New insights into intrinsic foot muscle morphology and composition using ultra‐high‐field (7-Tesla) magnetic resonance imaging

Background The intrinsic muscles of the foot are key contributors to foot function and are important to evaluate in lower limb disorders. Magnetic resonance imaging (MRI), provides a non-invasive option to measure muscle morphology and composition, which are primary determinants of muscle function. Ultra-high-field (7-T) magnetic resonance imaging provides sufficient signal to evaluate the morphology of the intrinsic foot muscles, and, when combined with chemical-shift sequences, measures of muscle composition can be obtained. Here we aim to provide a proof-of-concept method for measuring intrinsic foot muscle morphology and composition with high-field MRI. Methods One healthy female (age 39 years, mass 65 kg, height 1.73 m) underwent MRI. A T1-weighted VIBE – radio-frequency spoiled 3D steady state GRE – sequence of the whole foot was acquired on a Siemens 7T MAGNETOM scanner, as well as a 3T MAGNETOM Prisma scanner for comparison. A high-resolution fat/water separation image was also acquired using a 3D 2-point DIXON sequence at 7T. Coronal plane images from 3T and 7T scanners were compared. Using 3D Slicer software, regions of interest were manually contoured for each muscle on 7T images. Muscle volumes and percentage of muscle fat infiltration were calculated (muscle fat infiltration % = Fat/(Fat + Water) x100) for each muscle. Results Compared to the 3T images, the 7T images provided superior resolution, particularly at the forefoot, to facilitate segmentation of individual muscles. Muscle volumes ranged from 1.5 cm3 and 19.8 cm3, and percentage muscle fat infiltration ranged from 9.2–15.0%. Conclusions This proof-of-concept study demonstrates a feasible method of quantifying muscle morphology and composition for individual intrinsic foot muscles using advanced high-field MRI techniques. This method can be used in future studies to better understand intrinsic foot muscle morphology and composition in healthy individuals, as well as those with lower disorders.


Background
The intrinsic foot muscles are those that have their anatomical attachments within the foot, in contrast to the extrinsic muscles that originate on the leg and insert on the foot. Together with passive structures (such as the bony arches, fat pads, ligaments and plantar fascia), the intrinsic foot muscles assist attenuation of forces associated with the foot-ground collision and stiffening of the foot for propulsion [1,2]. The distinguishing contribution of the intrinsic foot muscles, compared to passive structures, is the ability to modulate the energetic function of the foot to respond to changing demands (e.g. acceleration and deceleration, surfaces and footwear) [1,3].
The important contribution of the intrinsic foot muscles to foot function suggests they should be considered when evaluating and treating patients with lower limb disorders. The challenge is to measure these muscles in a valid manner in clinical and research settings [4]. Measures of muscle strength cannot distinguish contributions from intrinsic and extrinsic muscles. The anatomical configuration of the intrinsic foot muscles, such as their small size and depth within the foot, limits electromyography studies to invasive intramuscular evaluations. Imaging modalities, such as ultrasound imaging (US) and magnetic resonance imaging (MRI), provide a non-invasive option to measure muscle morphology (size and shape) and composition, which are primary determinants of muscle function (force output) [5]. Threedimensional MRI is considered a gold standard for quantification of muscle morphology as it allows evaluation of the whole muscle (i.e. volume). This contrasts ultrasound, which typically involves 2-dimensional imaging and measures of cross-sectional area or thickness from transverse and longitudinal views. MRI also permits quantification of muscle composition (e.g. fat infiltration), which can also affect the force producing capacity of a muscle [6].
Advanced MRI technologies and techniques, such as ultra-high-field scanners and chemical-shift sequences, offer the opportunity to enhance current evaluation and understanding of intrinsic foot muscle morphology. With double the signal-to-noise-ratio of 3T, imaging at 7T offers greater precision and accuracy towards quantification of intrinsic foot muscle morphology, particularly the segmentation of individual muscles and has not yet been realised. ) produces wateronly and fat-only images from dual-echo and/or multiecho acquisitions, which permits accurate quantification of muscle composition (e.g. percentage muscle fat infiltration). The aim of this paper is to provide a proof-ofconcept method for the measurement of intrinsic foot muscle morphology and composition with high-field MRI. As part of a series of papers [34,35], here we also     (Fig. 1).
