Noncontrast Magnetic Resonance Angiography for the Diagnosis of Peripheral Vascular Disease

The assessment of the peripheral vascular system using magnetic resonance imaging has had extensive clinical validation in patients with lower extremity arterial disease.^1,2 Recent advances in magnetic resonance (MR) hardware (eg, high-performance gradients and multichannel receive coil arrays), pulse sequences (eg, rapid gradient-echo-based sequences), and image reconstruction algorithms (eg, parallel imaging) have facilitated rapid imaging with high spatial resolution and signal to noise ratio (SNR), particularly to image distal vessels. Contrast-enhanced MR angiography (CE-MRA) performed with gadolinium-based contrast is a worthy alternative to computed tomography angiography (CTA) and avoids exposing patients to iodinated contrast material and ionizing radiation. However, given the high prevalence of chronic kidney disease in patients with peripheral vascular disease³ and the association of gadolinium-based contrast media and nephrogenic systemic fibrosis in patients with severe chronic kidney disease,⁴ contrast-free alternatives to CE-MRA are appealing. Moreover, the recent concerns with long-term deposition of gadolinium in the brain and bones have further strengthened interest in noncontrast MRA (NC-MRA) approaches.^5–8

In the past several years, review articles summarizing NC-MRA techniques used for diagnosis of disease in various vascular districts have been written.^9,10 The purpose of our review is to summarize technological advances and the most appropriate techniques for the study of lower limb arteries in patients with peripheral arterial disease (PAD). Both established and emerging NC-MRA techniques for evaluating the lower extremities are contrasted, and a clinical perspective on the utility of these techniques and their advantages has been provided.

Inflow-Based Techniques

Time of Flight

Time of flight (TOF) is the oldest NC-MRA technique, with contrast in the technique based on the differences in magnetization (manifested as magnetic resonance imaging signal intensity) between the flowing blood and stationary tissue. Flow-related signal enhancement is created when the inflowing unsaturated blood enters the imaging slice, so it can be subject to an radiofrequency (RF) excitation pulses and can be replenished by free unsaturated blood before the next RF pulse, while the stationary tissue in the imaging slice experiences repeated RF pulses, suppressing longitudinal magnetization and therefore signal. Gradient-echo imaging readouts are used for TOF imaging, with the slice oriented in an axial plane approximately perpendicular to the primary direction of vascular flow. As needed, venous or arterial blood suppression is achieved by using selective saturation pulses that are applied parallel and adjacent to the imaging slices (either below or above) to create either arteriograms or venograms, respectively. Two-dimensional TOF of the lower extremities requires the use of cardiac gating to avoid ghost artifacts related to multiphasic flow in healthy lower extremity arteries. In TOF imaging, vascular signal intensity depends on the amount of unsaturated blood in the slice at the time of sampling, which in turn depends on blood velocity, repetition time (TR) between applied RF pulses, and thickness of the imaging slice (Figure I in the Data Supplement). In general, thinner slices, faster-flowing blood, vessels oriented perpendicular to the slice, and a longer TR increase vascular signal. The use of a long TR, however, increases signal from stationary tissue and lengthens the acquisition time, so an intermediate to short TR is often selected. Short echo time values (<7 ms) are used to minimize phase dispersion–related signal loss from accelerating or turbulent blood flow near stenosis. In addition, larger flip angle leads to reduced signal from stationary tissue, but it can also result in increased saturation of blood experiencing multiple excitation pulses. Thus, a flip angle that balances these 2 competing priorities is typically chosen (usually 30°−60° are used). If the distance traveled by blood in each TR is greater or equal to the slice thickness, the entire volume of blood within the slice is replaced by new fresh blood, allowing generation of high vascular signal. If a portion of the unsaturated blood volume is replaced by saturated blood, because of repeated RF pulses, this may result in lower vascular signal. Additional loss of signal may occur due to flow-induced spin dephasing caused by the magnetic field gradients used for imaging and may be reduced by the use of flow-compensated imaging gradients.¹¹

Clinical Utility and Preferred Vascular Application(s)

