Abstract
Physical therapists must be able to determine the activity and passive properties of the musculoskeletal system in order to accurately plan and evaluate therapeutic measures. Discussed in this article are imaging methods that not only allow for the measurement of muscle activity but also allow for the measurement of cellular processes and passive mechanical properties noninvasively and in vivo. The techniques reviewed are T1- and T2-weighted magnetic resonance (MR) imaging, MR spectroscopy, cine–phase-contrast MR imaging, MR elastography, and ultrasonography. At present, many of these approaches are expensive and not readily available in physical therapy clinics but can be found at medical centers. However, there are ways of using these techniques to provide important knowledge about muscle function. This article proposes creative ways in which to use these techniques as evaluative tools.
Physical therapists must be able to use knowledge of the activity and passive properties of the musculoskeletal system in order to accurately plan and evaluate the efficacy of therapeutic measures. This knowledge can be obtained in at least 2 ways. First, physical therapists can use knowledge gained by motor systems researchers about the structure and function of the musculoskeletal system. Second, physical therapists can use (or have applied by other clinicians) contemporary methods to understand the anatomy and physiology of the musculoskeletal system. For the knowledge of the musculoskeletal system to be most relevant and practical, these musculoskeletal analysis techniques should be noninvasive.
In the past, the noninvasive techniques available for examining human neuromusculoskeletal function were limited. The gold standard for the noninvasive measurement of muscle activity in human subjects has been surface electromyography (EMG) recording, and this method is excellent for monitoring temporal information about muscle activity.1–3 Unfortunately, the actual source of the electrical activity being detected by surface EMG electrodes is vague. For example, the signal from deeply located muscles, such as the tibialis posterior muscle, is quite faint when it reaches the surface of the posterior calf skin. Uncertainty about the source of the signal limits the use of EMG data for clinicians using biofeedback or clinicians using drugs such as botulinum toxin, for which precise injection into the proper muscle or subvolume of a muscle is necessary for success. Although knowing the location of muscle activity is critically important, other, non–activity-related factors, such as the passive properties of muscle, affect movement and also need to be explored in a noninvasive manner. For example, if the target of a physical therapy intervention is the elasticity of tissue, then the therapist needs to target the correct tissue and be able to evaluate whether the intervention truly affected the targeted tissue.
It is hoped that this article will give both physical therapist researchers and clinicians a glimpse of the progress made and the challenges that remain for obtaining appropriate information about the functions of the musculoskeletal system needed for normal movement and adaptations that occur with disease or trauma. New information attained with imaging can be used in the pretreatment evaluation for determination of the therapeutic regimen and can contribute to the posttreatment outcome evaluation.
Noninvasive Approaches for Monitoring Musculoskeletal Function
The advent of modern imaging techniques offers a variety of approaches to physical therapy researchers and clinicians for monitoring musculoskeletal function and not just structure. This article focuses on several uses and forms of magnetic resonance (MR) imaging (anatomical imaging, mapping of T2 times, magnetic resonance spectroscopy [MRS], cine–phase-contrast MR imaging, and magnetic resonance elastography [MRE]) and ultrasonography. All of these techniques offer potential insights into the structure and function of the musculoskeletal system and, more importantly, provide clinicians with quantitative measures of adaptation of muscle function as an outcome measure of and a rationale for therapeutic interventions.
MR Imaging
An overview of magnetic resonance (MR) physics is presented in the article by Kimberley and Lewis in this Special Series. There are many ways in which to use MR imaging to examine structure and function. This article emphasizes the use of some basic approaches for examining structure and function while acknowledging that there are many other functional approaches, such as perfusion time course, diffusion time course, and blood volume time course, which are different from the blood oxygenation T2-weighted contrast used in functional MR imaging.
