Walking is one of the most important activities of daily living. It is a complex orchestration of muscular contractions with the goal of moving the body forward while maintaining stance stability (25). Understanding the individual contributions of the muscles involved in walking is important for a variety of purposes (including rehabilitation, medical education, and the validation of musculoskeletal models), but until a few decades ago, it was not possible to measure the contributions of all muscles of the lower limb to walking because of limitations of existing measurement techniques.
Muscle activity during walking has been studied extensively using EMG (e.g., (4,9,16)). Although this method is convenient and noninvasive, it has several important limitations, as follows: deep-lying muscles cannot be measured unless needle electrodes are inserted invasively, only a limited number of muscles can be measured at the same time, a reference contraction is required if muscles are to be compared with each other, and adipose tissue and crosstalk from other muscles are confounding factors (7,32). [18F]-fluorodeoxyglucose in combination with positron emission tomography (FDG-PET) does not suffer from these drawbacks. It can be used to estimate muscle activity in all muscles of the lower limb (including deep-lying muscles) in one session. The working principle of FDG-PET is that exercising muscles take up glucose (and FDG) from the blood to replenish their expended energy. Using a concurrently made magnetic resonance imaging (MRI) or computed tomography (CT) scan for anatomical reference, the muscles can be identified and their individual FDG uptake can be quantified.
FDG-PET has been used for studying several exercises (e.g., cycling (11,12), walking with a stride assistance system (29,31), and simple flexion/extension exercises (20,23,27)) since Fujimoto et al. first used it to analyze muscle activity during running (10). Thus far, only a few studies have used FDG-PET for analyzing walking (21,30,31). Oi et al. (21) performed important work by comparing FDG uptake results with EMG and kinematic and kinetic studies and demonstrating the concurrent validity of measuring the muscular activity of the lower limb with FDG-PET. Shimada et al. (31) evaluated FDG uptake during walking with and without a stride assistance system and in older subjects compared with younger subjects (30). These studies, however, performed the quantification of the FDG uptake on one or only a few slices of the PET image. Thus, only a small portion of each muscle at a fixed cross-sectional level was analyzed. This raises the question whether analyzing such a small part provides an accurate representation of glucose uptake in the entire muscle. There are indications in the literature that muscle metabolism exhibits spatial heterogeneity. For instance, Pappas et al. (23) found an inhomogeneous distribution of glucose uptake in the transverse direction and along the longitudinal axis of the biceps brachii after an elbow flexion exercise. In the lower limb, Kalliokoski et al. (18) demonstrated heterogeneous glucose uptake between the four quadriceps femoris muscles during a knee extension exercise. Besides the spatial heterogeneity, certain muscles may not even be included at the cross-sectional level at which the slice is taken.
With the advent of new MR image analysis techniques, it is now feasible to semiautomatically segment not just one or a few slices of muscles but entire muscles. To our knowledge, volumetric glucose uptake in entire muscles during walking has not yet been investigated. The purpose of this study was to quantify the activity of the muscles in the lower limb during walking using FDG-PET scans and three-dimensional (3D) MRI segmentations. Furthermore, we aimed to determine whether the volumetric analysis would yield an appreciable benefit compared with analyzing only a single slice.
Ten healthy subjects participated in this study (Table 1). They had no history of major injury, had not undergone orthopedic surgery on the lower limb, and had no history of ailment related to CHO metabolism or to cardiac or muscular disease. One of the inclusion criteria was being able to walk for at least 90 min without any difficulty. The subjects were recruited from the research staff and a database of volunteers available at our department. We specifically chose to include a wide variety of subjects in this study in terms of age (23–60 yr), sex (five men and five women), and anthropometry (1.60–1.95 m, 55.5–91.7 kg). This was done because it was part of a larger research project that aimed to validate personalized musculoskeletal models of the subjects at a later stage (e.g., (5)). The study procedures were approved by the ethical committee of the region Arnhem–Nijmegen, The Netherlands. A written informed consent was obtained from each participant.
