Unenhanced MR Angiography of the Foot: Initial Experience of Using Flow-Sensitive Dephasing–prepared Steady-State Free Precession in Patients with Diabetes

Unenhanced MR angiography performed with flow-sensitive dephasing–prepared steady-state free precession enables clear depiction of the foot arterial tree and accurate detection of significant arterial stenosis.

Abstract

Purpose

To assess image quality and diagnostic performance of unenhanced magnetic resonance (MR) angiography with use of flow-sensitive dephasing (FSD)–prepared steady-state free precession (SSFP) of the foot arteries in patients with diabetes.

Materials and Methods

This prospective study was approved by institutional review board. Informed consent was obtained from all subjects. Thirty-two healthy volunteers and 38 diabetic patients who had been scheduled for lower-extremity contrast material–enhanced MR angiography were recruited to undergo unenhanced MR angiography with a 1.5-T MR unit. Image quality and diagnostic accuracy of unenhanced MR angiography in the detection of significant arterial stenosis (≥50%) were assessed by two independent reviewers. Contrast-enhanced MR angiography was used as the reference standard. The difference in the percentage of diagnostic arterial segments at unenhanced MR angiography between healthy volunteers and diabetic patients was evaluated with the McNemar test and generalized estimating equation for correlated data. Signal-to-noise ratio (SNR) and artery-to-muscle contrast-to-noise ratio (CNR) of pedal arteries were measured and compared between the two MR angiography techniques by using the paired t test.

Results

All subjects successfully underwent unenhanced MR angiography of the foot. Unenhanced MR angiography yielded a high percentage of diagnostic arterial segments in both healthy volunteers (303 of 320 segments, 95%) and patients (341 of 370 segments, 92%), and there was no difference in the percentage between the two populations (P = .195). In patients, the average SNR and CNR at unenhanced MR angiography were higher than those at contrast-enhanced MR angiography (SNR: 90.7 ± 38.1 vs 81.7 ± 34.7, respectively, P = .023; CNR: 85.2 ± 33.2 vs 76.6 ± 33.5, respectively, P = .013). The average sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of unenhanced MR angiography were 88% (35 of 40 segments), 93% (107 of 115 segments), 81% (35 of 43 segments), 96% (107 of 112 segments), and 92% (142 of 155 segments), respectively. Interobserver agreement between the two readers for diagnostic accuracy was good (κ = 0.83).

Conclusion

Unenhanced MR angiography with use of FSD-prepared SSFP allows clear depiction of the foot arterial tree and accurate detection of significant arterial stenosis. The technique has the potential to be a safe and reliable screening tool for the assessment of foot arteries in diabetic patients without the use of gadolinium-based contrast material.

© RSNA, 2014

Introduction

Foot complications associated with diabetes have a prevalence of 15%–20% in all patients with diabetes and remain the most common cause of nontraumatic amputations of the lower extremity (1). Previous studies have shown that 45%–85% of patients can be spared amputation if appropriate revascularization is performed on the basis of a reliable early diagnosis (2,3). Surgical and endovascular revascularization remains the most important therapeutic option for limb salvage in patients with severe arterial occlusive disease, particularly those with diabetes (4,5). Because the outcome of the therapy is in large part determined by the presence of adequate pedal blood flow, precise preoperative mapping of the vascular anatomy and the location, extent, and degree of stenosis is crucial for planning stent placement or selecting a site of graft anastomosis (6–8).

Contrast material–enhanced magnetic resonance (MR) angiography has been used as a routine procedure for the imaging of extremity arteries and offers advantages over computed tomographic angiography in that patients are not exposed to ionizing radiation or potentially nephrotoxic iodinated contrast material (9). However, with a conventional bolus-chase multistation technique, it is a challenge for contrast-enhanced MR angiography to depict pedal arteries without venous contamination owing to the short contrast material first-pass window in the arteries (10). Recently, a time-resolved contrast-enhanced MR angiography technique has been developed to prevent venous contamination in the foot (11); however, multiple injections of contrast material are required if it is necessary to image more regions of the lower extremities (12–15). In addition, because of the potential risk of nephrogenic systemic fibrosis, the use of gadolinium-based MR imaging contrast material should be avoided in patients with severe renal insufficiency (16), which is more prevalent in patients with diabetes (17,18). More recently, a group of unenhanced MR angiography techniques, such as fresh blood imaging (19–21) and quiescent-interval single-shot imaging (22,23), have been developed as an alternative to contrast-enhanced MR angiography for peripheral MR angiography. Their applicability has been successfully demonstrated in the major lower-extremity vasculatures but not in the pedal arteries, which pose particular challenges. These include the small caliber of the vessels, the presence of slower flow, and tortuous anatomy. An unenhanced MR angiography technique with use of steady-state free precession (SSFP) and flow-sensitive dephasing (FSD) magnetization preparation has been shown to allow good image quality in tortuous and small distal extremity arteries and to demonstrate clinical potential in the depiction of calf and hand arteries (24–27). To our knowledge, however, no previous studies have been performed to evaluate unenhanced MR angiography in the foot. The purpose of this study was to assess image quality and diagnostic performance of unenhanced MR angiography with use of FSD-prepared SSFP of the foot arteries in patients with diabetes.