In brief, when material (or a human subject) is placed in an MR scanner bore, the atoms with odd-numbered protons begin to align and spin along the axis of the magnet. This spinning is called “precessing,” which is simply presented in Principles of Neural Science.4 If this spinning is thought of as something that has magnitude and direction, then it can be expressed in vector coordinates (Mx, My, and Mz). A brief radio-frequency (RF) signal sent into the bore will cause the protons to spin off axis (wobble) in 3 dimensions. This proton displacement in response to an RF pulse has a phase when the proton spins furthest off axis (decay phase) and a phase when the proton starts returning toward the bore axis (recovery phase). This wobbling creates a magnetic field, with each atom having its own unique spin frequency based on its local environment.
There are time constants related to the decay (relaxation) of the spin (T2) and the recovery (T1) toward realignment with the bore of the magnet. The relaxation and recovery times can be influenced by the pulse sequences. The pulse sequences are a group of RF magnetic field pulses, and the timing of these pulses causes protons to spin in ways that emphasize certain tissues or physiological processes. The time constants are calculated on a pixel-by-pixel basis within the scanner field of view (3-dimensional calculations are made on voxels). Thus, each tissue within a section of an MR image is made of many pixels (or voxels, when 3-dimensional) whose time constant value is determined on the basis of its environment and history. Magnetic resonance pulse sequences that are aimed at the T1 time constant produce the typical anatomical images seen every day in clinics. Magnetic resonance pulse sequences that are aimed at the T2 time constant can provide information about muscle activity or muscle activity history. One sequence used for examining muscle function is very similar to the blood oxygen level–dependent technique used in brain functional MR imaging and described elsewhere in this Special Series. However, many other approaches are discussed later in this article.
Comparatively simple reviews of MRI basics are presented in Principles of Neural Science,4 MRI: Basic Principles and Applications,5 and Functional MRI: An Introduction to Methods.6 There are, however, other T2-related constants. For example, T2*, which is shorter than T2, is influenced not only by spin-spin interactions but also by magnetic field gradient irregularities (homogeneity of the external field7). Thus, T2 is relatively fixed by the milieu, whereas T2* can vary. Because fluid is probably the source of T2 changes with muscle activity, T2* may be the most sensitive constant.
Relatively standard MR pulse sequences are potentially important in the assessment of musculoskeletal function and adaptation and have been used with increasing frequency by investigators. The next section presents a review of MR imaging approaches for studying muscle properties that may be transferred easily to the clinical setting. Combinations of these approaches can be used, and the following sections are organized arbitrarily.
Imaging for Assessment of Muscle Structure
Muscle Mass (Cross-Sectional Area)
T1-weighted images (and, in some instances, T2-weighted images) are very useful for determining muscle cross-sectional area noninvasively and in vivo.8 Cross-sectional area is closely associated with force output9; therefore, this approach can provide baseline information about a client's strength. However, the relationship between cross-sectional area and functional capabilities is not simple and, in some instances, may not be highly correlated.10–12 Still, it may be possible to monitor adaptations in muscle cross-sectional area that occur with disease, trauma, immobilization, rehabilitation, or exercise8 and to use these adaptations as a measure of therapeutic outcomes.
Fiber Type Distribution
Knowledge of muscle mass and how it changes with treatment, exercise, disease, or trauma is potentially important information. For example, a loss of muscle mass (sarcopenia) is a major problem in aging men. In order to determine whether exercise protocols are helping to control sarcopenia, it is necessary to assess regional muscle mass.13 However, at least as important as knowledge of muscle mass is knowledge of normal muscle fiber–type composition within muscles and how fiber type changes with exercise, disease, trauma, or immobilization. The 3 broad skeletal muscle fiber types are categorized approximately by fatigability, force output, and contraction time; these fiber types are not equally distributed in the named muscles.14 Standard methods for determining fiber types in vivo require invasively acquired biopsy specimens, which, by nature, have to be limited in number and distribution. The biopsy procedure itself probably causes local adaptation and, because so few areas are sampled, investigators cannot be confident about correctly determining the fiber type composition of the entire muscle that was sampled. This limited sampling can be a particular problem in muscles that may be compartmentalized15 or that are known to have a nonhomogeneous distribution of fiber types. Thus, noninvasive in vivo methods for determining muscle fiber type composition and distribution could have clinical importance. For example, if a patient or client needs to be trained in fatigue-resistant activities, it would be desirable to know the substrate that is being exercised or whether the fiber type distribution has changed with an exercise regimen. Unfortunately, only small steps in this direction have been made.