TOF imaging is sparingly used for NC-MRA of lower extremity peripheral vascular disease because of the existence of newer protocols that provide improved image quality and shorter scan times and its susceptibility to artifacts (partial saturation effects related to in-plane flow and flow-related turbulence) that may mimic or exaggerate a stenosis.^12,13 The application of parallel imaging approaches such as compressed sensing may reduce scan times and improve image quality of TOF-based protocols, although the clinical utility of such approaches remains to be determined.¹⁴ TOF, however, remains a popular technique for the evaluation of the extracranial and intracranial arteries.¹³

Quiescent Interval Slice-Selective MRA

Quiescent interval slice-selective (QISS) MRA is a cardiac (ie, electrocardiogram [EKG])-gated technique initially described by Edelman et al¹⁵ for the evaluation of the lower extremities. A recent review article by Edelman et al¹⁰ focuses on new developments and implementations of this technique. Our purpose is to describe the basics principles and the evidence about the application in clinical practice for the evaluation of PAD of this approach.

This technique uses a single-shot 2D balanced steady-state free precession (bSSFP) readout, which addresses many of the limitations of TOF imaging by reducing scan times and arterial saturation effects and leveraging the pulsatile nature of blood flow to better visualize lower extremity arteries. After detection of the EKG R-wave, a slice-selective saturation RF pulse is applied to the imaging plane, and a tracking venous saturation RF pulse is applied below the imaging slice. The 2 saturation pulses are followed by a quiescent interval phase (≈280 ms) that overlaps systole during which fully magnetized arterial blood (not affected by the in-plane saturation pulse) flows into the imaging slice. Alignment of the quiescent interval with the systolic phase of rapid blood flow ensures adequate inflow into the imaging slice; the quiescent interval is also short enough to maintain good suppression of stationary background tissue (Figure II in the Data Supplement). Following the quiescent interval, a frequency-selective fat saturation RF pulse is applied immediately before data acquisition to suppress the appearance of fat signal, which would otherwise appear bright. A single-shot bSSFP image acquisition is thereafter applied because it provides high signal from blood, allowing for time-efficient imaging of a single slice in each heartbeat (Figure 1). The sequence of saturation pulses, quiescent interval, and single-shot acquisition is repeated over many sequentially acquired contiguous or overlapping slices to cover the anatomy of interest.

• Download figure
• Download PowerPoint

Due to the use of a quiescent interval that overlaps, systole QISS MRA is effective in low-flow situations and is, therefore, particularly suited for angiographic evaluation of the lower extremities. It is also robust against motion and cardiac arrhythmias due to the use of single-shot acquisition. Other advantages include ease of use, as no protocol tailoring is necessary, and the relatively quick scan time that allows full coverage of the entire peripheral arteries (from pelvis to ankles) in ≈7 minutes (excluding the time for shimming). Venography is possible with QISS MRA by flipping the location of the tracking saturation pulse (to suppress arteries instead of veins) and lengthening the quiescent interval to accommodate the inflow of slower venous blood into the imaging slice (Figure 1). Example of QISS in patients with PAD is depicted in Figure 2. Potential limitations include suppression of in-plane-oriented vessel segments that reside within the saturation region and limited superior-inferior resolution due to limited minimal slice thickness. The use of a bSSFP readout renders QISS sensitive to potential off-resonance artifacts from metallic implants (eg, prostheses and stents) and air-filled bowel loops in the pelvis located near the arteries.

• Download figure
• Download PowerPoint

Clinical Utility and Preferred Vascular Application(s)

When compared with either digital subtraction angiography (DSA) or CE-MRA as the reference standard examination, QISS MRA demonstrates high sensitivity and specificity for the detection of ≥50% stenosis of the lower extremities. Across 8 studies^16–22 using DSA as the reference standard exam (153 total patients), median values for the sensitivity and specificity of QISS NC-MRA were 91.4% and 96.4%, respectively. In 9 studies^{16–19,23–27} using CE-MRA as the reference standard examination (254 total patients), median values for the sensitivity and specificity of QISS were 89.2% and 96.0%, respectively (Table 1).