The subjects walked on a treadmill at their own comfortable pace for 90 min, 60 min before and 30 min after injection with FDG 53.6 ± 1.8 MBq (Fig. 1). The walking speed was 1.26 ± 0.11 m·s−1. The PET/CT scan was started 30 min after finishing the last walking phase, which allowed the subject to be transported from the treadmill to the nuclear medicine department in a wheelchair, to void any excess radioactivity, and to position the subject in the PET/CT scanner. In addition to the PET/CT scan, an MRI scan had been taken a few weeks earlier for anatomical reference. Before the PET/CT scan commenced, the subject’s legs were positioned on the table to reproduce as closely as possible their position as it was during the MRI scan using a ruler and a vacuum pillow that could be molded into the desired shape. Before the testing procedure, the subjects had to refrain from eating and drinking anything except water for at least 6 h and from participation in strenuous physical activity for 2 d.
FDG is a glucose analog, and as such, after entering the cell, FDG is phosphorylated into FDG-6-phosphate. However, FDG-6-phosphate is not a significant substrate for further metabolism. It is not converted into glycogen to any significant extent and is not further metabolized in the glycolytic pathway. It thus becomes trapped within the cells, which is convenient from an imaging viewpoint (6). It has a half-life time of 110 min.
PET and CT scans
The PET/CT scans were made with a Biograph 40 mCT (Siemens AG, Erlangen, Germany) (17), in 3D mode, using an extended axial field of view of 216 mm. The PET scans lasted for 4 min per bed position. Eight to 10 bed positions were acquired depending on the subject’s height. Thus, the scan time was 32–40 min. The scan region ranged from the feet to at least the iliac crest, with 50-mm overlap in the bed positions. The slice thickness was 2.0 mm. The PET images were reconstructed with the TrueX algorithm (with a spatially varying point spread function) and the incorporation of time-of-flight measurements (ultra-HD PET). Image reconstruction was performed using three iterations, 21 subsets, a matrix size of 512 × 512, and voxel spacing of 1.6 mm. After reconstruction, filtering was performed with a 3D Gaussian filter kernel with a full width at half maximum of 3.0 mm. The CT images used for attenuation correction were reconstructed with the B19f convolution kernel and a slice thickness of 5.0 mm, whereas the CT images used for anatomical reference were reconstructed with the B31f convolution kernel and slice thickness of 2.0 mm to correspond with the slice thickness of the PET images.
MRI scans were made with a Magnetom Skyra (Siemens AG, Erlangen, Germany) at 3 T and with a 400-mm transaxial field of view. The feet and ankles were placed in a head/neck coil containing 20 elements; the other areas of the lower limb were scanned with body-phased array coils containing 18 elements combined with a spine array coil containing 32 elements that was embedded in the table. The scan protocol was specifically designed to yield optimal distinction between muscle boundaries (based on Scheys ((28)). The settings were as follows: T1-weighted image (repetition time/echo time, 450–545/9 ms) with turbo spin echo; number of excitations, 3; two echoes per excitation; 92–195 echo trains per slice; matrix size, 512 × 512; no gap between slices; and in-plane resolution, 1.0 × 1.0 mm. Parallel imaging was not used, and there was no inversion pulse at the front end of the sequence. The hip, knee, and ankle regions were scanned with a 3.0-mm slice thickness, and the long bone regions in between were scanned with 8.0-mm slice thickness. This was done to obtain a higher level of detail in the areas where most muscles originate or insert. Six imaging stacks that covered the ankles (1), lower legs (1), knees (1), thighs (2), and hip region (1) were used. An overlap of at least two 3.0-mm slices was used to allow proper “stitching” of the image stacks into a single integrated scan.
To obtain FDG uptake values of each individual muscle, several sequential steps were performed using Mimics (Materialise N. V., Leuven, Belgium), Matlab 2012b (Mathworks, Natick, MA), and Insight Segmentation and Registration Toolkit 4.5.2 (ITK) (Kitware Inc., Clifton Park, NY).
- 1) 3D regions of interest (ROI) were drawn manually around the muscles of the left lower limb of one of the subjects from the axial images. The boundaries of the ROI thus reflected the muscle boundaries. In total, 39 muscles were segmented in this way (Table 2 and Fig. 2). The resulting segmentation of the muscles in the lower limb will henceforth be called “atlas.”
- 2) The atlas was used to semiautomatically segment the muscles in the left lower limb of the other nine subjects in Mimics. The workflow was as follows: 1) create an initial crude mask containing all the muscles using a threshold on the voxel intensities, 2) remove strong gradients from the mask, 3) remove remaining undesired structures (fat and bones) using pen strikes on approximately every fifth slice as user input, 4) register the atlas to the subject automatically using an algorithm based on nonrigid registration, and 5) in an automatic postprocessing step, the transformed segmentation was combined with the voxel intensity information from the MRI scan to obtain the muscle ROI of the subject. The muscle ROI were thoroughly checked and manually corrected by a researcher where necessary. Hence, this step produced a segmentation of all muscles of the left lower limb for each subject.
