7T bone perfusion imaging of the knee using arterial spin labeling MRI

Abstract

Purpose:

To evaluate the feasibility of arterial spin labeling (ASL) imaging of epiphyseal bone marrow in the distal femoral condyle of the knee at 7T MRI.

Methods:

The knees of 7 healthy volunteers were imaged with ASL using a 7T whole body MRI scanner and a 28-channel knee coil. ASL imaging used a flow-sensitive alternating inversion recovery method for labeling and a single-shot fast spin echo sequence for image readout. ASL imaging with a single oblique transverse slice was performed at 2 slice positions in the distal femoral condyle. Blood flow was measured in 2 regions of interest: the epiphyseal bone marrow and the overlying patellofemoral cartilage. To analyze perfusion SNR, 200 noise images were also acquired using the same ASL imaging protocol with RF pulses turned off.

Results:

Knee bone marrow perfusion imaging was successfully performed with all volunteers. The overall mean of blood flow in the knee bone marrow was 32.90 ± 2.41 mL/100 g/min, and the blood flow was higher at the more distal slice position. We observed significant B₀ and ${\text{B}}_1^{+}$ inhomogeneities, which need to be addressed in the future to improve the quality of ASL imaging and increase the reliability of knee bone marrow perfusion measurements.

Conclusion:

Bone marrow perfusion imaging of the distal femoral condyle is feasible using ASL at 7T. Further technical development is needed to improve the ASL method to overcome existing challenges.

1 |. INTRODUCTION

Perfusion (or blood flow) is an important physiological parameter that measures the level of microscopic blood supply in the capillary bed.¹ Adequate perfusion is critical to ensure sufficient delivery of nutrients and oxygenation to tissue and clearance of metabolic waste; therefore, perfusion measurements reflect vasculature status and tissue viability. Bone vasculature and blood flow have a significant role in bone development and repair as well as in the initiation and progression of skeletal diseases.^2–4

There is growing interest in measuring bone marrow perfusion using imaging methods. Bone marrow perfusion can provide essential knowledge about bone physiology,^5,6 improve our understanding of disease pathophysiology (e.g., osteoarthritis),^7–10 assist the differentiation between normal and abnormal bone marrow,^11,12 and assess the response to prescribed therapy (e.g., tumor anti-angiogenic therapy).¹³ Although DCE MRI has been applied to measuring bone marrow perfusion,^5–7,9–13 the health risks of MRI with gadolinium-based contrast agents (e.g., nephrogenic systemic fibrosis)¹⁴ have raised significant concerns for its routine applications.¹⁵ An alternative perfusion imaging method is desired.

In contrast to DCE-MRI, arterial spin labeling (ASL) MRI¹⁶ uses endogenous blood water spins as intrinsic agents to measure tissue perfusion. The fact that ASL is a non-invasive and non-contrast-enhanced method eliminates safety concerns for human research studies, making it potentially well-suited for routine assessment of bone marrow perfusion and longitudinal monitoring of disease progression and therapy response.

Although ASL imaging is an established technique and has been routinely applied for highly perfused organs, such as the brain and kidneys,^17,18 bone marrow perfusion is a challenging application. ASL imaging is a low SNR technique, and the perfusion level of bone marrow is lower than that of the brain or kidneys.¹⁹ As is well known, there are 2 types of bone marrow: red and yellow bone marrow. In adults, red bone marrow is predominantly found in axial skeleton and flat bones and mainly consists of hematopoietic cells with a relatively abundant and arborized vascular network; yellow bone marrow is primarily found in long bones and should be dominantly present at the distal end of long bones, the so-called epiphysis. Yellow marrow mainly consists of fat cells with a sparse network of capillaries.^20,21 Recently, the feasibility of bone marrow ASL imaging has been demonstrated in the vertebrae,²² which mainly contains red bone marrow. ASL imaging of skeleton regions primarily containing yellow bone marrow, such as the knee, will be challenging. In addition, the complicated architecture of arterial blood vessels in the skeletal system with relatively low blood flow velocities^23,24 make it difficult to use high-SNR ASL imaging methods, such as pseudo-continuous ASL.²⁵

Ultrahigh-field (≥7T) MRI can potentially help overcome these challenges and benefit ASL imaging at 7T²⁶ by increasing SNR,²⁷ prolonging longitudinal relaxation times of arterial blood,²⁸ and improving parallel imaging performance.²⁹ Today, 7T MRI scanners become more widely available since the Food and Drug Administration (FDA) approved its clinical use in the brain and knee imaging. However, challenges of 7T MRI, including B₀ and B₁ inhomogeneities and high specific absorption rate (SAR),³⁰ must also be addressed.