At 3 Tesla, the performance of QISS appears to be slightly better with 92.7% sensitivity and 95.6% specificity with respect to DSA (81 total patients) and 89.2% sensitivity and 96.1% specificity with respect to CE-MRA (129 total patients). QISS has also shown good accuracy in patients with glomerular filtration rates ≤60 mL/min per 1.73 m² and diabetes mellitus.^16,17 In comparison with CTA, QISS demonstrates better sensitivity and specificity for >50% stenosis (Table 1), with high inter-observer agreement. Moreover, QISS is reportedly less prone to artifacts than CTA and better portrays heavily calcified vessels.^21,29 Other more recent applications of QISS include the evaluation of the coronary, pulmonary, extracranial carotid, and intracranial arteries,^42–45 with some protocols leveraging radial k-space sampling and spoiled gradient-echo readouts to improve image quality and combat local magnetic field inhomogeneity.

The original implementations of QISS have been shown to be reliable for the evaluation of PAD although with some limitations related to suppression of in-plane-oriented vessel segments residing within the saturation region and the sensitivity to off-resonance artifacts (Table 2). These features render QISS suitable for lower limb arteries evaluation in patients with known or suspected PAD to assess the degree of stenosis. However, arterial districts where susceptibility artifacts are less likely to occur and especially in long vertical segments (eg, with no metallic elements or air), such as femoral or calf vessels, are ideal for this technique (Table 3). QISS is also the NC-MRA technique that is used increasingly for rapid imaging of the renal and pulmonary arteries.¹⁰

Cardiac Phase and Flow Dependent Techniques

Three-Dimensional Fast (Turbo) Spin Echo Techniques

Three-dimensional (3D) fast spin echo (FSE) techniques are versatile techniques that can be used to image the lower extremity circulation. While the precise implementation may vary across the manufacturer, they usually incorporate the following features: (1) 3D-fast (turbo) spin echo acquisition, (2) long echo train lengths, (3) ultrashort echo spacing, typically 3 to 4 msec, (4) nonselective refocusing pulses, (5) reduced flip angles (may be constant but usually variable, typically 30°–120°), (6) partial Fourier imaging schemes and parallel imaging approaches to reduce imaging time, (7) optimized/efficient k-space trajectories with sampling in both in-plane and through-slab phase-encode directions during signal evolution, and (8) reasonable imaging times (5–10 minutes).

Reduced flip angle refocusing approaches result in a complex combination of spin and stimulated echoes that produce a signal that persists and depends on both T1 and T2. The use of variable (low flip angle) spatially nonselective radiofrequency refocusing pulses help suppress blurring, lengthen the useable duration of the spin-echo train, and reduce RF energy deposition and specific absorption rate.^46,47 The variable flip angle pulse train typically starts with higher amplitude pulses and slowly decreases to approach a constant (asymptotic) value of longitudinal magnetization. The results are scans that are faster, with lower specific absorption rate, where the SNR (despite using lower flip angles) is close to those acquired with 180° refocusing pulses. The additional use of parallel imaging techniques and navigator gating may enable the acquisition of high-resolution 3D turbo spin echo data (isotropic <1 mm spatial resolution at 3T) that has been quite useful for the imaging of lower extremity arterial circulation.

EKG-Gated Techniques

The idea of using the insensitivity of spin-echo signal to flowing spins was first introduced 3 decades ago by Meuli et al⁴⁸ and advanced by Miyazaki et al.⁴⁹ Several identical approaches are available: NATIVE SPACE (NATIVE, noncontrast angiography of the arteries and veins; SPACE, sampling perfection with application-optimized contrast by using different flip angle

