. Forty-seven adult patients undergoing an MRI scan for a brain tumor indication were included. Contrast-enhanced MRI of the brain was performed at 3 T using both MPRAGE and Dixon uT 1 RESS. To control for any impact of contrast agent washout during the scan procedure, Dixon uT 1 RESS was acquired in approximately half the subjects immediately after MPRAGE, and in the other half immediately before MPRAGE. Image quality, artifacts, and lesion detection were scored by 3 readers, whereas lesion apparent signal-to-noise ratio and lesion-to-background Weber contrast were calculated from region-of-interest measurements. RESULTS: Image quality was not rated significantly different between MPRAGE and Dixon uT 1 RESS, whereas motion artifacts were slightly worse with Dixon uT 1 RESS. Comparing Dixon uT 1 RESS with MPRAGE, the respective values for mean lesion apparent signal-to-noise ratio were not significantly different (199.31 ± 99.05 vs 203.81 ± 110.23). Compared with MPRAGE, Dixon uT 1 RESS significantly increased the tumor-to-brain contrast (1.60 ± 1.18 vs 0.61 ± 0.47 when Dixon uT1RESS was acquired before MPRAGE and 1.94 ± 0.97 vs 0.82 ± 0.55 when Dixon uT 1 RESS was acquired after MPRAGE). In patients with metastatic disease, Dixon uT 1 RESS detected at least 1 enhancing brain lesion that was missed by MPRAGE on average in 24.7% of patients, whereas Dixon uT 1 RESS did not miss any lesions that were demonstrated by MPRAGE. Dixon uT 1 RESS better detected vascular and dural invasion in a small number of patients. CONCLUSIONS: In conclusion, brain tumors were significantly more conspicuous at 3 T using Dixon uT 1 RESS compared with MPRAGE, with an approximately 2.5-fold improvement in lesion-to-background contrast irrespective of sequence order. It outperformed MPRAGE for the detection of brain metastases, dural or vascular involvement. These results suggest that Dixon uT 1 RESS could prove to be a useful adjunct or alternative to existing neuroimaging techniques for the postcontrast evaluation of intracranial tumors.

Super-resolution head and neck MRA using deep machine learning.

PURPOSE: To probe the feasibility of deep learning-based super-resolution (SR) reconstruction applied to nonenhanced MR angiography (MRA) of the head and neck. METHODS: High-resolution 3D thin-slab stack-of-stars quiescent interval slice-selective (QISS) MRA of the head and neck was obtained in eight subjects (seven healthy volunteers, one patient) at 3T. The spatial resolution of high-resolution ground-truth MRA data in the slice-encoding direction was reduced by factors of 2 to 6. Four deep neural network (DNN) SR reconstructions were applied, with two based on U-Net architectures (2D and 3D) and two (2D and 3D) consisting of serial convolutions with a residual connection. SR images were compared to ground-truth high-resolution data using Dice similarity coefficient (DSC), structural similarity index measure (SSIM), arterial diameter, and arterial sharpness measurements. Image review of the optimal DNN SR reconstruction was done by two experienced neuroradiologists. RESULTS: DNN SR of up to twofold and fourfold lower-resolution (LR) input volumes provided images that resembled those of the original high-resolution ground-truth volumes for intracranial and extracranial arterial segments, and improved DSC, SSIM, arterial diameters, and arterial sharpness relative to LR volumes (P < .001). The 3D DNN SR outperformed 2D DNN SR reconstruction. According to two neuroradiologists, 3D DNN SR reconstruction consistently improved image quality with respect to LR input volumes (P < .001). CONCLUSION: DNN-based SR reconstruction of 3D head and neck QISS MRA offers the potential for up to fourfold reduction in acquisition time for neck vessels without the need to commensurately sacrifice spatial resolution.

Nonenhanced arterial spin labeled carotid MR angiography using three-dimensional radial balanced steady-state free precession imaging.