The relationship between T1 and T2 times and muscle fiber type distribution is complex and not fully understood. Because MR signals are strongly related to the histochemical composition of tissue, all tissues have specific ranges of T1 and T2 times (ie, bone is different from muscle). Even within a muscle, there may be differences in T1 and T2 times. For example, the resting T2 time for the biceps brachii muscle is shorter than that for the quadriceps femoris muscle.16 Similarly, the T2 time for the tibialis anterior (TA) muscle is probably different from that for the soleus muscle.
Kuno et al17 found a significant correlation between T1 and T2 times and fiber type for the vastus lateralis muscle. Houmard et al18 observed that the T1 (relaxation) time was correlated with the percentage of type I (slow-twitch) fibers for the lateral gastrocnemius (LG) muscle in humans, whereas there was no correlation between the T2 (relaxation) time and fiber type composition. In contrast, Le Rumeur et al19 found an increase in the T1 time with increased anterior thigh power output, a finding that would suggest that type II fibers are correlated with the T1 time. Parkkola et al20 found no correlation between T1 and T2 times and fiber types in cadaver multifidus and psoas muscles but did find that both T1 and T2 times were longer in the multifidus muscle than in the psoas muscle. Thus, there does not appear to be a clear relationship between T1 or T2 time and fiber type in humans. These studies were complicated by the choice of muscle and the problems inherent in the biopsy techniques or the use of cadaver material.
In 2 animal studies, there appeared to be a longer T2 time in slow-twitch fibers than in fast-twitch fibers (the rabbit study of Adzamli et al21 and the rat hind-limb study of English et al22). A clear relationship between fiber types and MR time constants has yet to emerge from these data. To further explain the relationship between fiber types and MR time constants, future studies with animal models, consistent experimental approaches, and more comprehensive numbers and functions of muscles chosen are needed. Recruitment of motor units usually follows the size principle,23 in which small motoneurons are recruited before larger motoneurons (ie, small motor units before larger ones). Because motor unit types and their constituent muscle fibers provide the core elements for the recruitment of force during functional tasks, knowledge of the distribution of muscle fiber types within muscles is important for understanding the results of motor control experiments, and using this knowledge is important for determining physical therapy intervention regimens.
Imaging for Assessment of Muscle Activity: Functional Muscle MR Imaging
Two recent review articles (Patten et al24 and Meyer and Prior25) examined the use of T2 times to study muscle activity. Several approaches for examining muscle activity are related to T2 times. These include multiecho sequences, T2* measurements, or fast techniques, such as echoplanar imaging (EPI), which also is very sensitive to T2* times. The advantage of multiecho sequences is that some anatomical resolution is retained, whereas the advantage of EPI sequences is less loss of signal, allowing one to investigate recovery from exercise.
Probably the key feature to extract from noninvasive in vivo approaches is determining which muscles or regions of muscles are active during functional tasks. In addition, approaches that allow for the serendipity of research would be desirable. For example, among the challenges of the use of EMG recording are the requirements to use a sufficient number of sensors (electrodes) and to place them properly in order to detect and analyze muscle activity accurately. That is, what can be recorded with EMG electrodes is determined mostly by the locations and recording areas of the electrodes. For discrete recording of regions of the leg, for example, many electrodes, whether surface or indwelling, are needed. With MR approaches, the entire volume of leg muscles can be recorded (scanned) simultaneously. Thus, finding unexpected (serendipitous) activity in muscles or combinations of muscles is probably more likely with MR approaches than with EMG recording. However, it is difficult to scan the leg and the shoulder at the same time; therefore, multisegment analysis currently is limited with MR approaches.