evolution) and TRANCE (triggered angiography noncontrast enhanced).⁵⁰ This technique uses EKG-gated 3D segmented FSE readouts, consisting of a 90° RF pulse followed by multiple repetitive 180° pulses (echo train). By adjusting the cardiac trigger delay time to take advantage of the differential arterial flow velocities in systolic and diastolic phases of the cardiac cycle or R-R interval, images with either bright or black arterial signal are acquired. During the systolic phase, the arterial blood is fast and yields no signal as the blood flows out of the imaged volume before experiencing the 180° RF refocusing pulses (Figure III in the Data Supplement). The slow venous blood and the stationary tissue contribute significant signal as their transverse magnetization is fully refocused through the train of 180° RF pulses. In contrast, during diastole, when arterial flow is low, all spins in the imaging volume are experiencing the refocusing RF echo train and contribute signal to the image (Figure III in the Data Supplement). When the systolic acquired image is subtracted from the diastolic image, the resultant image will depict only signal in the arterial compartment (Figure 3). Theoretically, fat signal should be zero after subtraction, but additional prepulse such as short tau inversion recovery or spectral attenuated inversion recovery is often used for more robust fat suppression.⁴⁹ The flow suppression effect is generally the largest in the frequency encode direction because of the large bipolar gradient for frequency encoding. Hence, the frequency-encoded direction is typically prescribed parallel to the primary direction of arterial flow.³³ The flip angle of the refocusing pulse is meant to control the degree of flow sensitivity and should be carefully adjusted. A general trade-off is that large refocusing flip angle (>160°) tends to yield more intense signal in large vessels but tends to fail to delineate small branch vessels.⁵¹ The 3D-FSE technique and its variants are relatively insensitive to B0 field variation, allowing their use with large field of view and high field strength. The limitations arise from the need for 2 separate acquisitions and properly timing the systolic and diastolic acquisitions. Moreover the degree of flow spoiling, and subtraction artifact from motion can be quite problematic in the pelvis (Table 2).

• Download figure
• Download PowerPoint

Clinical Utility and Preferred Vascular Application(s)

The relatively low sensitivity to B0 inhomogeneity allows the use of these techniques for studies of vascular districts contiguous with high susceptibility zones (eg, lungs, metallic prosthesis, and valves) or large field of view. The technique can, therefore, be used for pulmonary angiography and thoracic and abdominal aortic angiography,⁹ and it has demonstrated value in peripheral arterial imaging as well as whole body MRA.⁵² Comparison with CTA³³ and CE-MRA^30,32 in peripheral run-off studies has demonstrated high sensitivity and negative predictive value (Table 1). Limitations include visualization of small-caliber arteries, particularly of calf and foot district.

Gutzeit et al³¹ studied cardiac gated 3D FSE in the lower limb arteries using DSA as reference standard. The study demonstrated a very high negative predictive value (>95%) and high diagnostic accuracy (>90%) for both thigh and calf segments. Schubert et al³⁴ used the cardiac gated 3D FSE and CE-MRA for the evaluation of pedal vasculature in 20 PAD patients, with DSA as standard reference. The rate of nondiagnostic vessel segments was considerably higher for the NC-MRA technique. Sensitivity and negative predictive value for the NC-MRA were 96% and 97%, respectively, while specificity and positive predictive value were 80% and 69% for >50% stenosis, respectively. The performance of this technique in patients with severe PAD such as critical limb ischemia appears to be sub-optimal. Altaha et al²² compared cardiac gated 3D FSE with QISS in the evaluation of lower limb arteries in 19 patients with critical limb ischemia. Although QISS provided good to excellent image quality in 80% to 96% of arterial segments without any nondiagnostic arterial segments, 90% to 96% of segments obtained with 3D-FSE MRA were rated as nondiagnostic or poor, precluding the calculation of diagnostic performance measures for 3D-FSE MRA. Similarly, Ward et al²⁴ reported inferior image quality and specificity of cardiac gated 3D-FSE (NATIVE SPACE) as compared with QISS in a cohort of patients with chronic lower limb ischemia (Table 1). These findings might be related to the low-flow velocity in systole compared with diastole in the stenotic arterial segments that can determine the suppression of the signal in the final subtracted image (Figure 4). The TRANCE technique has also been recently implemented for the lower limb arteries imaging on open, low-field MR systems showing good image quality but underestimation of larger vessels and overestimation of smaller caliber vessels compared with DSA.⁵³

• Download figure
• Download PowerPoint

3D-FSE sequences such as SPACE or VISTA (volume isotropic turbo spin echo acquisition) can also be used without subtraction and cardiac gating, but with additional use of a double inversion pulse, allowing for isotropic 3D of the lower extremity circulation. The sequence is useful in providing dark blood images that allow visualization of the arterial wall (Figure 5). Mihai et al⁴⁶ compared SPACE to CE-MRA for the evaluation of lower limb arteries in a small group of PAD patient, showing high correlation with no significant bias in the vessel area estimation between the 2 techniques.