- 3) The MRI scan and its corresponding ROI were subsequently registered onto the CT scan using ITK. The registration results were again thoroughly checked and manually corrected by a researcher where necessary.
- 4) The ROI were exported per muscle as Digital Imaging and Communications in Medicine (DICOM) mask files. The DICOM file format is a standard method for transferring images and associated information between devices and software manufactured by various vendors, including those used in this study.
- 5) The ROI DICOM files and the subject’s PET scan (obtained directly from the PET scanner in DICOM format) were opened in Matlab 2012b and multiplied using a custom algorithm. In the algorithm, the ones (reflecting the muscle’s position) and zeros (surrounding area) in the ROI DICOM files were multiplied with the PET scan (gray values reflecting the FDG uptake (Bq·mL−1)). The algorithm also extracted the rescale slope and intercept values from the PET DICOM headers, which were used to correct the FDG uptake values of the PET scan (Bq·mL−1) for decay time since the start of the scan.
We emphasize that the PET DICOM files from which we extracted the FDG uptake values were obtained directly from the Biograph 40 mCT and that they were thus not “morphed” or modified in any other way. Furthermore, we only analyzed the left lower limb in this study.
Analysis of volumetric uptake
The FDG uptake values per muscle were evaluated primarily with the volumetric %FDG uptake value. To obtain this value, we first calculated the volumetric mean standardized uptake value (SUV) of each muscle. The mean SUV reflects the average FDG uptake in a muscle, normalized by body mass and injected activity.
In this equation, the numerator is the mean radioactivity count from all the voxels in the muscle divided by the muscle volume. The denominator reflects the total injected activity divided by the total body mass. The SUV was corrected for decay between the injection and the start of the scan. Because the SUV is normalized by the muscle volume, small muscles can exhibit similar or higher SUV values than large muscles, even though the large muscles plausibly exert much more force and/or perform much more work during gait. Therefore, we multiplied each muscle’s SUV by its volume. The resulting value was expressed as a percentage of the sum of the product of each muscle’s SUV times its volume, yielding %FDG uptake as our primary outcome measure:
where i is the index of the muscle and j is a summation index that includes all 39 muscles. The %FDG uptake values thus reflect the amount of FDG in each muscle relative to the total amount of FDG in the muscles of the limb.
Analysis of volumetric versus slice-based uptake
To allow comparison of our volumetric analysis with the commonly used method in the literature of analyzing single slices, the %FDG uptake values were calculated both for entire muscles and at three discrete slice heights, as follows: 3 cm above the femoral head, 50% of the distance between the joint space of the knee and the top of the femoral head, and at 30% of the distance between the joint space of the knee and the tip of the lateral malleolus. These slice heights were based on commonly used slice heights in the literature for this type of study (21,29–31,33). The percentage difference between the methods was calculated using the following formula:
where i is the index of the muscle. We did not want to choose either the volumetric or the slice-based data to be the “reference value”; therefore, this unbiased version of the percentage difference was used. The comparison between methods was made only among the five most active muscles in each segment (based on the %FDG uptake values observed in the volumetric analysis) because differences in these muscles were considered the most relevant.
To examine the sensitivity of our results with regard to the effects of errors in identification of muscle boundaries, we took an offset of the muscles on the PET images by 1 mm (both dilated and eroded). We chose 1 mm because this is the pixel size on the MRI on which the original muscle ROI were drawn. To perform the sensitivity study, we created 3D models of the muscles as triangulated surfaces (Stereolithography (STL) file format). Then, we computed new muscle masks from these STL. The dilated and eroded STL were then multiplied by the PET data (step 5 as described earlier) to yield a new set of results. The results in terms of %FDG uptake were virtually insensitive to this manipulation; the differences between the original, dilated, and eroded results were in the order of mere tenths of percents (0.0%–0.7%). This was true for all muscles, including the very smallest, where the offset of 1 mm had a relatively large effect in terms of volume (in the order of 30%–50%). This lends confidence that our method of volumetrically extracting %FDG uptake values from semiautomatically processed scans was sufficiently accurate.