Muscle MR imaging with analysis of T2 times appears to be capable of examining the relative amount of activity of muscles or portions of muscles participating in a task. For example, in one study of ankle dorsiflexion, T2 (relaxation) times increased in a relatively linear manner with the amount of load.26 Thus, the measurement of T2 (relaxation) times could be used to examine the effect of resistance training on various muscle portions or groups. Like the report by Fisher et al,26 reports by Jenner et al27 and Disler et al28 supported a linear relationship between T2 times and exercise intensity and work level, respectively. Studies by Fleckenstein29 and Cheng et al,30 however, did not support a linear relationship between T2 times and work across all levels of activity but rather supported a sigmoid-shape relationship. Cheng et al30 suggested that at low levels of muscle fiber activation, there is minimal or no change in T2 times. However, the results reported by Yue et al31 supported the suggestion that changes in T2 (relaxation) times can be detected with as few as 2 repetitions of a task. My own data tend to support a threshold effect similar to that described by Cheng et al,30 in which a low level of muscle activity appears to be detected by EMG electrodes before it can be detected by MR imaging (Fig. 1).
The relationship between T2 time and effort may not be linear. There may be a threshold for change in T2 time and then a linear relationship with effort, followed by a plateau in T2 time. Thus, the composite relationship between T2 time and effort may be a sigmoid-shape relationship. On the other hand, electromyographic (EMG) activity has a linear relationship with effort.
The relationship between T2 times and EMG activity is less clear for several reasons, including the placement of EMG electrodes and the cumulative nature of MR imaging sampling for some T2 protocols. That is, standard multiecho T2 sequences may take approximately 6 minutes to scan a leg. Some investigators, such as Adams et al,32 studying the biceps brachii muscle found very high correlations between T2 times and integrated EMG data. The relationships among resistance, integrated EMG activity, and T2 times were assessed, and all were found to be significant.
Other investigators found significant but weaker relationships between EMG data and T2 times than Adams et al32 and did not find significant correlations for all leg muscles (eg, Kinugasa and Akima33). Several factors could contribute to these discrepancies among studies, including that different limb segments were being studied, surface electrodes may record biceps brachii muscle activity more specifically than the activity of some of the calf muscles, and there may have been functional heterogeneities in some of the muscles being studied. Indeed, the relationships between T2 times and the EMG activity of calf muscles differ depending on the study.33,34 One possible interpretation is that MR imaging analysis allows the detection of functional heterogeneities in muscles that are beyond the resolution of surface EMG electrodes. Another interpretation is that the underlying mechanisms for changes in T2 times simply are not correlated with the electrical activity of some muscles (note that Vandenborne et al35 showed with spectroscopy and T2* times that metabolic changes correlate well with T2* times).
It is clear that changes in T2 times with exercise are multifactorial, including fiber type distribution, differences in regional perfusion, and aerobic capacity. The factors affecting changes in T2 times with exercise in rats were explored and completely discussed by Prior et al.36 One important caution that they stated was that because different muscle fiber types have different mechanisms of changes in T2 times, a perfect correlation with recruitment will not occur. However, they acknowledged that this problem may not be as significant in humans because almost all muscles are a mix of fiber types. In addition, the time course and the persistence of the changes in T2 times recently were modeled by Damon and Gore.37
Types of T2-Weighted Imaging Sequences
In many T2 time studies, investigators used multiecho sequences and then fit the data to a curve to map the T2 time constant accurately.5 During multiecho sequence studies, multiple axial or sagittal sections can be sampled, and images with reasonable anatomical clarity can be produced. The disadvantage of this approach is that scans are performed exclusively after exercise and are the measure of the cumulative effects of exercise. Recently, fast imaging procedures (such as EPI) that are sensitive to T2* (typically used in functional MRI) were applied to muscle and provided significantly better temporal resolution (see the study of TA muscle dorsiflexion activity by Akima et al38). However, fewer sections can be sampled, and the sampled area therefore is smaller. Finally, the anatomical clarity is inferior to that obtained with multiecho sequences or T1 times.