• Download figure
• Download PowerPoint

The relatively low sensitivity to B0 inhomogeneity and the limitations related to the low image quality in the visualization of small caliber vessels of the lower extremity make this technique useful for the evaluation of the distal abdominal aorta and iliac arteries in patients with and without metallic stent or prosthesis (Table 3).

Flow-Sensitive Dephasing Magnetization Preparation Technique

Another noncontrast approach to visualize peripheral vasculature is the 3D bSSFP using flow-sensitive dephasing (FSD) magnetization preparation, which provides high arterial blood SNR and blood-tissue CNR, isotropic submillimeter resolution, and different velocities and flow directions suppression using a flexible FSD module. The principle of this technique is similar to the spin-echo-based technique in that bright artery, and dark artery images are acquired separately for the same imaging volume, and image subtraction is used to yield artery-only images. While the sensitivity of spin-echo readout to flow is used in the former approach to suppress the appearance of arteries in systole, with this technique, an FSD prepulse is used, allowing generation of dark blood images.⁵³ The FSD prepulse consists of a 90°_x–180°_y–90°_-x sequence interleaved with bipolar gradients. After excitation by the initial 90°_x pulse, spins get dephased to a varying degree by the bipolar gradients depending on their velocities. The amount of this dephasing determines how constructively the last 90°_-x pulse works on the spins, which results in different effective flip angles ranging from 0° to 180° for spins with different velocities. The resultant longitudinal magnetization is a cosine function over spin’s velocity, where the frequency of the sinusoidal fluctuation is proportional to the size of the bipolar gradient. Then, with sufficiently large bipolar gradients, spin velocity variations within each voxel cause large intravoxel dephasing and make the net signal nearly zero. To obtain dark artery images, the FSD prepulse is played during systole using cardiac triggering, followed by the 3D bSSFP readout. To obtain bright artery images, the prepulse is played with all the bipolar gradients turned off, which becomes the same as typical T2 preparation prepulse (Figure IV and V in the Data Supplement).

The strength and duration of the bipolar gradient determine the degree of the flow sensitivity; this parameter must be carefully chosen by the user. The use of large bipolar gradients (and therefore too large flow sensitivity) may result in signal from unwanted contamination from venous blood and even stationary tissues due to the diffusion effect, whereas the use of small bipolar gradients may fail to delineate small vessels with low-flow velocity. The pros and cons of the FSD-based NC-MRA are similar to those of the spin-echo-based subtractive technique (Table 2). The use of 3D encoding allows for high spatial resolution in all 3 dimensions, but the need for 2 acquisitions increases the scan time and the risk of motion-related artifacts. A practical drawback of FSD NC-MRA is that it is not commercially available at this time.

Clinical Utility and Preferred Vascular Application(s)

FSD-prepared 3D bSSFP provides good SNR and image resolution due to the T2-preparation before FSD module but requires high B0 homogeneity. Liu et al³⁷ showed that, when compared with traditional CE-MRA in the evaluation of foot arteries, the technique has a high percentage of diagnostic arterial segments (92%), with high interobserver agreement and good values of SNR and CNR. The average sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of FSD MRA were 88%, 93%, 81%, 96%, and 92%. The technique has been used in the evaluation of the calf arteries at both 1.5T^35,36,38 and 3T^39,40 B0 field strength, demonstrating good sensitivity and negative predictive value compared with CE-MRA as reference standard and high sensitivity and positive predictive value compared with DSA (Table 1). The relatively low specificity in the evaluation of the calf arteries may be explained by flow-related artifacts due to the long acquisition times affecting patient compliance, the inhomogeneity of magnetic field at 3T, and the imprecise choice of acquisition parameters, in particular the amplitude and the direction of the bipolar gradients. Zhang et al²⁸ compared QISS and FSD MRA, using CE-MRA as standard reference. They found no significant difference in the number of diagnostic arterial segments and higher SNR and CNR values of FSD MRA than QISS. Authors showed no difference in sensitivity (95% versus 93%; P>0.05) and negative predictive value (98% versus 97%; P>0.05) between FSD and QISS for detecting stenosis >50%. FSD showed higher specificities (99% versus 92%) and diagnostic accuracy (98% versus 92%) compared with QISS. The main limitation of FSD MRA when compared with QISS is the longer scanning time, which makes this technique less useful in noncompliant patients.