Materials and Methods

Subjects

This prospective study was approved by the local institutional review board. Written informed consent was obtained from all participants. All MR angiography studies were performed in a single medical center from May 2010 to December 2011. Subjects were composed of 32 healthy volunteers aged 24–72 years (mean age, 43 years; 20 men aged 25–72 years [mean age, 45 years] and 12 women aged 24–68 years [mean age, 41 years]) without a history of peripheral vascular disease and 38 consecutive diabetic patients aged 34–79 years (mean age, 58 years; 22 men aged 39–79 years [mean age, 59 years] and 16 women aged 34–77 years [mean age, 56 years]) who were referred for conventional contrast-enhanced MR angiography of the lower extremities. All patients had type 2 diabetes mellitus, which was diagnosed according to 2006 World Health Organization diabetes criteria (28). Eleven of the 38 diabetic patients (29%) had diabetic foot ulcer, 22 (58%) had diabetic retinopathy, 23 (60%) had diabetic neuropathy, 20 (53%) had hypertension, two (5%) had undergone amputation of one leg, and 20 (53%) had kidney damage (glomerular filtration rate, <60 mL/min). Fifteen patients had intermittent claudication, seven complained of rest pain, and 11 had diabetic foot ulcer. The mean ankle-brachial index (ABI) was 0.7 (range, 0.3–1.6). Characteristics of the 38 patients are detailed in Table 1. The exclusion criteria for the MR angiography study were general contraindications to MR examination (claustrophobia, pacemaker, aneurysm clip, and intraauricular metallic implants) and severe renal failure (glomerular filtration rate, <30 mL/min) according to the hospital policy for the use of gadolinium-based contrast material. Ten patients were excluded from the study owing to severe renal failure (n = 4), claustrophobia (n = 3), and pacemaker (n = 3).

Table 1.

Characteristics of the 38 Patients with Diabetes

Note.—Except where indicated, data are numbers of patients, with percentages in parentheses.

^*Numbers in parentheses are the range.

MR Angiography

All examinations were performed with a 1.5-T MR unit (Magnetom Avanto; Siemens Healthcare, Erlangen, Germany). Unenhanced MR angiography was performed at the foot by using a 12-element phased-array head coil for both patients and volunteers. Contrast-enhanced MR angiography was performed only in patients, with use of the same head coil for the foot and two six-element phased-array body coils for the pelvis, thigh, and calf. All subjects were placed in the supine position, and the feet were positioned in slight plantar flexion and immobilized with foam padding to prevent movement. Unenhanced MR angiography was performed before contrast-enhanced MR angiography.

Unenhanced MR angiography was performed by using electrocardiographically triggered three-dimensional segmented SSFP coupled with FSD magnetization preparation (24). The preparation consists of a 90°_x-180°_y-90°_−x radiofrequency pulse series with balanced gradient pulses (ie, FSD gradients) straddling the center 180° pulse. Such a sequence module is capable of generating flow sensitivity (measured by means of the first-order gradient moment, a quantity derived from the FSD gradient duration and strength) and suppressing flow signals on the basis of intravoxel dephasing among flowing spins that have large intravoxel velocity variability. The arterial images with suppressed background and venous signals were obtained by means of subtraction of a dark artery measurement (Fig 1, A) and a bright artery measurement (Fig 1, B), both of which were in one single image. The bright artery and bright vein data set was acquired by applying the T2 preparation during mid-diastole when arterial flow is substantially slow as the venous flow. The dark artery and bright vein data set was acquired during systole by using the FSD magnetization preparation with an appropriate first-order gradient moment to suppress the signal of fast-flowing arterial blood while causing minimal effects on the venous blood and static tissues. The imaging parameters were as follows: repetition time msec/echo time msec = 3.8/1.9, receiver bandwidth = 965 Hz/pixel, field of view = 300 × 200 × approximately 80–90 mm³, voxel size = 0.9 × 0.9 × 0.9 mm³ without interpolation, flip angle = 90°, generalized autocalibrating partially parallel acquisition acceleration factor = two in the phase-encoding direction, 60 lines per heartbeat, acquisition time = approximately 3–4 minutes per image (depending on heart rate) including two consecutive measurements (ie, bright artery and dark artery), oblique coronal acquisition with the readout direction coinciding with the principal arterial flow of the foot artery (ie, the lateral plantar artery), and FSD gradients applied along the readout direction only. Simultaneous applications of FSD gradients in all three logic directions would render the flow sensitivity limited to the direction of the sum vector among the three directions. This could make the flow suppression ineffective in some arterial segments that are perpendicular to the sum vector, as suggested in previous work (26). To achieve sufficient signal suppression in the arterial blood but little in the venous blood, the first-order gradient moment value of the FSD gradient was individually optimized in a scout image (approximately 100–150 mT ⋅ msec²/m) by using a previously proposed approach (29).