Functional Experimental Results of T2-Weighted Imaging Studies
Several studies in which T2 times were used as a measure of muscle activity now have been published. Livingston et al39 demonstrated clearly the coordinated activation of forearm muscles necessary to effect different wrist movements (Fig. 2). As expected, activity related to wrist radial deviation was distributed among the extensor carpi radialis longus and flexor carpi radialis muscles and possibly the pronator teres muscle. However, the highlight of the technique is its spatial resolution. Livingston et al39 showed no changes in T2 times for the brachioradialis muscle during wrist movements, whereas some cross talk from the brachioradialis muscle normally would have been expected if surface EMG recording of the extensor carpi radialis longus muscle had been used (Fig. 2). Similarly, MR imaging allows the confident sampling of smaller leg muscles, such as the peroneus longus muscle. Some recent studies of the peroneus longus muscle with MR imaging demonstrated this muscle to be very active in plantar-flexion tasks.34,40
T2-weighted magnetic resonance images across sections for the same exercise task (radial deviation at 60% of the maximal voluntary contraction; sections 4–10 are distal to proximal). Note that the extensor carpi radialis longus (ECRL) and extensor carpi radialis brevis (ECRB) muscles showed much greater activity (lighter in image) than the adjacent brachioradialis and extensor digitorum communis muscles, which were not active. EDC=extensor digitorum communis, FCR=flexor carpi radialis, FDP=flexor digitorum profundus. (Reprinted with kind permission of Springer Science and Business Media from Livingston BP, Segal RL, Song A, et al. Functional activation of the extensor carpi radialis muscles in humans. Arch Phys Med Rehabil. 2001;82:1164–1170.)
Several other studies also investigated task-specific muscle activation patterns with the mapping of T2 times. For example, Richardson et al41 conducted an experiment in which participants performed traditional knee extension exercises in one condition and then lower-limb cycling for comparison. Both of these conditions obviously involved knee extension, but very different patterns of activation based on the task were noted. Surprisingly, the uniarticular vastus muscle showed similar levels of activation in both tasks, whereas the biarticular rectus femoris muscle was much more active during the knee extension exercises. My research group also has found that T2 times are effective indicators of changes for defining the coordinated activation of leg muscles during motor tasks. For example, Segal and Song40 reported longer T2 times for the medial gastrocnemius (MG) muscle than for the LG muscle during a unilateral heel raise task, whereas Giordano and Segal42 reported that during a supine plantar-flexion task, the LG muscle was more active.
One of the potentially important clinical uses of precise spatial information about muscle activity is to provide appropriate biofeedback regarding muscle activity to patients with difficulty controlling muscle activation (eg, patients after stroke or spinal cord injury [SCI]). Studies done over the last 20 to 30 years revealed that some muscles may have functional compartments (reviewed by English et al43). Can MR imaging be used to examine this type of organization?
Prior et al16 searched for regionally localized activity within several muscles by examining the variance of T2 times across increasing levels of effort. Although there were qualitative increases in the variance of T2 times within histograms, none of these was statistically significant. These researchers used Monte Carlo simulations to show that under certain circumstances, such as sprouting with denervation, the variance approach could detect anatomically clustered regions of motor unit territories within a muscle. Indeed, Hillegass and Dudley44 showed clustering of motor unit activity following electrical stimulation in patients with SCI. Moreover, Adams et al45 reported regional clustering of changes in T2 times with electrical stimulation in subjects who were healthy.