Shaida³⁸ et al reported a sensitivity of 65% and a specificity of 93% for the calf district evaluation in a cohort of 24 patients. This result could be explained by the high rate of multifocal disease in PAD patients, which make the image interpretation more challenging.

Although this technique is still not commercially available, its strengths reside in the possibility to modulate the characteristics of the bipolar gradient to determine the degree of the flow sensitivity and in the sub-millimetric spatial resolution. These features make the technique suitable for evaluation of distal arterial segments (calf or foot) in patients with known history or suspected PAD (Table 3).

Velocity-Sensitive Flow-Dependent Approaches

Phase Contrast MRA

This MRA approach was originally developed starting from the idea that spins in moving protons in blood incur a net phase difference that is greater than stationary protons.^54,55 The application of a bipolar gradient to an imaging volume containing vessels will then result in stationary protons experiencing no net phase shift, in contrast to the moving protons that experience a net phase shift proportional to their velocity and direction. Bipolar gradients can be applied along one anatomic axis or along 2 or more axes simultaneously to measure flowing spin phase shift in an arbitrary direction stenosis. The net phase shift experienced by the moving protons is proportional to the velocity of blood, with the maximal phase shift occurring at a ±180° phase shift. The amplitude, duration, and spacing of the bipolar gradients are controlled by an operator-dependent parameter known as velocity encoding. When a certain velocity encoding is selected, the MR system modifies the strength of the flow encoding (bipolar) gradients so that protons moving at the desired velocity will have a full 180° phase shift. The main disadvantage of phase contrast (PC)-MRA are the long scan time and the high sensitivity to signal losses caused by turbulence and vessel tortuosity. Given that the technique does not have imaging quality information, its utility is primarily in assessment of stenosis. A different approach to PC technique is the combination of 3D spatial encoding, 3-directional velocity encoding and cine acquisition, proposed by Wigström in 1996.⁵⁶ This 4D PC approach allows time-resolved flow and depiction of temporal evolution of flow patterns in an anatomic region.⁵⁷ Due to long scan times and relatively low spatial resolution, the applications of the technique in the study of the lower limbs’ arteries are currently limited (Table 2). Four-dimensional technique can be used for in vivo evaluation of atherosclerotic plaques in large vessels.⁵⁸

Velocity-Selective MRA

Velocity-selective MRA (VS-MRA) has been recently introduced as another promising noncontrast-enhanced angiography method.^59,60 The key component of this method is a velocity-selective magnetization preparation pulse which modulates each magnetic spin as an explicit and flexible function of its velocity, but without regards to its spatial location.⁶¹ This prepulse may sound similar to the FSD prepulse in that both excite spins in a flow-sensitive fashion. The important difference, however, is that the VS prepulse works by suppressing the background tissues and slow venous blood while preserving arterial blood signal and, thus, generating positive arterial contrast directly in a single acquisition. However, the FSD prepulse can only suppress flowing spins, thereby requiring acquisition of reference (bright artery) image and subtraction of 2 images. The initial version of VS preparation pulse consists of multiple hard RF pulses of small flip angle, interleaved with repeated bipolar gradients. The pulse sequence for VS-MRA is cardiac gated and plays the VS prepulse near the time of peak systolic flow by adjusting a trigger delay typically based on a prior PC flow measurement. For a segmented 3D acquisition that follows the VS prepulse, balanced SSFP readouts are preferred to further enhance arterial signal, while gradient-echo readout can also be used to avoid potential banding artifacts at high fields.

Due to the use of 3D encoding, VS-MRA can achieve high spatial resolution and large field of view in all three dimensions, similarly to the 3D subtractive methods based on FSE readout or FSD prepulse. In addition, VS-MRA requires only one acquisition, which makes it less time-consuming and with less potential for motion-induced artifacts compared with the other subtractive methods. Moreover, because the velocity sensitization of VS-MRA happens during systolic phases only, this technique is not sensitive to cardiac arrhythmia, which alters the diastole significantly but the systole minimally.^62,63 Despite these compelling features, the effects of B0 and B1 inhomogeneity may manifest as arterial signal loss and stripe artifacts particularly at high field (Table 2). Relevant recent studies include application of optimized composite pulse to enhance the immunity to the field offsets and alternate application of phase-cycled VS prepulses to remove the stripe artifact.^64,65