Example PET and MRI scans with muscle ROI are shown in Figure 2.
Volumetric %FDG uptake
The muscles with the highest median %FDG uptake over all subjects were the soleus (17.1%), gluteus maximus (7.1%), vastus lateralis (5.8%), gastrocnemius medialis (5.6%), and adductor magnus (5.4%) (Fig. 3). In general, the larger muscles had higher %FDG uptake whereas the smaller muscles had lower uptake (Table 2).
In the hip region, besides the gluteus maximus, the gluteus medius and minimus also had high %FDG uptake (4.4% and 2.2%, respectively). The uptake in the gluteus medius, in particular, varied widely between subjects (2.5%–19.5%). The iliacus and psoas had lower %FDG uptake than the gluteal muscles (2.0% and 1.4%, respectively), and the uptake also varied less between subjects (1.2%–3.0% and 0.8%–1.7%, respectively).
In the thigh, the vastus lateralis (5.8%), adductor magnus (5.4%), vastus intermedius (4.3%), and vastus medialis (4.2%) had the highest median uptake. The FDG uptake in these muscles was relatively similar between subjects compared with that in the gluteal muscles.
In the lower leg, the soleus was by far the most active muscle (17.1%). The uptake in this muscle varied widely between subjects (8.8%–40.5%). The gastrocnemius medialis (5.6%) and tibialis anterior (4.8%) also exhibited a high median uptake. The uptake in the gastrocnemius medialis varied widely between subjects (1.9%–13.7%). The gastrocnemius lateralis had a much lower median uptake (1.9%) than the gastrocnemius medialis, and it also varied less (1.0%–4.6%).
The %FDG uptake data of all muscles in all subjects are available in the supplemental digital content (see Tables (page 1), Supplemental Digital Content 1, FDG Uptake Values and Standardized Uptake Values of All Muscles, https://links.lww.com/MSS/A484).
The muscles in the lower leg tended to exhibit greater SUV than those in the hip or thigh regions (see Tables (page 2), Supplemental Digital Content 1, FDG Uptake Values and Standardized Uptake Values of All Muscles, https://links.lww.com/MSS/A484); the mean SUV of all the muscles in the lower leg over all subjects was 2.04 g·mL−1, whereas it was 1.12 g·mL−1 in the hip and 0.76 g·mL−1 in the thigh. The tibialis anterior (3.20 g·mL−1), extensor digitorum longus (2.57 g·mL−1), extensor hallucis longus (2.39 g·mL−1), soleus (2.34 g·mL−1), and gastrocnemius medialis (2.31 g·mL−1) had the highest SUV overall. Several muscles that had high %FDG uptake, such as the gluteus maximus and the vastus lateralis, had comparatively low median SUV of 0.58 g·mL−1 and 0.64 g·mL−1, respectively.
The SUV data of all muscles in all subjects are available in the supplemental digital content (see Tables (page 2), Supplemental Digital Content 1, FDG Uptake Values and Standardized Uptake Values of All Muscles, https://links.lww.com/MSS/A484).
Differences between volumetric and single slice analyses
There were large differences between the %FDG uptake as measured volumetrically compared with the %FDG uptake measured using the single slice technique. The differences of the five muscles in each segment that had the largest %FDG uptake are shown in Figure 4.
In the hip region, the %FDG uptake in the gluteus medius and minimus was severely overestimated when only a single slice was analyzed (between 45.7% and 80.4% and between 65.9% and 123.5%, respectively), whereas the uptake in the psoas tended to be underestimated (between −95.0% and 13.8%).
In the thigh, the %FDG uptake was underestimated when analyzing only a single slice in all muscles shown in Figure 4. The underestimation was most severe in the rectus femoris (between −90.8% and −32.1%).
In the lower leg, we did not find consistent under- or overestimation in %FDG uptake, but the ranges of differences between methods were roughly similar to those observed in the hip and thigh. The gastrocnemius lateralis had a particularly wide range of between −53.9% (underestimation) and 56.2% (overestimation).