One important concern that was particularly well addressed in the study of Prior et al16 is that each pixel does not represent a single muscle fiber. A pixel overlies tens of muscle fibers, so that each pixel overlies muscle fibers from multiple motor units. Thus, detecting the differential activation of motor units within the same general anatomical area is virtually impossible. However, if there is an anatomical localization of function in a muscle, as with compartments, the mapping of T2 times should be successful. Indeed, several investigators have reported the regional localization of function.38–40,42 Using EPI scanning, Akima et al38 showed an interesting phenomenon within the TA muscle during dorsiflexion. During the experiment, there were different levels of signal intensity in the superficial (anterior) and deep (posterior) portions of the TA muscle. These areas appear to correspond to compartments of the TA muscle identified in cadaver dissections by Wolf and Kim.46 Damon et al47 used a cluster analysis approach to search for the spatial localization of T2 time changes and found that their approach could qualitatively reveal spatial localization like that seen by Akima et al.38 A note of caution is that the study of Akima et al38 revealed that the changes observed with MR imaging may have complex mechanisms, as discussed earlier in this article.
In summary, the mapping of T2 times has been used to examine the coordination of limb muscles with reasonable success, particularly at the whole-muscle level. Although there are restrictions related to temporal resolution and fiber type, the described techniques offer distinct advantages over surface EMG recording for spatial localization. It is possible under certain circumstances for T2 time mapping to reveal the spatial localization of changes in activity within portions of muscles. Moreover, because many of these MR approaches use routine sequences, they are potentially available in local radiological clinics with scanners. However, analysis of the data is somewhat complex and may require collaboration with specialists outside local radiological clinics.
Adaptation of Muscle Activity
The availability of scanners to clinicians makes MR imaging a viable tool for assessing patients’ baseline muscle function and their progression with a therapeutic intervention or regression because of a disease process or immobilization. Although the cost of doing these muscle tests as stand-alone tests is currently high and not reimbursable, if their benefit can be established carefully with research, muscle activity imaging may be achievable, particularly if merged with other imaging sequences.
One potentially important use for MR imaging is monitoring of the adaptation of patterns of muscle activation in response to different insults, ranging from fatigue to central nervous system injuries or disease to therapy. In one human study of adaptation, the use of electrical stimulation to fatigue the vastus lateralis portion of the quadriceps femoris muscles resulted in an adaptive activation of the other quadriceps femoris muscles, allowing for the successful completion of a knee extension task.48 These results are similar to those obtained from animal studies that also demonstrated the development of adaptive mechanisms in muscles denervated by nerve transection or botulinum toxin exposure.49,50 As in nonhuman studies, uncertainties remain as to how similar the movements were before and after adaptive muscle activation. That is, were the same forces and kinematics achieved after adaptation as were used before adaptation? For MR imaging data or any other data on adaptive muscle activation to be valid, investigators will need to have precise behavioral and kinetic data available. The T2 time mapping approach can be very important in these studies of adaptation of patterns of muscle activation because electrode placement is not an issue. Thus, changes in activity can be detected in muscles that were not anticipated to be important for an experiment or in the clinic.
Other Imaging Approaches for Studying Cellular and Mechanical Muscle Function
There is important information about muscle function that cannot be captured by the mapping of T2 times and EMG recording. Discussed here are other imaging approaches that supply useful information about cellular and mechanical functions that complements or supplants the information gained from EMG recording and T2 time mapping.
MRS (or Nuclear MRS)
Magnetic resonance spectroscopy (or nuclear MRS) offers the opportunity to explore changes in the cellular milieu brought about by muscle activity. Magnetic resonance spectroscopy was first discovered in the middle 1940s independently by future Nobel laureates Felix Bloch51 and Edward M Purcell.52 The technique quickly became useful for the analysis of individual molecules or molecules in solution. In 1986, Radda53 showed how MRS could be used with whole-body magnets. The emphasis in MRS is to search for resonant frequencies of protons of different molecules within the same large magnetic field. These differences in frequencies are referred to as chemical shifts. Magnetic resonance spectroscopy allows an investigator to obtain metabolic information, including the concentration and molecular environment of a metabolite, from a localized area of sample tissue. A standard chemical species is used as a reference, and the differences in resonant frequencies between a molecule being studied and this reference are expressed in parts per million.