Clinical Utility and Preferred Vascular Application(s)

The first clinical testing of VS-MRA in a cohort of PAD patients was recently reported with DSA as the reference standard.⁴¹ The diagnostic sensitivity for detecting significant stenosis was 79.4% (averaged over 3 independent reviewers) when calculated on a per-segment basis and 92.1% when calculated on a per-region basis. The specificity was 95.1% on a per-segment basis and 94.0% on a per-region basis. The average of image quality score was 2.77 in the scale of 0 (nondiagnostic) to 3 (excellent). Figure 6 illustrates representative VS-MRA and DSA images in 3 PAD patients with moderate, advanced, and severe stenoses, respectively. While this initial test demonstrates the potential of VS-MRA, its clinical value needs to be further verified through comparisons with gadolinium-enhanced MRA and other nongadolinium-enhanced techniques.

• Download figure
• Download PowerPoint

Despite the need of further clinical verifications, the low sensitivity to cardiac arrhythmia and the high sensitivity to the effects of B0 and B1 inhomogeneity may render this sequence suitable for the evaluation of lower limb arteries in arterial districts not affected by sources of magnetic field inhomogeneity (air, metallic stets, and prosthesis) such as femoral or calf arteries in patients with knows history of cardiac disease (Table 3).

Future Directions

There are several new NC-MRA techniques that appear promising and may serve a role in the future.^66,67

Arterial spin labeling technique relies on the application of a tagging pulse applied to a slab of tissue upstream from the imaging volume that inverts the magnetization of water molecules. A control image is acquired on the selected region before the application of the tagging pulse, and a new image is acquired after the application of the tagging pulse in the selected region of interest to measure the signal difference of the tagged blood flow from the untagged image. With this approach, the evaluation of lower limbs perfusion is possible.^68,69 Blood oxygenation level–dependent imaging relies on the evaluation of regional differences in blood flow features. When a region increases its perfusion, the extraction of oxygen from the local vascular bed determines an initial drop in oxygenated hemoglobin and an increase in local carbon dioxide and deoxygenated hemoglobin levels. The ability to detect the difference between the amount of these products resides in the different magnetic properties of oxygenated and deoxygenated hemoglobin. Deoxygenated hemoglobin is paramagnetic, whereas oxygenated hemoglobin is not, so the former will cause local dephasing of protons, reducing signal intensity from imaged tissues. Lower limb perfusion in patients with PAD has been studied with this approach,^70–72 which has been extensively used for brain functional imaging.⁷³ Atherosclerotic plaque imaging is an important challenge for magnetic resonance imaging technology because plaque composition influences the subsequent risk for vascular occlusion.⁶⁷ Several approaches have been implemented and are under investigation for atherosclerotic plaques imaging: (1) susceptibility imaging aiming to study plaque composition taking advantage of the different magnetic properties of plaque components (eg, calcification and intraplaque hemorrhage lead to increased magnetic susceptibility). Highly susceptive components have short T2*, which can be measured using multi-echo imaging techniques. (2) Plaque components, such as lipoproteins and collagen, may have short T2 values to be directly imaged with standard FSE techniques. Ultra-short echo time and magnetization transfer imaging techniques have been introduced to obtain signal from tissues with this characteristic. (3) Microscopic structure of the plaque components can influence the magnetic resonance imaging signal on diffusion-weighted imaging. Regions with smaller diffusion coefficients undergo less dephasing and appear brighter on the resulting diffusion-weighted imaging image (eg, necrotic core regions show lower diffusion coefficients than fibrous tissue). In the future, multiple approaches could provide stenoses severity and phenotypic characterization of lower extremity plaque and could leverage machine learning and artificial intelligence approaches that may be warranted given the need for accurate and reproducible data extraction and comparisons.

Conclusions

The application of NC-MRA techniques has seen a recent resurgence of interest in patients with PAD. In contrast to older techniques, such as TOF and PC-MRA, newer approaches such as QISS, 3D-FSE, and VS-MRA provide superior ability especially in patients with PAD and are theoretically an attractive alternative to CTA or DSA and may provide a practical platform to assess PAD patients.

Footnotes