Figure 1:

Pulse sequence diagrams of, A, dark artery imaging with FSD-prepared (FSD_prep) SSFP triggered at systole and, B, bright artery imaging with T2-prepared (T_2prep) SSFP triggered at diastole. Both FSD preparation and T2 preparation consist of a 90°_x-180°_y-90°_−x radiofrequency (RF) series and are of the same duration (ie, T2 preparation) to maintain the same T2 weighting for venous blood and static tissues. FSD gradient pulses applied in readout direction are switched on in FSD preparation but switched off in T2 preparation. The readout direction coincides with main blood flow direction. FSD or T2 preparation is followed by a spectrally selective fat saturation radiofrequency module and 10 preparation radiofrequency pulses with linearly increasing flip angles before SSFP data acquisition. G_RO = readout gradient.

Contrast-enhanced MR angiography was performed with the bolus-chase three-station technique (10) from the anterosuperior iliac spine to the feet on the basis of the institutional contrast-enhanced MR angiography protocol for the lower extremities. A three-dimensional gradient-echo fast low-angle shot pulse sequence was used for data acquisition. Imaging parameters were as follows: 1.14/3.3, flip angle = 25°, generalized autocalibrating partially parallel acquisition factor = two, field of view = 320 × 320 × 96 mm³, base matrix = 284 × 284, phase resolution = 100%, 80 partitions, and section resolution = 60%, giving a nearly isotropic voxel size of 1.1 × 1.1 × 1.2 mm³. Contrast-enhanced acquisition was initiated when the contrast material (gadopentetate dimeglumine [Magnevist; Schering, Berlin, Germany]) arrived at the iliac artery as detected with the Care Bolus technique (Siemens Healthcare). The contrast material dose was 0.2 mmol per kilogram body weight. The injection flow rate (2–2.5 mL/sec) was adjusted to cover 70% of the total acquisition time. The acquisition time for one data set was 18 seconds. The table was moved automatically (30 cm) to the next table position, increasing the total acquisition time to 64 seconds.

Image Analysis

All MR angiograms of the foot were transferred to a workstation (Leonardo; Siemens Medical Solutions, Erlangen, Germany) for postprocessing. Maximum intensity projection (MIP) images were created by a radiologist (X.L.) with 10 years of experience in cardiovascular radiology by using standardized postprocessing procedures for both contrast-enhanced and unenhanced MR angiography.

The quality of the MIP images was evaluated independently by two radiologists (P.L. [reader 1] and Q.Y. [reader 2], with 10 years of experience in interventional radiology and 6 years of experience in cardiovascular imaging, respectively). The readers were not involved in performing the examinations, and both readers were unaware of other clinical test information. A four-point scale was used to rate the image quality for five pedal arterial segments, including the dorsal pedal artery, medial plantar artery, lateral plantar artery, pedal arch, and metatarsal arteries, as follows: score 4, excellent diagnostic arterial display with sharp delineation of pedal arteries in full length and no minor venous contamination or soft-tissue artifacts; score 3, good diagnostic arterial display with good delineation of the vessel structures and no or minor venous contamination or soft-tissue artifacts; score 2, fair diagnostic arterial display with moderate delineation of the arterial vasculature with medium soft-tissue artifacts or venous contamination and detection of lesions still possible; and score 1, poor or nondiagnostic arterial display due to severe soft-tissue artifacts or no exact arterial delineation. An arterial segment with an image quality score of 2 or higher was classified as diagnostic.