Multiple MRS protocols can be used to examine muscle metabolism; these include 31P MRS, 1H MRS, and 13C MRS (reviewed by Prompers et al54). For example, 31P MRS can allow for the examination of adenosine triphosphate (ATP) metabolism. Different peaks in the chemical shift histogram are shown in Figure 3 for different phosphorous-containing metabolites found with 31P MRS. 1H MRS can be used to assess lactate formation, tissue oxygenation, and intramuscular lipid content, among other measurements. Investigators use 13C MRS for a variety of purposes, but one important use is the measurement of glycogen content and synthesis. Indeed, Taylor et al55 suggested that 13C MRS is comparable to, if not better than, needle biopsy of muscle. These powerful approaches allow researchers and physical therapist clinicians to make judgments about the physiological status of muscle in clients or patients and to observe how physiological status may change with therapeutic interventions.
Magnetic resonance spectra obtained with a surface coil in 3 minutes from the lateral gastrocnemius (LG), medial gastrocnemius (MG), and soleus (SOL) muscles. ATP=adenosine triphosphate, Pi=inorganic phosphate, PCr=phosphocreatine, γ=gamma-ATP peak, UN=unlocalized. (Reprinted with kind permission of Springer Science and Business Media from Vandenborne K, Walter G, Ploutz-Snyder L, et al. Relationship between muscle T2* relaxation properties and metabolic state: a combined localized 31P-spectroscopy and 1H-imaging study. Eur J Appl Physiol. 2000;82:76–82.)
The area sampled in MRS is dependent on the sequences (hard pulse versus adiabatic) and type of coil (surface or volume) used. The hard pulse is a brief strong pulse that is used to excite a large area and can result in a uniform excitation if used with a volume coil. It may not result in a uniform excitation if used with a surface coil. The advantage of the adiabatic pulse is that it is a relatively long pulse that does not require a uniform transmission field, like that needed with a surface coil. The surface coil is highly sensitive, but only over a small area, whereas a volume coil is less sensitive but can be used to excite a larger area.
One example of the usefulness of spectroscopy is the study of the physiology of fatigue. With EMG electrodes, the signal can be used to search for changes in firing frequency and hence fatigue but cannot define changes in the ionic or chemical milieu of the muscle fibers that may contribute to fatigue and help differentiate between central fatigue and peripheral fatigue. 31P MRS has been successfully used to measure phosphorus-containing metabolites (inorganic phosphates) created by muscle during fatiguing protocols (reviewed by Bendahan et al56 and Roy57). This type of knowledge can be used to understand the mechanisms of disease and the functional capacity of patients. For example, Ljungberg et al58 used MRS to evaluate fatigue in patients who had had poliomyelitis. They used 31P MRS with a surface coil and an adiabatic RF pulse to examine the TA muscle in patients and in volunteers who were healthy. The patients who had had poliomyelitis had appropriate EMG changes related to poliomyelitis, such as large motor unit potentials attributable to sprouting. The main MRS findings obtained with fatiguing exercise were many individual variations and higher inorganic phosphate and inorganic phosphate/phosphocreatine values for patients during the recovery phase.
Magnetic resonance spectroscopy also may be useful for evaluating fiber types within muscles and how they are used during particular tasks.59 Using MRS, Crowther and Gronka59 found that task requirements were related to different values for the reciprocal of the time constant of phosphocreatine (KPCr) recovery that they suggested were related to the muscle fiber types recruited within the TA muscle (please note that Crowther and Gronka59 considered the TA muscle to be a “mixed” muscle, but data from Gregory et al60 classify the TA muscle as a “slow” muscle). Although these effects of different muscle fiber types may make it complicated to obtain an overall view of the metabolism of a muscle, this type of analysis may allow clinicians and researchers to verify how a subject is performing a task, to obtain an idea of the functional (fiber type) composition of a muscle, or both. In addition, it is very likely that MRS provides a better regional estimate of fiber types and metabolism than invasive biopsies. “Functional biopsies” were done with MRS to show that highly trained athletes who perform either endurance or sprint running did not have that diverse of a composition of muscle fiber types.61
Another example of the use of MRS is related to the lipid content of muscles. Magnetic resonance spectroscopy was used to determine the amount of intramyocellular lipid content (IMCL), which can be determined with either 1H MRS or 13C MRS, although 1H MRS is used more commonly. 1H MRS revealed that there is a negative correlation between IMCL and the degree of insulin resistance at rest in patients with diabetes.62 Interestingly, not all muscles normally have the same amounts of IMCL or total lipid contents.63 The soleus muscle has the most IMCL, whereas the gastrocnemius muscles have none and the peroneus muscle has the largest total amount of lipids. These details about the compositions of individual muscles may provide important information about changes in muscle with disease, rehabilitation potential, or both. More information about the baseline capabilities of patients certainly may be obtained. In addition, because these are in vivo measurements, clinicians can obtain data specific to an individual.