PURPOSE: To optimize and preliminarily evaluate a three-dimensional (3D) radial balanced steady-state free precession (bSSFP) arterial spin labeled (ASL) sequence for nonenhanced MR angiography (MRA) of the extracranial carotid arteries. MATERIALS AND METHODS: The carotid arteries of 13 healthy subjects and 2 patients were imaged on a 1.5 Tesla MRI system using an undersampled 3D radial bSSFP sequence providing a scan time of ∼4 min and 1 mm(3) isotropic resolution. A hybridized scheme that combined pseudocontinuous and pulsed ASL was used to maximize arterial coverage. The impact of a post label delay period, the sequence repetition time, and radiofrequency (RF) energy configuration of pseudocontinuous labeling on the display of the carotid arteries was assessed with contrast-to-noise ratio (CNR) measurements. Faster, higher undersampled 2 and 1 min scans were tested. RESULTS: Using hybridized ASL MRA and a 3D radial bSSFP trajectory, arterial CNR was maximized with a post label delay of 0.2 s, repetition times ≥ 2.5 s (P < 0.05), and by eliminating RF energy during the pseudocontinuous control phase (P < 0.001). With higher levels of undersampling, the carotid arteries were displayed in ≤ 2 min. CONCLUSION: Nonenhanced MRA using hybridized ASL with a 3D radial bSSFP trajectory can display long lengths of the carotid arteries with 1 mm(3) isotropic resolution.

Magnetic resonance imaging evaluation of cerebral cavernous malformations with susceptibility-weighted imaging.

BACKGROUND: Cerebral cavernous malformations (CCMs) can be sporadic or inherited, the latter characterized by multiple lesions. Novel imaging sequences have increased the sensitivity of detecting multiple CCMs. OBJECTIVE: To compare T2-weighted gradient echo (T2*GRE) and susceptibility-weighted imaging (SWI) sequences in familial and sporadic CCM to assess their respective sensitivity. METHODS: This prospective study included 23 consecutive cases grouped as multifocal/familial CCMs (n=14), solitary/clustered sporadic CCMs with developmental venous anomaly (n=8), and postirradiation CCMs (n=1). Brain magnetic resonance imaging included T2*GRE and SWI sequences. Two radiologists independently counted the number of lesions on each sequence. The difference in the number of lesions on both sequences was compared, and interobserver agreement was evaluated. RESULTS: In multifocal/familial cases, a mean of 34.7 lesions were detected on T2*GRE and 66.9 on SWI (P=.001). The difference of lesion prevalence with the 2 techniques was significant (P=.006), with strong interobserver correlation for the T2*GRE sequence (P<.001) and SWI sequence (P<.001). Patients with solitary/clustered sporadic CCMs, including those associated with venous anomaly, had no difference in lesion prevalence in the 2 sequences. CONCLUSION: SWI is more sensitive than T2*GRE in detecting CCM in multifocal/familial CCMs. Among cases classified as solitary/clustered with conventional imaging, including those associated with venous anomaly, the SWI did not impart additional sensitivity or reveal occult lesions not evident on T2*GRE sequence. No case was changed from the solitary/clustered to the multifocal clinical category because of SWI.

Cerebral venography using fluid-suppressed STARFIRE.

Fluid-suppressed STARFIRE (Signal Targeting Alternative Radiofrequency and Flow-Independent Relaxation Enhancement) is a noncontrast method for flow-independent MR venography (MRV). It uses magnitude subtraction of two inversion recovery-prepared segmented three-dimensional (3D) balanced steady-state free precession acquisitions to obtain isotropic cerebral venograms in which both fat and cerebrospinal fluid signals are suppressed. Unlike two-dimensional time-of-flight (2D TOF) MRV, it is insensitive to the flow velocity of the cerebral veins. The method provided excellent depiction of the dural venous sinuses and cortical veins on maximum intensity projection images. Fluid-suppressed STARFIRE and 2D TOF were compared with contrast-enhanced 3D MRV as the reference standard in seven healthy subjects at 1.5 Tesla. Fluid-suppressed STARFIRE compared favorably to 2D TOF on both quantitative and qualitative analyses. Contrast-enhanced MRV provided the highest vein-background relative contrast and best demonstrated the straight sinus, whereas STARFIRE depicted the most venous branches. Further investigation will be required to determine the accuracy for cerebral venous thrombosis.