The signal-to-noise ratio (SNR) of the artery and the artery-to-muscle contrast-to-noise ratio (CNR) were assessed from the original unenhanced and contrast-enhanced MR angiograms in three main foot arteries (dorsal artery, lateral plantar artery, and pedal arch) with diagnostic image quality (score, ≥2). Signal intensity measurements for the blood were performed by one author (N.Z., with 5 years of experience in MR imaging research) by using circular regions of interest (area, approximately 0.08–0.1 cm²) placed within the center of cross section at the middle portion of the arterial segments, defined as a region with the highest but homogeneous signal intensity (30). The signal intensity of muscle was measured from the adjacent muscle tissue immediately next to the arteries (region of interest area is the same as that for the vessel). Because of the difficulty in measuring noise in parallel imaging techniques with inhomogeneous noise distribution, the mean of the signal intensity standard deviations from four larger regions of interest (area, approximately 0.5–1.2 cm²) in airspace free of image artifacts and close to the site of signal intensity measurement were computed as noise (31). Blood SNR was defined as the mean signal intensity in the arterial segments divided by the noise. Artery-to-muscle CNR was defined as the difference between the mean signal intensity in the arterial segments and that in the surrounding muscle tissue divided by the noise.

The severity of arterial stenosis was evaluated on MIP images visually by the two readers independently as less than 50% luminal narrowing, including no narrowing, or at least 50% luminal narrowing, including occlusion in the three main pedal arteries (dorsal pedal artery, lateral plantar artery, and pedal arch). Stenosis of at least 50% luminal narrowing or occlusion was considered hemodynamically significant. The readers were allowed to create additional MIP images or check source images for the arteries if a significant stenosis was suspected. To reduce the potential memory effect, the paired unenhanced and contrast-enhanced MR angiograms were assessed in a random order, with an interval of 4 weeks in between. In patients with lesions at multiple sites, only the site with the most severe lesion was scored.

Statistical Analysis

A McNemar test was performed to evaluate the difference in the percentage of diagnostic arterial segments at unenhanced MR angiography between healthy volunteers and diabetic patients. The generalized estimating equation was applied to model diagnosis judgment with independent variables of healthy volunteers and diabetic patients, two independent readers, and the correlation between five arterial segments per subject. In this generalized estimating equation model, the working correlation matrix was defined as autoregression first order, and the link function was defined as logit for binary distribution cluster. In the comparison of quantitative measurements, the independent and paired t tests were used to assess the difference in the SNR and CNR between unenhanced and contrast-enhanced MR angiography. Two-tailed P ≤ .05 was considered to indicate a significant difference.

The diagnostic sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of unenhanced MR angiography (only for arterial segments that were considered diagnostic at contrast-enhanced MR angiography) in the detection of significant stenosis were calculated on a segmental basis by using contrast-enhanced MR angiography as the standard of reference. An intention-to-diagnose approach was used to calculate diagnostic accuracy, and nondiagnostic segments at unenhanced MR angiography were considered as negative (32). The Cohen κ statistic was used to evaluate interobserver agreement in the assessment of significant arterial stenosis at unenhanced MR angiography.

A statistical software package (SPSS 17.0; SPSS, Chicago, Ill) was used for all statistical analyses in the study.

Results

We evaluated 64 feet in the 32 healthy volunteers and 74 feet in the 38 diabetic patients (two patients had undergone amputation of one leg). All subjects successfully underwent unenhanced MR angiography of the foot. The number of diagnostic arterial segments for each of the five foot arteries at unenhanced MR angiography is shown in Table 2. For all arterial segments, unenhanced MR angiography showed a high percentage of diagnostic segments in both healthy volunteers (303 of 320 segments, 95%) and patients (341 of 370 segments, 92%), and there was no significant difference in the percentage between the two populations (P = .195). Of the total 46 nondiagnostic segments (dorsal pedal artery in two segments, medial plantar artery in 12, lateral plantar artery in 12, pedal arch in 16, and metatarsal arteries in four), 22 segments (48%) were affected by severe soft-tissue artifacts (presumably due to motion), 16 (35%) were affected by poor arterial blood SNR, and eight (17%) were affected by severe venous contamination.

Table 2.

Number of Diagnostic Arterial Segments at Unenhanced MR Angiography of the Foot in 32 Healthy Volunteers and 38 Patients with Diabetes

Note.—All data are the average of the two readers. Numbers in parentheses are raw data (numbers of feet). A total of 74 feet were evaluated in 38 diabetic patients because two patients had undergone amputation of one leg. An arterial segment with an image quality score of at least 2 (delineation of the arterial vasculature with medium artifacts of soft tissue or venous contamination) was defined as a diagnostic arterial segment.

