Introduction

Prior studies of isometric contraction in calf muscle using velocity encoded phase contrast (VE-PC) imaging have used either single slice imaging to extract 2D strain and strain rate tensors [1] or sequentially acquired single slice images which were concatenated to form a volume to extract 3D strain and strain rate tensors [2]. The advantage of 2D VE-PC is that it is fast but requires acquiring the images in the plane of the fibers to leverage the fact that muscles exhibit planar deformation; a 3D acquisition circumvents the need for this positioning. A volume acquisition VE-PC sequence is usually too long for subjects to maintain consistent contractions through the scan duration. In this study, we explored a highly accelerated 4D Compressed Sensing flow sequence to image the much smaller motion seen in muscle tissue.

Methods

Six young subjects were recruited after IRB approval and scanned on a 3-T Siemens Prisma scanner with a specially designed 8-Ch phased array coil. The leg was constrained in a foot pedal device with a strain sensor embedded in the foot-plate and a visual display of the current force curve superposed on a template allowed subjects to exert the target sub-maximal forces ~30%MVC. The force was continuously recorded during the acquisition. The prototype sequence was a highly accelerated (×7.7) Cartesian 4D flow imaging sequence using L¹ regularized wavelet compressed sensing sequence [3,4]. Dynamic images of calf muscles were acquired under isometric contractions using a gated 4D CS Flow sequence (TE: 4.43 ms, TR: 51.2 ms, FA: 80), velocity encoded in all three directions (VENC=15 cm/sec), 2 k-space lines per segment, prospectively gated 50 temporal phases (cycle length of 2.9 sec), slab of 24 sagittal slices (5 mm thick slice partitions), FOV: 150 (PE) × 300 (RO) mm, Matrix Size: 66 (PE) × 160 (RO), 1.9 × 1.9 mm in-plane resolution. 3D strain and SR tensors were calculated for the acquired slices. Prior to the analysis, the phase-contrast images were corrected for phase shading artifacts and denoised using a 3D anisotropic diffusion filter. Voxels in the entire volume were tracked to obtain displacements and locations in subsequent temporal frames. The 3D strain and SR images were generated from the spatial gradients of the velocity and displacement images and diagonalized to yield Eulerian, E, and Lagrangian, L, strains and SR in the principal basis. Two invariants were calculated from the rank 2 strain and SR tensors (volumetric and maximum shear strain, E_max, L_max and maximum shear strain rate SR_max). Quantitative analysis was performed for a ROI placed in medial gastrocnemius and soleus muscles. Indices were extracted at the frame corresponding to peak (negative) E_λ1(fiber) [displacements and strains] and SR_λ1(fiber) [velocities and strain rates] during the contraction part of the cycle; here E_λ1, L_λ1 or SR_λ1 correspond to the component of the strain and SR tensor that is closest to the fiber.

Results

Figure 1 shows the displacements obtained from the whole volume tracking as a function of the dynamic cycle. Maximum displacements are seen in the foot-head direction (y-axis) and peak displacements occurs around frame 30 (peak of the force curve). Figure 2 shows the plot of the strains (Eulerian) eigenvalues for an ROI in the MG and in the soleus. E_λ1 is negative and represents compressive strains (direction of E_λ1 is closest to the muscle fiber direction). E_λ3 is positive and represents radial expansion in the fiber cross-section. The planar deformation in both the soleus and medial gastrocnemius is clearly evident in the very small values of E_λ2. Figure 3 shows a movie of the strain rate indices: SR_λ1, SR_λ3 and SR_max. A negative peak in SR_λ1 occurs in the compression half while a positive peak occurs in the same time period in SR_λ3. The maximum shear strain has peaks at the same temporal location. SR_λ2 is not shown as its value similar to E_λ2 is very small. Table 1 shows the values of all the strain and strain rate indices calculated from the 3D strain and strain rate tensors for ROIs placed in the MG and SOL.

Discussion

The 4D CS Flow sequence was originally implemented for blood flow imaging and tested for high values of VENC (> 100 cm/sec). This is one of the first adaptations of the high VENC flow imaging to map muscle contractions. The fact that this was a highly accelerated scan allowed true 3D acquisition (24 partitions) in ~5 minutes. This time was short enough to enable consistent isometric contractions at 30%MVC for young subjects. The values of strain, strain rate indices matched those of earlier dynamic studies on the plantarflexors [2]. The volumetric strain is small in the MG and soleus as it should be for an incompressible material. The non-zero volumetric strain values may be related to blood/other fluids moving in and out of the tissue.

Conclusions

The 4D CS-Flow sequence can be applied to image calf muscle motion during isometric contraction in order to extract 3D Strain tensors. The ability to extract the full 3D strain tensors should enable one to answer questions related to muscle deformation anisotropy in the transverse plane, volumetric strain and maximum shear strains.

Acknowledgements

This work was supported by the National Institute on Aging Grant No. R01AG056999.