This study aimed to determine muscular activity during walking using a volumetric analysis of glucose uptake in entire muscles. The muscles with the highest median %FDG uptake were the soleus, gluteus maximus, vastus lateralis, gastrocnemius medialis and adductor magnus. We found a wide range of %FDG uptake values between subjects, including in some of the most important muscles involved in walking (e.g., soleus, gluteus medius, gastrocnemius medialis). Compared with the volumetric analysis, the single slice analysis did not yield an accurate estimate of the %FDG uptake in many of the most active muscles, including the gluteus medius and minimus (overestimated) and all the thigh muscles (underestimated).
The soleus and the gastrocnemius medialis were among the muscles that had the highest median uptake of FDG of the muscles in the lower limb, both in terms of %FDG uptake and SUV. As these muscles are both plantarflexors, this is consistent with literature stating that the generation of plantarflexion torque is key to forward propulsion and to prevent the body from falling forward while walking (19,24,36). The gluteus maximus and vastus lateralis also had high %FDG uptake values. These muscles are two of the primary hip and knee extensors, respectively. As such, they play a crucial role in providing support in the first half of stance (22,35). The vastus lateralis took up more FDG than the vastus intermedius and medialis because of its larger volume. Combined, the median uptake of all three vasti was 14.3%, which demonstrates their importance in the walking motion.
The adductor magnus, much to our surprise, was one of the most active muscles during walking (median, 5.4% FDG uptake). The adductor magnus also had the second highest SUV (0.75 g·mL−1) of the thigh muscles, indicating that its high %FDG uptake was not merely due to its large volume (509 mL). Although the temporal activation pattern of the adductor magnus during walking has been reported (using EMG) (13), the quantitative activity of this muscle has remained understudied (1). The consistently high %FDG uptake we observed in this muscle across subjects suggests that this muscle performs more work or exerts more force than commonly presumed. Previous work most often designated the hip abductors as the primary muscles controlling the hip in the frontal plane (e.g., (1,19,24)), with the adductors only tending to participate during the transfer phases between stance and swing (25). Therefore, we had expected that the largest hip abductor (gluteus medius, 4.4% FDG uptake) would be more active than any of the adductor muscles. Indeed, the gluteus medius had a higher median SUV (0.86 g·mL−1) than the adductor magnus (SUV, 0.75 g·mL−1) but the adductor magnus had a larger volume, resulting in higher %FDG uptake. This result demonstrates the added value of the FDG-PET technique in revealing the activity of an entire muscle during gait, which would have been difficult to do with traditional measurement techniques (e.g., surface EMG) that measure only a small part of the muscle. Although other studies that used FDG-PET did analyze the FDG uptake in the hip adductors (21,30,31), they examined the hip adductors only as a muscle group rather than differentiate between the muscles that comprise the group.
Like the adductor magnus, the gluteus minimus was surprisingly active. In spite of its small size (86 mL) compared with the functionally similar gluteus medius (330 mL), our results suggest that it made a large contribution to hip abduction and stabilization, as it took up roughly 50% of the FDG relative to the gluteus medius (Table 2). The gluteus minimus also had the highest SUV among all the muscles in the hip region (1.56 g·mL−1). This is in line with Oi et al. (21), who found that the SUV of the gluteus minimus during walking was the highest among the three gluteal muscles.
The iliacus and psoas are two muscles that flex the hip but have been difficult to examine with traditional methods (e.g., surface EMG) because of their anatomical position deep in the body. Although some authors have used intramuscular EMG to record the activity of these muscles, quantitative data about their contribution to walking have remained scarce (e.g., (2,25)). In this study, we found that the %FDG uptake values of the iliacus and psoas were 2.0% and 1.4%, respectively. Even when added together, our data suggest that the activity of these muscles is rather low in normal walking compared with the ankle plantarflexors and dorsalflexors, hip extensors and abductors, and knee extensors and flexors. This is in agreement with Perry and Burnfield (25), who stated that the iliacus plays a small role in normal walking, although walking faster or slower increases its importance.