To enable positioning of the foot within the coil, and allow maximal visualisation of the foot on the MR images, the participant was positioned in prone with the knee supported in slight flexion by a foam pad under the tibia, and the ankle and foot supported in plantar flexion within the receiver coil. Foam padding was placed between the dorsum of the foot and the receiver coil. The position of the foot and ankle was supported by foam wedges and sandbags to minimise movement. A T1-weighted VIBEradio-frequency spoiled 3D steady state GREsequence of the whole foot was acquired on each scanner. For comparison of 3T and 7T image resolution, images were imported to OsiriX MD (Pixmeo SARL, Switzerland) and coronal plane images from each T1weighted series were selected at five anatomical locations (Fig. 2): (A) the level of the sustentaculum tali, (B) immediately distal to the talo-navicular joint, (C) immediately distal to the first tarso-metatarsal joints, (D) the mid-shaft of the first metatarsal and (E) the distal shaft of the first metatarsal (up to the metatarsal head). These locations were selected to provide example images from the rearfoot, midfoot and forefoot regions, as well as to include visualisation of all intrinsic foot muscles.
For the quantitative assessment of intrinsic foot muscle structure and composition, muscle volume and percentage muscle fat infiltration were measured from the 7T fat-water images using 3D Slicer software [36]. Regions of interest were manually contoured on each slice for each muscle (abductor hallucis, adductor hallucis, flexor digitorum brevis, quadratus plantae, abductor digiti minimi, flexor hallucis brevis, flexor digiti minimi, lumbricals, and plantar and dorsal interossei). Muscle volumes and the percentage of muscle fat infiltration (muscle fat infiltration % = Fat/(Fat + Water) x 100) were calculated from the segmented images for each muscle using established methods [37,38]. For illustrative purposes only (viewing the three-dimensional reconstruction), the bones of the leg and foot were also segmented on images. Figure 3 presents the coronal plane images at each of the five locations for the 3T and 7T sequences. For reference, Fig. 3 also displays the region of interest (segmentation) for each muscle on the 7T images. As demonstrated, 7T produced higher image resolution, which enhanced the visualisation of individual muscle borders for manual segmentation. At the rearfoot (Fig. 3a), the abductor hallucis was easily visualised on both 3T and 7T images, but 7T was required to visualise muscle borders between quadratus plantae and abductor digiti minimi, as well as between abductor digiti minimi and flexor digitorum brevis. At the midfoot (Fig. 3b), although individual muscle borders were enhanced on 7T images, muscle borders of extensor digitorum brevis, abductor hallucis, flexor digitorum brevis, quadratus plantae, abductor digiti minimi were visible on both 3T and 7T images. At the forefoot (Fig. 3c, d and e), the enhanced visualisation of individual muscle borders on 7T was required to distinguish flexor digiti minimi from abductor digiti minimi, adductor hallucis from flexor hallucis brevis, and abductor hallucis from flexor hallucis brevis. 7T images also provided sufficient resolution to  visualise and individually segment discrete portions of some muscles, for example, the transverse and oblique portions of the adductor hallucis (Fig. 3E), as well as the medial and lateral portions of the flexor hallucis brevis (Fig. 3c, d). The muscle border of the first dorsal interossei was visible on 7T and could be segmented individually, however differentiation of the other individual plantar and dorsal interossei was challenging, even on 7T images. Consequently, these muscles were segmented as a group (Fig. 3c, d, e). Figure 4 illustrates fat and water images, in the coronal plane, from the same five anatomical locations (Fig. 2). For reference, the in-phase image is also included with the region of interest (segmentation) displayed for each muscle. Measurements of muscle volume and percentage muscle fat infiltration are  Table 2. Muscle volume ranged from 1.5 cm 3 (lumbricals) to 19.8 cm 3 (dorsal and plantar interossei) and percentage muscle fat infiltration ranged from 9.2% (extensor digitorum brevis) to 15.0% (abductor digiti minimi). Figure 5 displays a three-dimensional reconstruction of the intrinsic foot muscles from 3D Slicer.