We hypothesize that after addressing the challenges of 7T MRI, bone marrow perfusion in the knee can be measured using ASL. The purpose of this study is to investigate the feasibility, as well as the challenges, of bone marrow perfusion imaging in the knee using a pulsed ASL method at 7T.

2 |. METHODS

2.1 |. Participants

Seven young healthy volunteers (3 males and 4 females, 23 ± 4 years old) were recruited and provided written informed consent before the imaging studies in accordance with a protocol approved by our local Institutional Review Board.

2.2 |. MRI

Knee imaging was performed on a 7T whole body MRI scanner (Siemens Healthcare, Erlangen, Germany) with a 28-channel receive, 1-channel transmit knee coil (QED, Mayfield Village, OH). The imaging protocol included scout localizer, high-resolution anatomic imaging in 3 orthogonal orientations (sagittal, oblique coronal, and axial), B₀ and ${\text{B}}_1^{+}$ mapping, and bone marrow perfusion imaging. High-resolution anatomic imaging used a proton-density weighted turbo spin echo sequence, B₀ mapping used a standard dual echo gradient recalled echo (GRE) sequence, and ${\text{B}}_1^{+}$ mapping used Siemens’ work-in-progress ${\text{B}}_1^{+}$ mapping sequence.

ASL perfusion imaging used a flow-sensitive alternating inversion recovery (FAIR) method³¹ for labeling and a single-shot fast spin echo (ss-FSE) sequence as perfusion image readout (referred to as FAIR ss-FSE). The sequence diagram and relative spatial locations of each RF pulse in the sequence are illustrated in Figure 1. In this sequence, a single ss-FSE image is first acquired to measure the fully relaxed bone marrow tissue magnetization (referred to as the M₀ image) before the acquisition of label and control images. The ASL preparation module consists of: a pre-saturation RF pulse applied at the location of the imaging slice; an inversion RF pulse that alternates between a small inversion slab for control images and a large inversion slab for label images; a specified delay time (TI₁ in Figure 1A) after the inversion RF pulse to define the temporal bolus width of the labeled spins; saturation RF pulses targeting a slab proximal and parallel to the imaging slice; and a post-bolus delay (PBD = TI–TI₁, Figure 1A) to allow the labeled blood to travel down the vascular tree into the small arterioles and minimize undesired hyper-intense intravascular signals. Background suppression³² was not applied in this study to reduce RF power deposition and comply with the long-term SAR limit, which was found to be the primary constraint for knee FAIR ss-FSE imaging.

FIGURE 1.

Sequence diagram (A) and RF pulse spatial positions at the distal femoral condyle (B) of the FAIR ss-FSE method. A single ss-FSE image is first acquired to measure fully relaxed bone marrow magnetization (called M₀ image) before perfusion image acquisitions. This sequence consists of a pre-saturation RF pulse at the slice location, adiabatic inversion RF pulses for either label or control images, and 4 proximal saturation RF pulses to define the temporal bolus width (TI₁) followed by a post-bolus delay (PBD) (TI–TI₁). Knee perfusion imaging was performed at 2 slice positions with a distal slice (solid green) and a proximal slice (dashed green), respectively. The relative slab positions of ASL preparation components are illustrated for imaging with a distal slice

To minimize the peak amplitude of the inversion pulse, an optimized HS4 adiabatic pulse with a 20-ms duration and a time-bandwidth product of 20 was used.³³ For ss-FSE readout, variable-rate selective excitation (VERSE) RF pulses³⁴ were used for refocusing with a hyper echo phase and amplitude modulation scheme.³⁵ To improve fat suppression within the bone marrow, strong fat suppression was achieved by applying 3 fat saturation RF pulses with altered phases before image readout. To boost the water signal level in the bone marrow, the signal gain was specifically elevated for ss-FSE image acquisition.