The large differences between subjects in %FDG uptake values in some muscles could indicate that each subject activated his/her muscles in a unique fashion. This notion is supported by previous studies that used surface EMG; Arsenault et al. (3) found large and highly significant intersubject differences in the normalized (to maximum voluntary contraction) signal amplitude among healthy subjects, for instance in the soleus, rectus femoris, vastus medialis, and tibialis anterior. Yang and Winter (38) and Winter and Yack (37) found intersubject coefficients of variation that are in the same order of magnitude as ours (Supplemental Digital Content, page 1, https://links.lww.com/MSS/A484). Given that we specifically chose to include subjects with a wide range of body types in this study, several factors could explain the variance in %FDG uptake, for instance, age. Subject 8 provides an illustration of the effect that a relatively high age (60 yr) might have had. The soleus had the highest uptake of all muscles and in all subjects except subject 8 (data shown in Supplemental Digital Content, https://links.lww.com/MSS/A484). That subject concurrently had the highest %FDG uptake in the gluteus maximus among all subjects. Thus, this subject might have shifted from exerting ankle plantarflexion torque to exerting more hip extension torque to achieve forward progression. In addition, this subject had the lowest psoas and iliacus activity of all subjects as well as the second lowest summed activity in the quadriceps of all subjects. These combined findings suggest an “elderly walking pattern”; they match exactly a redistribution pattern of joint torques observed by DeVita and Hortobagyi (8) in elderly subjects when compared with that in younger subjects. However, we did not observe the same pattern in any of the other three relatively old subjects (age, 55–60 yr), so other factors (e.g., general physical fitness level) may also influence the degree to which each muscle is active during walking.
The volumetric and single slice analyses yielded systematically discrepant results in some of the muscles that were most active during walking (e.g., gluteus medius, vastus lateralis). Moreover, the muscles in which no systematic discrepancy was found (e.g., gastrocnemius lateralis, iliacus) still exhibited a rather large range (from −53.9% to 56.2% and from −18.9% to 37.5%, respectively) around the “ideal” 0% difference. These findings illustrate the added value of analyzing entire muscles rather than only single slices on the quantification of %FDG uptake values. For this comparison, we took slices taken at discrete heights on the basis of the literature (21,29–31,33). It may be argued that other slice heights might resemble the results more closely as obtained in the volumetric analysis. Therefore, in an additional sequence of assessments, we also examined the effects of taking slices at different heights but we were unable to find a single slice height in each segment that accurately represented the volumetric FDG uptake in all subjects.
Our study has some limitations. First, the muscle segmentation might have some errors despite extensive examination of the intermediate and end results. Despite the fact that the MRI scan protocol was specifically designed for imaging muscle boundaries, the boundaries of some muscles were difficult to distinguish. However, the sensitivity analysis indicated that the effects of these possible errors on the volumetric %FDG uptake were small. Second, we had no control condition such as similar measurements of FDG uptake after a resting period. This could have revealed whether the baseline FDG uptake was similar between muscles and subjects. Unfortunately, it remains difficult to extend this type of experiment with control measurements because of ethical issues regarding the radiation dose. Third, the comfortable walking speed was unique for each subject and this might have influenced the %FDG uptake values. The intensity level of the protocol might also have been experienced differently by different subjects, although none of the subjects indicated that it was too intense or asked to have the speed of the treadmill reduced. We deliberately chose comfortable walking rather than a fixed walking speed to prevent subjects from being “forced” to walk at an unnatural pace; consequently, this might lead to unnatural muscle FDG uptake.
Another limitation inherent to studies that use FDG-PET to determine muscle activity is the unknown relation between FDG uptake and the actual work performed or force exerted by a muscle. Muscle cells can use various substrates, including internal glycogen and triglyceride stores and plasma free fatty acids and glucose. The relative importance of these substrates depends on factors such as the intensity of the exercise (26), the remaining muscle glycogen stores (34), and the fiber type composition of the muscle (15). In our study, the exercise intensity was constant and rather low and the muscle glycogen stores were likely not depleted as a result. Thus, plasma glucose uptake likely remained constant (26) in our study. The fiber type composition might have an influence on FDG uptake; muscles that have a high proportion of Type I fibers tend to be more metabolically active during exercise (14). The concurrent validity, however, of studying walking by FDG-PET compared with EMG and kinematic and kinetic studies has previously been shown by Oi et al. (21). The present results corroborate these findings, as the %FDG uptake in several known prime movers of gait (e.g., soleus, gluteus maximus, vasti, gastrocnemius medialis) was high, which also corresponds well with the literature on contributors to joint torques and powers (1,19,36).