Discussion
Ultra-high-field (7T) MRI enabled the establishment of quantitative methods to provide more detailed measures of intrinsic foot muscle morphology (volume) and composition (percentage muscle fat infiltration) than has been possible with previous technology. In this proof-ofconcept study, we compared images acquired on 7T and 3T scanners, with similar acquisition times, in a single participant on the same day. As 7T MRI has double the signal-to-noise ratio of 3T, the image contrast was, as expected, superior at 7T scanning. The higher resolution was required to enable clear visualisation of anatomical borders for segmentation of individual muscles, particularly in the forefoot. Our observation is supported by previous reports from several authors that segmenting individual intrinsic foot muscles was challenging at lower field strength due to lower image quality [12, 29,   32], as well as the wider literature on morphological musculoskeletal imaging that reports increased spatial resolution at 7T [39]. Although the resolution was lower for the images at 3T than 7T, the lower resolution might be sufficient to enable the segmentation and measurement of crosssectional area of some muscles at some locations. For example, abductor hallucis at the rearfoot and midfoot, and extensor digitorum brevis, flexor digitorum brevis, quadratus plantae and abductor digiti minimi at the midfoot, were visible on the 3T images. Previous investigations have reported measurements of cross-sectional area of these muscles of the rearfoot using lower field strengths (1.5T) [20,31]. However, our findings, and reports from others (using lower magnetic field strengths) [12,29,32,33] of the difficulty to clearly visualise borders of the smaller muscles in the forefoot, draws into question the clinical meaningfulness of individual muscle volumes previously reported from 3T or lower field scanners [24,25,27,28,30]. For future studies using magnetic field strengths of 3T or less, we caution the segmentation of individual intrinsic foot muscles in the forefoot, and recommend that 7T MRI is used to ensure segmentation precision in that regard. Alternatively, muscle volumes could be reported for muscle groups (e.g. abductor hallucis together with flexor hallucis brevis, abductor digiti minimi together with flexor digiti minimi) [16,32], or the intrinsic foot muscles as a whole [7,8,13,14,[17][18][19]26], similar to that reported by previous studies.
In this report of a new method, we follow suggestions from the earlier papers in this series [34,35] towards quantification of muscle fat infiltration for individual intrinsic foot muscles. Using semi-quantitative methods, previous MRI studies at the foot have reported low fatty infiltration in healthy participants [11,16,20]. One study that used a quantitative method [13,14], reported that 22.7% of total intrinsic foot muscle volume was intramuscular adipose tissue (muscle fat infiltration) in 12 healthy adults (mean age 57 years). This was evaluated for the intrinsic foot muscles as a group, not individual muscles. At other areas of the body, muscle fat infiltration has been observed to differ between individual muscles, within a muscle group, in patient populations [40,41]. Considering different pain, injury and disease conditions of the foot may affect individual muscles differently, individual muscle segmentation would provide more specific information. We recommend that future studies perform individual muscle segmentations when evaluating muscle fat infiltration, however we recognise this may be limited by availability of ultra-high-field MRI (e.g. 7T).