To minimize B₀ inhomogeneity within the imaging region and improve fat saturation, Siemens’ advanced B₀ shimming was applied, in which B₀ shimming was performed twice using sequentially acquired B₀ maps. To overcome ${\text{B}}_1^{+}$ inhomogeneity and ensure the adiabatic condition for the inversion RF pulse throughout the entire labeling region at the proximal side of imaging slice, a nominal flip angle (i.e., an RF transmit voltage) at least 2 times that needed for an adiabatic inversion was applied for the inversion RF pulse.³³

A single oblique transverse slice was used for knee bone marrow perfusion imaging using the FAIR ss-FSE sequence (Figure 1B). Blood flow in large vessels of the posterior knee can produce ghosting artifacts. Such flow-related artifacts spread along the phase-encoding direction and can overlap the bone marrow area. To avoid the potential undesired effects of flow-related artifacts on bone marrow perfusion measurements, left–right phase-encoding direction was used and oriented roughly perpendicular to the anterior–posterior direction of the knee. The ss-FSE image acquisition used fat saturation and elevated the signal gain to boost the water signal level in the bone marrow. Perfusion imaging was performed at 2 slice positions: 1 distal slice with all 7 volunteers and 1 proximal with 5 of the volunteers (Figure 1B). For imaging scans at either of these 2 slice positions, the knees of participants were always positioned to place the image slice at the proximal side of the coil with at least ¾ coverage of the knee coil over the labeling region. In addition, to avoid excessive perfusion signal decay,¹⁹ an intermediate inflow time or total delay time was applied in this feasibility study.

ASL imaging acquisition parameters were: TR/TE = 5000/16 ms; hyper echo flip angle = 90°; FOV = 140 × 140 mm²; matrix size = 70 × 70; in-plane resolution = 2 × 2 mm²; phase encoding direction = left to right; slice thickness = 10 mm; partial Fourier = 5/8; GRAPPA parallel acceleration factor (R) = 4 with 24 separately acquired reference lines; labeling time (TI₁)/total delay time (TI) = 600/1200 ms; control/labeling inversion slab size = 30/210 mm; number of label and control images = 60; slab thickness/RF duration/interval of proximal saturation = 60 mm/25 ms/50 ms; and total imaging time = ~5 min. To facilitate perfusion SNR analysis, 200 noise images were acquired using the same protocol with RF pulses turned off.²⁵

2.3 |. Image processing and data analysis

For each ASL series, 2D motion correction was performed using FSL toolbox (FMRIB, Oxford, UK).³⁶ After motion correction, label and control images were pairwise subtracted to obtain perfusion-weighted images. The perfusion-weighted images were subsequently averaged to get a mean perfusion-weighted image that was used for perfusion quantification. The calculation of the blood flow map and region of Interest (ROI) analyses were performed using in-house scripts implemented in MATLAB (The MathWorks, Natick, MA).

Blood flow maps were calculated by applying the single compartment blood flow quantification model³⁷

$$\text{BMBF}(\text{r})=\Delta \text{M}(\text{r})/\left[2\alpha \times {\text{M}}_0(\text{r})\times {\text{TI}}_1\times \text{exp}\left(-\text{TI}/{\text{T}}_{1\text{b}}\right)\right],$$ (1)

where r is the spatial location of imaging voxel, ΔM the perfusion signal from a mean perfusion-weighted image, M₀ the fully relaxed magnetization of bone marrow, TI₁ the sequence-defined temporal bolus width, TI the total delay time, T_1b the longitudinal relaxation time of the arterial blood (assumed to be 2041 ms),²⁸ and α the labeling efficiency (assumed to be 0.95 for the applied HS4 adiabatic RF pulse).³³

Spatial and temporal SNR analyses were performed for perfusion signals within the bone marrow area. The SD map of 200 noise images was used as the estimate of spatial noise.²⁵ The spatial SNR map was then estimated as follows

$${\text{SNR}}_{\text{spatial }}(\text{r})=\text{S}(\text{r})\times \surd \left({\text{N}}_{\text{avg }}/2\right)/\sigma (\text{r}),$$ (2)

where S(r) represents perfusion signals, N_avg is the number of temporal perfusion signal averages, and σ(r) the spatial noise. The temporal SNR map was calculated as the ratio of mean perfusion-weighted image to the temporal SD of all perfusion-weighted images.

ROIs were defined in 2 regions: one covering the patellofemoral cartilage with surrounding joint fluid, which does not have blood flow, and another covering the bone marrow, as illustrated in Figures 2 and 3. To further reduce the impact of subtraction errors resulting from residual small motion and labeled blood signals in the large arteries, trimmed mean signals within ROIs were used in the final perfusion signal measurements by excluding the 5% of voxels with the lowest and highest values.^38,39

FIGURE 2.