For the generalized estimating equation analysis, there are 70 clusters (32 healthy volunteers and 38 diabetic patients). Each reader has a judgment to five arterial segments on both feet for each subject. The results show that there is no significant correlation between healthy volunteers and diabetic patients (P = .34), the two independent readers (P = .30), and five arterial segments for each subject (P = .16). The results are shown in Table 3.

Table 3.

Results of Generalized Estimating Equation Analysis

Note.—Subject group is the category for healthy volunteers and diabetics patients. P < .05 indicates significant difference.

For the quantitative measurements, the average SNR of the artery and the artery-to-muscle CNR at unenhanced MR angiography were higher than those at contrast-enhanced MR angiography (SNR: 90.7 ± 38.1 vs 81.7 ± 34.7, respectively, P = .023; CNR: 85.2 ± 33.2 vs 76.6 ± 33.5, respectively, P = .013). At the dorsal artery and lateral plantar artery, however, there was no difference in the SNR or CNR between unenhanced MR angiography and contrast-enhanced MR angiography. The details of the average SNR and CNR for each of the three main pedal arteries are shown in Table 4.

Table 4.

Comparison of SNR and CNR of Diagnostic Arterial Segments at Unenhanced and Contrast-enhanced MR Angiography in 38 Patients with Diabetes

Note.—SNR and CNR data refer only to the three main foot arteries (dorsal artery, lateral plantar artery, and pedal arch) with diagnostic image quality at unenhanced MR angiography (n = 205) and contrast-enhanced MR angiography (n = 155).

Of the 222 main pedal arteries evaluated in diabetic patients, 155 (70%) were classified as diagnostic on contrast-enhanced MR angiograms. Of these 155 arterial segments, 40 were identified as having significant stenosis by both readers on the basis of consensus. Reader 1 identified 35 of the 40 stenoses (88%) at unenhanced MR angiography and reader 2 identified 36 (90%). With use of contrast-enhanced MR angiography as the standard of reference, there were nine false-negative findings. Eight false-negative findings occurred at the dorsal artery, and one false-negative finding occurred at the lateral plantar artery. In the remaining 115 arterial segments that were either normal or had insignificant stenosis at contrast-enhanced MR angiography, 109 (95%) and 106 (92%) segments were correctly identified by reader 1 and reader 2, respectively, at unenhanced MR angiography. There were 15 false-positive findings, eight in the lateral plantar artery and seven in the pedal arch. The average sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of the two readers for unenhanced MR angiography were 88% (35 of 40 segments), 93% (107 of 115 segments), 81% (35 of 43 segments), 96% (107 of 112 segments), and 92% (142 of 155 segments), respectively. Interobserver agreement between the two readers for diagnostic accuracy was good (κ = 0.83).

Figures 2–5 illustrate results from four patient studies.

Figure 2a:

Example source images obtained in 39-year-old man with diabetes and diabetic neuropathy for 5 years. Mean ABI was 1.2. (a, b) Spatially matched source images acquired in dark artery (a) and bright artery (b) measurements. (c) Corresponding subtraction image. In this case, prescribed FSD gradients (or first-order gradient moment) are sufficiently sensitive to both large (arrows in a–c) and small (arrowhead in a–c) arterial segments, providing consistent flow suppression effects in them. (d) Subtraction MIP MR angiogram enables delineation of all relevant arterial segments. However, venous blood is not suppressed by FSD owing to substantially slow flow velocity and thus is not depicted on image. DA = dorsal artery, LPA = lateral plantar artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Figure 5a:

MIP images from (a) unenhanced and (b) contrast-enhanced MR angiography of the foot in 67-year-old man with diabetes, diabetic retinopathy and nephropathy, intermittent claudication, and foot ulcer on top of right second toe. ABI was 0.4. On unenhanced image, occlusions of dorsal artery are present in right foot, and peroneal arteries provide collateral circulation distal to occluded dorsal arteries in both feet (arrowheads). There are local arterial stenoses at distal dorsal artery in left foot (arrows). All findings at unenhanced MR angiography are consistent with those at contrast-enhanced MR angiography. Soft-tissue artifacts are present on unenhanced image but have no negative effect on delineation of arteries, which have much higher SNR than artifacts. Spatial resolution of unenhanced and contrast-enhanced MR angiography was 0.9 × 0.9 × 0.9 mm³ and 1.0 × 1.0 × 2.0 mm³, respectively. Arch = pedal arch, DA = dorsal pedal artery, LPA = lateral plantar artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Figure 2b:

Example source images obtained in 39-year-old man with diabetes and diabetic neuropathy for 5 years. Mean ABI was 1.2. (a, b) Spatially matched source images acquired in dark artery (a) and bright artery (b) measurements. (c) Corresponding subtraction image. In this case, prescribed FSD gradients (or first-order gradient moment) are sufficiently sensitive to both large (arrows in a–c) and small (arrowhead in a–c) arterial segments, providing consistent flow suppression effects in them. (d) Subtraction MIP MR angiogram enables delineation of all relevant arterial segments. However, venous blood is not suppressed by FSD owing to substantially slow flow velocity and thus is not depicted on image. DA = dorsal artery, LPA = lateral plantar artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Figure 2c:

Example source images obtained in 39-year-old man with diabetes and diabetic neuropathy for 5 years. Mean ABI was 1.2. (a, b) Spatially matched source images acquired in dark artery (a) and bright artery (b) measurements. (c) Corresponding subtraction image. In this case, prescribed FSD gradients (or first-order gradient moment) are sufficiently sensitive to both large (arrows in a–c) and small (arrowhead in a–c) arterial segments, providing consistent flow suppression effects in them. (d) Subtraction MIP MR angiogram enables delineation of all relevant arterial segments. However, venous blood is not suppressed by FSD owing to substantially slow flow velocity and thus is not depicted on image. DA = dorsal artery, LPA = lateral plantar artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Figure 2d:

Example source images obtained in 39-year-old man with diabetes and diabetic neuropathy for 5 years. Mean ABI was 1.2. (a, b) Spatially matched source images acquired in dark artery (a) and bright artery (b) measurements. (c) Corresponding subtraction image. In this case, prescribed FSD gradients (or first-order gradient moment) are sufficiently sensitive to both large (arrows in a–c) and small (arrowhead in a–c) arterial segments, providing consistent flow suppression effects in them. (d) Subtraction MIP MR angiogram enables delineation of all relevant arterial segments. However, venous blood is not suppressed by FSD owing to substantially slow flow velocity and thus is not depicted on image. DA = dorsal artery, LPA = lateral plantar artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Figure 3a:

MIP images from (a) unenhanced and (b) contrast-enhanced MR angiography of the foot in 51-year-old man with diabetes and diabetic neuropathy. ABI was 0.6. Unenhanced MR angiography enables clear delineation of entire foot arterial trees with good image quality. All arterial segments are normal, which is consistent with findings on contrast-enhanced MR angiogram. Compared with contrast-enhanced MR angiography, unenhanced MR angiography depicts more distal arterial branches. Spatial resolution of unenhanced MR angiography and contrast-enhanced MR angiography was 0.9 × 0.9 × 0.9 mm³ and 1.1 × 1.1 × 2.0 mm³, respectively. Arch = pedal arch, ATA = anterior tibial artery, DA = dorsal pedal artery, DMA = dorsal metatarsal artery, LPA = lateral plantar artery, LTA = lateral tarsal artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Figure 3b:

MIP images from (a) unenhanced and (b) contrast-enhanced MR angiography of the foot in 51-year-old man with diabetes and diabetic neuropathy. ABI was 0.6. Unenhanced MR angiography enables clear delineation of entire foot arterial trees with good image quality. All arterial segments are normal, which is consistent with findings on contrast-enhanced MR angiogram. Compared with contrast-enhanced MR angiography, unenhanced MR angiography depicts more distal arterial branches. Spatial resolution of unenhanced MR angiography and contrast-enhanced MR angiography was 0.9 × 0.9 × 0.9 mm³ and 1.1 × 1.1 × 2.0 mm³, respectively. Arch = pedal arch, ATA = anterior tibial artery, DA = dorsal pedal artery, DMA = dorsal metatarsal artery, LPA = lateral plantar artery, LTA = lateral tarsal artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Figure 4a:

MIP images from (a) unenhanced and (b) contrast-enhanced MR angiography of the foot in 71-year-old woman with diabetes, diabetic nephropathy, intermittent claudication, and ulceration in first toe of left foot. ABI was 0.7. Unenhanced MR angiogram shows stenosis at left distal anterior tibial artery (arrowhead in a), which is consistent with findings on contrast-enhanced MR angiogram at corresponding site of artery (arrowhead in b), and diffuse severe luminal narrowing in left lateral plantar artery. Peroneal arteries provide collateral circulation distal to occluded lateral plantar artery in left foot. Contrast-enhanced MR angiogram does not depict diffuse lesion in lateral plantar artery. Spatial resolution of unenhanced MR angiography and contrast-enhanced MR angiography was 0.9 × 0.9 × 0.9 mm³ and 1.1 × 1.1 × 2.0 mm³, respectively. Arch = pedal arch, ATA = anterior tibial artery, DA = dorsal pedal artery, DMA = dorsal metatarsal artery, LPA = lateral plantar artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Figure 4b:

MIP images from (a) unenhanced and (b) contrast-enhanced MR angiography of the foot in 71-year-old woman with diabetes, diabetic nephropathy, intermittent claudication, and ulceration in first toe of left foot. ABI was 0.7. Unenhanced MR angiogram shows stenosis at left distal anterior tibial artery (arrowhead in a), which is consistent with findings on contrast-enhanced MR angiogram at corresponding site of artery (arrowhead in b), and diffuse severe luminal narrowing in left lateral plantar artery. Peroneal arteries provide collateral circulation distal to occluded lateral plantar artery in left foot. Contrast-enhanced MR angiogram does not depict diffuse lesion in lateral plantar artery. Spatial resolution of unenhanced MR angiography and contrast-enhanced MR angiography was 0.9 × 0.9 × 0.9 mm³ and 1.1 × 1.1 × 2.0 mm³, respectively. Arch = pedal arch, ATA = anterior tibial artery, DA = dorsal pedal artery, DMA = dorsal metatarsal artery, LPA = lateral plantar artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Figure 5b:

MIP images from (a) unenhanced and (b) contrast-enhanced MR angiography of the foot in 67-year-old man with diabetes, diabetic retinopathy and nephropathy, intermittent claudication, and foot ulcer on top of right second toe. ABI was 0.4. On unenhanced image, occlusions of dorsal artery are present in right foot, and peroneal arteries provide collateral circulation distal to occluded dorsal arteries in both feet (arrowheads). There are local arterial stenoses at distal dorsal artery in left foot (arrows). All findings at unenhanced MR angiography are consistent with those at contrast-enhanced MR angiography. Soft-tissue artifacts are present on unenhanced image but have no negative effect on delineation of arteries, which have much higher SNR than artifacts. Spatial resolution of unenhanced and contrast-enhanced MR angiography was 0.9 × 0.9 × 0.9 mm³ and 1.0 × 1.0 × 2.0 mm³, respectively. Arch = pedal arch, DA = dorsal pedal artery, LPA = lateral plantar artery, MPA = medial plantar artery, PTA = posterior tibial artery.

Discussion

Our study demonstrated that unenhanced MR angiography with use of FSD-prepared SSFP enabled complete depiction of the foot arterial tree within a clinically acceptable imaging time of 3–4 minutes. All five foot arteries had a high percentage of diagnostic segments and excellent image quality in both healthy volunteers and diabetic patients. This was especially true in the two main foot arteries—the dorsal artery and the lateral plantar artery. The average percentage of diagnostic segments in the two arteries at unenhanced MR angiography was more than 95%, which is similar to results reported in previous studies with use of time-resolved or dedicated single-station contrast-enhanced MR angiography of the foot (7,12,33). Meanwhile, although the flow of the pedal arteries may be impaired in the context of diabetes, the unenhanced MR angiography technique yielded a similar percentage of diagnostic arterial segments in both healthy volunteers and diabetic patients (303 of 320 segments [95%] vs 341 of 370 segments [92%], P = .195), which indicates the reliability of this technique in the depiction of complex arterial flow.

In our study, the pedal arch showed relatively lower image quality, SNR, and CNR at unenhanced MR angiography compared with the dorsal artery and lateral plantar artery. The flow direction, in addition to the slower speed of blood flow, may be the cause of the reduced contrast and lower image quality in the segment of the pedal arch. This is likely due to the reduced flow signal suppression effect of the FSD preparation in the segment that is not in parallel with the FSD gradient direction. The use of a multidirectional FSD preparation might help alleviate this issue by using FSD gradients in at least the readout and partition phase-encoding directions (26). Because the pedal arch is an important anastomotic pathway between the two main foot arteries and the patency of the arterial arch is critically important for preoperative planning, further technical optimization for the improvement of image quality in the pedal arch is needed.