Although the benefits obtained from 7T compared to 3T for musculoskeletal imaging, such as faster scan time (by a mean factor of 2) and increased spatial resolution (2 to 4-fold), have been reported [39], the availability of ultra-high-field MRI is a consideration for research or clinical utility. Availability may be influenced by several factors including cost and patient safety. In 2015 the estimated cost to install a 7T system in the US was approximately $10 million, although cost is expected to go down with widespread adoption of systems by institutions/clinics [42]. The field of ultra-high-field imaging is rapidly evolving with growth from approximately 40 scanners worldwide in 2014 [43] to over 80 scanners worldwide in 2019 [44]. This evolution also includes transition from use primarily for scientific and medical research to the first clinical use at the Mayo Clinic Rochester in 2017, as well as the number of scanners available in the marketplace. User costs are also a consideration for researchers and patients. To provide context, at our institution 7T scan costs are 30-40% higher than 3T for the same scanning time period (e.g. 30 minute scan). In terms of patient safety, relatively few biomedical implants have undergone proper testing at 7T and therefore to mitigate risks related to increased forces on metallic implants and the unpredictable radiofrequency-induced tissue heating with ultra-highfield strengths, a significant portion of research participants/patients may be precluded from currently undergoing 7T imaging [45].
Operational parameters can influence the quality of images that are acquired from MRI. As is evident in Table 1, a wide range of parameters have been used to evaluate intrinsic foot muscle morphology in previous work and this has been reported with variation in detail. To enable comparisons between foot imaging studies in future, standardised reporting of parameters is required. We propose reporting of the following operation parameters: field strength (e.g. 7-Tesla); sequence type (e.g. 2point DIXON (3D fast-field echo T1) whole foot); repetition time (e.g. TR 11 ms); echo time (e.g. TE 3.06 ms and 5.61 ms); flip angle (e.g. 3°); field of view (e.g. FOV 111 × 223 mm 2 ); acquired voxel dimensions (e.g. 0.38 × 0.38 mm 2 ); reconstructed voxel dimensions (e.g. 0.38 × 0.38 mm 2 ); bandwidth (e.g. 434 Hz/Px); acquisition time (e.g. TA 11 min 49 s); slice thickness (e.g. 0.38 mm) and number of sliced (e.g. 640 slices). Additionally, to improve transparency and reproducibility, image measurement methods should be described in detail, with the strong recommendation to include figures to illustrate muscle segmentation.
We report images of the intrinsic foot muscles acquired using 3T and 7T with one healthy individual, towards establishing the feasibility of quantifying intrinsic foot muscle structure and composition using higherfield (7T) MRI. It is our hope that this foundational effort can facilitate future studies using these methods, both in healthy individuals and in those with lower limb pain, injury or disease, in order to: (i) enhance understanding of structure and compositional changes in the intrinsic foot muscles across different populations and (ii) generate a body of literature that can be pooled and further scrutinised in systematic reviews with metaanalyses. Although our data illustrates the difference in image quality between 7T and 3T scanners, further research is required to assess the influence of field strength on measurement properties, such as the repeatability of measurements, an aspect that is lacking from the existing literature (see Table 1). Validity of measures should be considered by comparison with anatomical atlases from cadaveric tissue; as reported in the previous methods papers [34,35]. As manual segmentation of the individual intrinsic foot muscles is time consuming (approximately 6 hours per foot for our data) and raterdependent, development of automated methods may improve the efficiency and objectivity of muscle measures. Deep learning convolutional neural network models for muscle segmentation and automatic muscle fat infiltration calculation using fat-water images have demonstrated high test reliability and accuracy for muscles in other body regions [46]. This is an area that should be explored in future studies, as it may improve the ability to quantify and monitor foot muscle structure and composition in individuals with lower limb pain, injury or disease; saving time from arduous manual segmentation techniques, and increase likelihood of clinical implementation.

Conclusions
We have used ultra-high-field (7T) MRI to establish a method to quantify muscle morphology and composition of individual intrinsic foot muscles. This method can be used in future studies to better understand intrinsic foot muscle morphology and composition in healthy individuals, as well as those with lower limb disorders. Following on from papers at the cervical and lumbar spine [34,35], we also emphasise that improved reporting of image acquisition parameters and image measurement procedures is required to assist consistency and allow accurate comparison between studies.