Perfusion imaging results from 1 participant using a proximal slice (please refer to Figure 1B): (A) label and control images and blood flow map; (B) quantitative blood flow measurements in regions of interests (ROIs) covering the avascular patellofemoral cartilage (yellow) and bone marrow (green). Error bars represent standard errors

FIGURE 3.

Perfusion imaging results from the same participant as shown in Figure 2 using a distal slice (please refer to Figure 1B): (A) label and control images and blood flow map; (B) quantitative blood flow measurements in regions of interests (ROIs) covering the avascular patellofemoral cartilage (yellow) and bone marrow (green). Error bars represent standard errors

Statistical analyses were performed within GraphPad Prism software (GraphPad Software, La Jolla, CA) to compare bone marrow perfusion between distal and proximal slices using a 2-tailed paired t-test and to evaluate whether perfusion measurements from the ROI covering the patellofemoral cartilage were significantly different than 0 using a 1 sample Student’s t-test. Statistical significance was defined as P < .05.

3 |. RESULTS

Knee bone marrow perfusion imaging was successfully performed with all volunteers. Figure 2 shows perfusion imaging results from 1 participant using a proximal slice, and Figure 3 presents results from the same participant using a distal slice. Although there was measurable blood flow in the bone marrow areas, mean blood flow in the avascular patellofemoral cartilage was not significantly greater than 0 (−0.04 ± 3.08 mL/100 g/min, P = .989).

Quantitative measurements of bone marrow blood flow, as well as the estimates of spatial and temporal SNRs of perfusion measurements, are shown in Figure 4. In the 5 volunteers who were imaged at both slice locations, bone marrow blood flow measurements were significantly greater in distal slices than those in proximal slices (38.32 ± 3.65 vs. 26.80 ± 2.37 mL/100g/min, P = .002) (please refer to Supporting Information Figure S1 for bone marrow perfusion maps from these subjects). Spatial and temporal SNRs were also significantly greater in the distal slices than those in proximal slices (spatial SNR: 6.73 ± 1.27 vs. 3.98 ± 0.58, P = .024; and temporal SNR: 2.09 ± 0.19 vs. 1.72 ± 0.12, P = .027). The overall mean blood flow within bone marrow areas of distal and proximal slices was 32.90 ± 2.41 mL/100 g/min, which was significantly greater than 0 (P < .0001).

FIGURE 4.

Measurements of bone marrow blood flow, as well as spatial and temporal SNR, at distal (N = 7) and proximal (N = 5) slice positions

We observed large B₀ and ${\text{B}}_1^{+}$ inhomogeneities during our knee bone marrow ASL imaging at 7T. Figures 5 and 6 show representative relative B₀ and ${\text{B}}_1^{+}$ maps and field coefficient of variance (CV) values within bone marrow areas for individuals. The bar graph in Figure 7 shows the CV values of B₀ and B₁ fields at the distal slice position from 7 participants and those at the proximal slice position from 5 participants. Further analyses using data from 5 volunteers imaged at both slice locations show that the CV values of B₀ and B₁ fields within bone marrow areas were significantly greater in distal slices than those in proximal slices (B₀ field: 12.08 ± 2.51% vs. 6.64 ± 1.25 %, P = .0304; B₁ field: 10.01 ± 0.55% vs. 7.32 ± 0.90%, P = .0448).

FIGURE 5.

One typical participant’s proton-weighted (PD) anatomic image and relative B₀ and B₁ maps at the proximal slice position (A) and coefficient of variance (CV) values of B₀ and B₁ fields within the green ROI covering bone marrow (B)

FIGURE 6.

One typical participant’s proton-weighted (PD) anatomic image and relative B₀ and B₁ maps at distal slice position (A) and coefficient of variance (CV) values of B₀ and B₁ fields within the green ROI covering bone marrow (B)

FIGURE 7.

Coefficient of variance (CV) values of B₀ and B₁ fields within bone marrow areas at distal (N = 7) and proximal (N = 5) slice positions

4 |. DISCUSSION

ASL perfusion imaging is a potentially valuable tool to assess bone perfusion noninvasively without the need for intravenous contrast administration. However, low bone marrow perfusion level¹⁹ makes the application of ASL imaging challenging, especially in skeletal regions mainly containing yellow bone marrow, such as the epiphysis of the distal femur of the knee. Ultrahigh-field (≥7T) MRI can benefit ASL imaging because of increased SNR,²⁷ prolonged longitudinal relaxation times of arterial blood,²⁸ and improved parallel imaging performance.²⁹ Recent study suggested that after overcoming imaging challenges, 7T renal ASL imaging could be achieved within a single breath-hold,²⁶ which motivated us to evaluate the feasibility of measuring knee bone marrow perfusion using ASL imaging at 7T. To our knowledge, this is the first ASL imaging of bone marrow perfusion in the knee at 7T.