The volumetric analysis of FDG uptake in the muscles of the lower limb yielded an extensive overview of the most active muscles during walking. The most active muscles in terms of %FDG uptake generally corresponded well with literature that used other methods such as inverse dynamics and EMG. Several muscles (e.g., adductor magnus, gluteus minimus) that were difficult to measure with traditional methods (e.g., surface EMG) because of their large size or their deep-lying position in the body were found by FDG-PET to be very active during gait. The FDG uptake as measured in single slices did not correspond well to the FDG uptake obtained by volumetric analysis, which strongly illustrates the added value of our novel image analysis techniques. We intend to use these data to validate musculoskeletal models of these healthy subjects. In the future, it might be possible to analyze the gait of patients with orthopedic or neurological deficits or to enable the construction of patient-specific rehabilitation protocols. To enable such next steps in patient populations, the protocol may need to be shortened. The innovative FDG-PET method as described in this article will assist in future steps toward better understanding of musculoskeletal behavior of healthy subjects and patients.
We thank Dr. Toshihiko Fujimoto for useful comments regarding the design of the study.
We gratefully acknowledge financial support by the Seventh Framework Programme of the European Union for the TLEMsafe project (http://www.tlemsafe.eu).
The authors declare no conflicts of interest.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Anderson FC, Pandy MG. Individual muscle contributions to support in normal walking. Gait Posture
. 2003; 17 (2): 159–69.
2. Andersson EA, Nilsson J, Thorstensson A. Intramuscular EMG from the hip flexor muscles during human locomotion. Acta Physiol Scand
. 1997; 161 (3): 361–70.
3. Arsenault AB, Winter DA, Marteniuk RG. Is there a ‘normal’ profile of EMG activity in gait? Med Biol Eng Comput
. 1986; 24 (4): 337–43.
4. Burden A. How should we normalize electromyograms obtained from healthy participants? What we have learned from over 25 years of research. J Electromyogr Kinesiol
. 2010; 20 (6): 1023–35.
5. Carbone V, van der Krogt MM, Koopman HF, Verdonschot N. Sensitivity of subject-specific models to errors in musculo-skeletal geometry. J Biomech
. 2012; 45 (14): 2476–80.
6. Cherry SR, Sorenson JA, Phelps ME. Physics in Nuclear Medicine
. 4th ed. Philadelphia (PA): Saunders; 2012.
7. De Luca CJ. The use of surface electromyography in biomechanics. J Appl Biomech
. 1997; 13 (2): 135–63.
8. DeVita P, Hortobagyi T. Age causes a redistribution of joint torques and powers during gait. J Appl Physiol (1985)
. 2000; 88 (5): 1804–11.
9. Ericson MO, Nisell R, Ekholm J. Quantified electromyography of lower-limb muscles during level walking. Scand J Rehabil Med
. 1986; 18 (4): 159–63.
10. Fujimoto T, Itoh M, Kumano H, Tashiro M, Ido T. Whole-body metabolic map with positron emission tomography
of a man after running. Lancet
. 1996; 348 (9022): 266.
11. Fujimoto T, Kemppainen J, Kalliokoski KK, Nuutila P, Ito M, Knuuti J. Skeletal muscle glucose uptake response to exercise in trained and untrained men. Med Sci Sports Exerc
. 2003; 35 (5): 777–83.
12. Gondoh Y, Tashiro M, Itoh M, et al. Evaluation of individual skeletal muscle activity by glucose uptake during pedaling exercise at different workloads using positron emission tomography
. J Appl Physiol (1985)
. 2009; 107 (2): 599–604.
13. Green DL, Morris JM. Role of adductor longus and adductor magnus in postural movements and in ambulation. Am J Phys Med
. 1970; 49 (4): 223–40.
14. Heinonen I, Nesterov SV, Kemppainen J, Fujimoto T, Knuuti J, Kalliokoski KK. Increasing exercise intensity reduces heterogeneity of glucose uptake in human skeletal muscles. PLoS One
. 2012; 7 (12): e52191.
15. Henriksen EJ, Bourey RE, Rodnick KJ, Koranyi L, Permutt MA, Holloszy JO. Glucose transporter protein content and glucose transport capacity in rat skeletal muscles. Am J Physiol
. 1990; 259 (4 Pt 1): E593–8.
16. Inman VT, Ralston HJ, Todd F. Human Walking
. Baltimore (MD): Williams & Wilkins; 1981, p. 100–17.
17. Jakoby BW, Bercier Y, Conti M, Casey ME, Bendriem B, Townsend DW. Physical and clinical performance of the mCT time-of-flight PET/CT scanner. Phys Med Biol
. 2011; 56 (8): 2375–89.