In our study, a substantial number of arterial segments (48%) were considered nondiagnostic owing to soft-tissue artifacts. A possible reason for this could be the slight spatial shift of tissues between the consecutive acquisitions for the subtraction due to either small bulk motion or drift in the magnetic field over time. Nevertheless, 644 of the 690 arterial segments (93.3%) were still diagnostic in the study.

Without the time constraint of contrast material washout, the imaging time of unenhanced MR angiography would be mainly determined according to the isotropic spatial resolution needed. This is especially important for arteries in the feet, as they are both small and tortuous and would require spatial resolution in the submillimeter range. Isotropic spatial resolution of 0.9 × 0.9 × 0.9 mm³ was achieved in our study and greatly improved depiction of pedal arteries and the ability to detect arterial stenosis. Comparable spatial resolution was achieved in a previous study in which a blood pool contrast agent was used to provide a longer imaging time window (34). More recently, Haider et al (11) described a time-resolved contrast-enhanced MR angiography technique that provides 0.75 × 0.75 × 0.90-mm³ spatial resolution for the imaging of foot arteries.

High SNR is also important for depicting foot arteries. In our study, SSFP was used to acquire arterial blood signal. This pulse sequence was ideally suited for MR angiography as it is fast, SNR efficient, and less sensitive to fast flow and complex flow patterns (35,36). Despite the known limitations regarding signal intensity measurements on images acquired with use of a parallel imaging technique, the overall SNR and CNR of the three main foot arteries (dorsal artery, lateral plantar artery, and pedal arch) at unenhanced MR angiography proved to be higher than that at contrast-enhanced MR angiography in our study and that at time-resolved contrast-enhanced MR angiography in a previous study (13). The inherently high SNR of SSFP enables adequate differentiation of arterial vasculature from the potential soft-tissue artifacts associated with the subtraction procedure.

There were several limitations to this study. First, because all arteries of the lower extremity, including the iliac region, are required to be screened in diabetic patients, a bolus-chase three-station technique was used in our study according to the institutional standard clinical contrast-enhanced MR angiography protocol of the lower extremities. The foot data set was acquired at the last station in which venous contamination cannot be completely avoided during contrast-enhanced MR angiography. This may lead to lower image quality in the comparison between unenhanced and contrast-enhanced MR angiography. Furthermore, conventional angiography was not available as a standard of reference in our comparison study. There may be bias in the assessment of diagnostic accuracy of unenhanced MR angiography with use of multistation contrast-enhanced MR angiography as the standard of reference. Alternative methods such as time-resolved contrast-enhanced MR angiography or dedicated single-station contrast-enhanced MR angiography targeted at the feet may be better standards of reference because venous contamination for imaging of the infragenual arteries may be avoided. However, all patients in this study were referred for the diagnosis of diseases of all lower-extremity arteries. Angiography with either time-resolved contrast-enhanced MR angiography or single-station contrast-enhanced MR angiography would require the use of an additional dose of gadolinium-based contrast material. Because the potential risk of nephrogenic systemic fibrosis increases in patients with severe renal insufficiency and diabetic patients are known to have a high incidence of renal insufficiency, the institutional review board did not approve of the study protocol involving additional injection of gadolinium. Multistation contrast-enhanced MR angiography was therefore used as the standard of reference in this study. Thus, a direct comparison of image quality between the two techniques was not performed in our study. Further studies using time-resolved contrast-enhanced MR angiography or invasive conventional angiography as the standard of reference would be helpful to validate the clinical significance of the unenhanced MR angiography technique. Last, unenhanced MR angiography was always performed before contrast-enhanced MR angiography, which could be favorable to unenhanced MR angiography image quality and might lead to increased motion artifacts on contrast-enhanced MR angiograms.

In conclusion, unenhanced MR angiography performed with FSD-prepared SSFP enables clear depiction of the foot arterial tree and accurate detection of significant arterial stenosis. The technique has the potential to be a safe and reliable screening tool for the assessment of foot arteries in diabetic patients without the use of gadolinium-based contrast material.

Advance in Knowledge

• ■ Unenhanced MR angiography performed with flow-sensitive dephasing (FSD)–prepared steady-state free precession (SSFP) enables clear depiction of the foot arterial tree and accurate detection of significant arterial stenosis (≥50%), with a sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of 88% (35 of 40 segments), 93% (107 of 115 segments), 81% (35 of 43 segments), 96% (107 of 112 segments), and 92% (142 of 155 segments), respectively.

Implication for Patient Care

• ■ Unenhanced MR angiography performed with FSD-prepared SSFP has the potential to be a safe and reliable screening tool for the assessment of foot arteries in diabetic patients without the use of gadolinium-based contrast material.