4.1 |. Bone marrow blood flow

Our study indicates that it is feasible to measure knee bone marrow perfusion using the FAIR ss-FSE method at 7T (Figure 4). On average, we measured blood flow in the knee bone marrow to be ~33 mL/100 g/min, which was lower than that in highly perfused organs, such as the brain (70 to 90 mL/100 g/min for gray matter in cerebral cortex) and kidneys (~300 mL/100 g/min for renal cortex).^25,26 This is primarily because our perfusion measurements are within the epiphyseal segment that mainly contain yellow bone marrow, and yellow bone marrow primarily contains fat cells and a sparse network of capillaries.^20,21

We also found that bone marrow blood flow was greater more distally in the femoral condyle, which could be related to the intrinsic differences in blood supply to the distal and proximal bone marrow spaces of the knee.²³ Importantly, the blood flow within the avascular patellofemoral cartilage was ~0 as expected and served as an internal control. These results show that ASL perfusion imaging using the FAIR ss-FSE sequence can differentiate a region with expected absent blood flow from a region with expected high blood flow.

Although PET imaging and DCE MRI have been used to investigate bone perfusion, these imaging modalities cannot provide direct estimates of blood flow. Rather, they provide indirect information associated with bone perfusion, such as ¹⁸F-fluoride uptake in PET imaging or pharmacokinetic parameters in DCE MRI. One in vivo study with dogs using the gold standard perfusion imaging method, 99Tcm-labeled microspheres, found that the perfusion of the distal epiphysis in non-tumoral bone was ~27 mL/100 g/min,⁴⁰ which is slightly lower than our knee bone marrow perfusion measurements from young healthy human participants (32.90 ± 2.41 mL/100 g/min). To date, only 1 other study has investigated ASL imaging of bone marrow perfusion.²² This prior study found that ASL-measured bone marrow perfusion in the spine was correlated to pharmacokinetic parameters from DCE MRI. Our bone marrow perfusion measurements in the knee are lower than those in the spine, which may be because in adults, the vertebrae mainly contains red bone marrow that has a relatively abundant and arborized vascular network.^20,21

4.2 |. Fat suppression and signal gain

Effective fat suppression and proper signal gain are critical for bone marrow ASL perfusion imaging using the FAIR ss-FSE method. In contrast to ASL perfusion imaging in the brain and kidneys where the parenchyma tissue itself for perfusion measurements does not contain fat, knee bone marrow ASL imaging measures labeled blood signals from bone marrow that is highly rich in fat. Fat suppression not only ensures that signals in label and control images are dominantly from water spins, but also decreases the signal intensity of raw label and control images, which helps reduce subtraction errors. In our study, to improve fat suppression within bone marrow, enhanced fat suppression was applied before the image readout using 3 fat saturation RF pulses with altered phase. Furthermore, maximizing signal dynamic range for fat-suppressed bone marrow tissue signals by increasing signal gain is also helpful for successful bone marrow ASL perfusion imaging. Using a large signal dynamic range can increase the signal intensity difference between bone marrow tissue and the background and avoid fat-suppressed bone marrow tissue signals that are very close to background signal level.

4.3 |. B₀ and ${\text{B}}_1^{+}$ inhomogeneity

It is well known that the inhomogeneities of the B₀ and ${\text{B}}_1^{+}$ fields increase with the strength of magnetic field of a MRI scanner. As expected, we observed large B₀ and ${\text{B}}_1^{+}$ inhomogeneities during our knee bone marrow ASL imaging at 7T (Figures 5–7).

Although susceptibility-associated image distortion and ghosting artifacts were minimal by using ss-FSE readout, B₀ inhomogeneity can adversely affect fat suppression within bone marrow, resulting in large subtraction errors for ASL signal measurements. In our study, to minimize B₀ inhomogeneity within the imaging region, the B₀ field was optimized through 2 iterative rounds of B₀ shimming based on sequentially acquired 3D phase maps by using Siemens advanced B₀ shimming method. As shown in Figures 5 and 6, even with advanced B₀ shimming, the B₀ field within the knee is still quite inhomogeneous, especially along the anterior to posterior direction.