18. Kalliokoski KK, Boushel R, Langberg H, et al. Differential glucose uptake in quadriceps and other leg muscles during one-legged dynamic submaximal knee-extension exercise. Front Physiol
. 2011; 2: 75.
19. Liu MQ, Anderson FC, Pandy MG, Delp SL. Muscles that support the body also modulate forward progression during walking. J Biomech
. 2006; 39 (14): 2623–30.
20. Masood T, Bojsen-Moller J, Kalliokoski KK, et al. Differential contributions of ankle plantarflexors during submaximal isometric muscle action: a PET and EMG study. J Electromyogr Kinesiol
. 2014; 24 (3): 367–74.
21. Oi N, Iwaya T, Itoh M, Yamaguchi K, Tobimatsu Y, Fujimoto T. FDG-PET imaging of lower extremity muscular activity during level walking. J Orthop Sci
. 2003; 8 (1): 55–61.
22. Pandy MG. Computer modeling and simulation of human movement. Annu Rev Biomed Eng
. 2001; 3: 245–73.
23. Pappas GP, Olcott EW, Drace JE. Imaging of skeletal muscle function using (18)FDG PET: force production, activation, and metabolism. J Appl Physiol (1985)
. 2001; 90 (1): 329–37.
24. Perry J. Kinesiology of lower extremity bracing. Clin Orthop Relat Res
. 1974; (102): 18–31.
25. Perry J, Burnfield JM. Gait Analysis: Normal and Pathological Function
. 2nd ed. Thorofare (NJ): SLACK; 2010 xvi, p. 551.
26. Romijn JA, Coyle EF, Sidossis LS, et al. Regulation of endogenous fat and carbohydrate metabolism in relation to exercise intensity and duration. Am J Physiol
. 1993; 265 (3 Pt 1): E380–91.
27. Rudroff T, Kindred JH, Benson JM, Tracy BL, Kalliokoski KK. Greater glucose uptake heterogeneity in knee muscles of old compared to young men during isometric contractions detected by [(18)F]-FDG PET/CT. Front Physiol
. 2014; 5: 198.
28. Scheys L. Personalized musculoskeletal modeling based on magnetic resonance images: use for biomechanical analysis of gait [dissertation]. Leuven (Belgium): K.U. Leuven; 2009.
29. Shimada H, Hirata T, Kimura Y, et al. Effects of a robotic walking exercise on walking performance in community-dwelling elderly adults. Geriatr Gerontol Int
. 2009; 9 (4): 372–81.
30. Shimada H, Kimura Y, Lord SR, et al. Comparison of regional lower limb glucose metabolism in older adults during walking. Scand J Med Sci Sports
. 2009; 19 (3): 389–97.
31. Shimada H, Kimura Y, Suzuki T, et al. The use of positron emission tomography
and [18 F]fluorodeoxyglucose for functional imaging of muscular activity during exercise with a stride assistance system. IEEE Trans Neural Syst Rehabil Eng
. 2007; 15 (3): 442–8.
32. Staudenmann D, Roeleveld K, Stegeman DF, van Dieen JH. Methodological aspects of SEMG recordings for force estimation—a tutorial and review. J Electromyogr Kinesiol
. 2010; 20 (3): 375–87.
33. Tashiro M, Fujimoto T, Itoh M, et al. 18F-FDG PET imaging of muscle activity in runners. J Nucl Med
. 1999; 40 (1): 70–6.
34. Wahren J, Felig P, Ahlborg G, Jorfeldt L. Glucose metabolism during leg exercise in man. J Clin Invest
. 1971; 50 (12): 2715–25.
35. Winter DA. Overall principle of lower limb support during stance phase of gait. J Biomech
. 1980; 13 (11): 923–7.
36. Winter DA. Energy generation and absorption at the ankle and knee during fast, natural, and slow cadences. Clin Orthop Relat Res
. 1983; (175): 147–54.
37. Winter DA, Yack HJ. EMG profiles during normal human walking: stride-to-stride and inter-subject variability. Electroencephalogr Clin Neurophysiol
. 1987; 67 (5): 402–11.
38. Yang JF, Winter DA. Electromyographic amplitude normalization methods: improving their sensitivity as diagnostic tools in gait analysis. Arch Phys Med Rehabil
. 1984; 65 (9): 517–21.