${\text{B}}_1^{+}$ inhomogeneity can result in either under-flipping or over-flipping magnetization excitation across the bone marrow area, both of which can cause perfusion signal loss, resulting in regional poor perfusion signals within a perfusion map. Using the single-channel transmit QED knee coil, we found that the characteristics of the ${\text{B}}_1^{+}$ field were determined by the fixed geometry configuration of the coil, and the pattern of the ${\text{B}}_1^{+}$ field inhomogeneity remained consistent across participants. As shown in Figures 5 and 6, the ${\text{B}}_1^{+}$ field was uniform in the majority of the bone marrow region, but became more inhomogeneous at the left-bottom area. When the flip angle for the ss-FSE readout is calibrated for magnetization excitation based on the mean ${\text{B}}_1^{+}$ field within the majority of bone marrow region where ${\text{B}}_1^{+}$ field is relatively uniform and low, over flipping will occur in a region with a relatively high ${\text{B}}_1^{+}$ field.

As mentioned in the Results section, the B₀ and ${\text{B}}_1^{+}$ inhomogeneities at distal slice position were significantly worse than those in proximal slice position. Based on our observations, such significant differences are not because of spatial changes of B₀ and ${\text{B}}_1^{+}$ field characteristics but because of bone marrow area changes across slices.

${\text{B}}_1^{+}$ field drops around the edges of coils. To address ${\text{B}}_1^{+}$ inhomogeneity in the labeling region at the proximal side of imaging slice and ensure the adiabatic condition for the inversion RF pulse throughout the labeling region, a nominal flip angle (or an RF transmit voltage) at least 2 times of that needed for an adiabatic inversion was applied for the inversion RF pulse.³³

4.4 |. Strategies to comply with SAR limits

Because of the high power deposition from the inversion and refocusing RF pulses, the long-term SAR limit was found to be the dominant constraint for knee bone marrow perfusion imaging using the FAIR ss-FSE method despite using a long TR of 5 s. To manage the long-term SAR, the following strategies were used: (1) high parallel imaging acceleration factors (e.g., 4) and partial Fourier (e.g., 5/8) were used to reduce the FSE echo train length and therefore the number of refocusing RF pulses; (2) reference lines for parallel imaging were acquired separately and once at the beginning of perfusion scan and before the acquisition of the M₀ image; and (3) VERSE RF pulses and hyper-echoes were used within the ss-FSE imaging readout.

4.5 |. Study limitations

Although the feasibility of knee bone marrow ASL imaging at 7T has been demonstrated, our study has several limitations. First, the implemented FAIR ss-FSE method only allowed single slice coverage, and ASL imaging of knee bone marrow at different locations within the distal femoral condyle of the knee had to be achieved through multiple acquisitions. Second, constrained by the long-term SAR, it was impractical to perform background suppression³² for our knee bone marrow ASL studies at 7T. Finally, with the single-channel transmit knee coil, ${\text{B}}_1^{+}$ shimming could not be performed to improve ${\text{B}}_1^{+}$ field homogeneity for either ASL labeling or image readout.²⁶

Further technical development is needed to improve 7T knee bone marrow ASL imaging. For example, better imaging efficiency with reduced repetition times can be achieved by using alternative adiabatic RF pulses (e.g., gradient offset independent adiabatic [GOIA] RF pulses)³³ and imaging readouts with lower total RF power requirements. To address B₀ issues, 3rd order shimming can be applied in the future.⁴¹

${\text{B}}_1^{+}$ field homogeneity can be improved by performing ${\text{B}}_1^{+}$ shimming separately for ASL labeling and image readout with a multi-channel transmit knee coil, and the specific needs of different RF pulses in the sequence can be addressed by dynamically applying ${\text{B}}_1^{+}$ shimming solutions that are optimized separately for ASL labeling and image readout.²⁶ Furthermore, using parallel transmit RF pulses that simultaneously minimize ${\text{B}}_1^{+}$ inhomogeneity and reduce RF power deposition may further improve UHF knee bone marrow perfusion imaging.⁴²

5 |. CONCLUSIONS

Bone marrow perfusion imaging of the distal femoral condyle of the knee is feasible using ASL at 7T. Further technical development is needed to improve the ASL method to overcome existing challenges.

Supplementary Material

Supp figS1

FIGURE S1ASL control images and bone marrow perfusion maps from 7 subjects