Magnetic resonance spectroscopy also can be used to assess changes in ATP-producing pathways. Recently, Lanza et al64 examined age-related changes in these pathways by using 31P MRS. Interestingly, although there were some changes with age, there were some similarities between younger and older people, and again there were many individual variations. Oxidative capacity was unaltered with age, whereas peak glycolytic flux and overall ATP production from anaerobic glycolysis were lower in older men during high-intensity contractions. These measurements are made continuously; whereas muscle biopsies provide only a transitory view of ATP synthesis.
These are but a few examples of how noninvasive in vivo images can provide excellent cellular information related to muscle functional capacity. However, some limitations of these techniques must be considered. The MRS approach resembles obtaining a real-time biopsy specimen from the muscle, although it is noninvasive. That is, information may be gained from a relatively small portion of a whole muscle (however, see Vandenborne et al35). This is a particularly important limitation if the muscle being studied is nonhomogeneous in fiber type composition or is anatomically compartmentalized. The combination of MRS and surface EMG recording may help to overcome this limitation somewhat because spectral changes in EMG correlate with spectroscopy findings. Finally, Vandenborne et al35 combined MRS with the mapping of T2 times and found that this combination showed great potential for simultaneously revealing the distribution of activity and metabolic changes.
Cine–Phase-Contrast MR Imaging
Cine–phase-contrast MR imaging is an approach used to synchronize MR imaging data acquisition with motion cycles to allow the measurement of particle velocity in 3 dimensions.65,66 In cine MR imaging, MR imaging data acquisition is synchronized with motion cycles so that tissue movement can be captured,67 whereas phase-contrast MR imaging allows the measurement of 3-dimensional particle velocity.68 For example, the technique can be used to visualize the differential displacement of muscle fibers within a muscle.69 Pappas et al69 showed the differential movement of muscle fibers within the biceps brachii muscle during repeated flexion. Specifically, muscle shortening was uniform along anterior muscle fascicles and nonuniform along centerline fascicles. However, Pappas et al69 did not take into account data from physiological and anatomical studies that suggested a possible medial-to-lateral organization of the recruitment of motor units for functional tasks.70–72
The cine–phase-contrast MR imaging approach not only can show the differential movement of muscle fibers within a given muscle but also can provide excellent architectural information. For example, Finni et al73 were able to show that the human soleus muscle is not architecturally homogeneous; the architecture of the proximal soleus muscle differs from that of the distal soleus muscle. Moreover, these differences appear to be associated with nonuniform strains within the aponeurosis-tendon complex.74 Differential activation by the nervous system of these different regions of the soleus muscle may have functional consequences. In most instances, attaining this kind of information is beyond routine clinical practice because of expense and inconvenience. However, in the foreseeable future, this information may be important for the placement of electrodes that are part of a neural prosthesis to aid locomotor recovery in people with SCI. In such a case, it would be money well spent to be able to place electrodes precisely in order to elicit the desired movement.
MRE
Approaches in which clinicians can measure simultaneously both passive and active properties of muscles are desirable. Magnetic resonance elastography is a method that allows for the estimation of the mechanical properties of tissues during both passive and active movements. The process involves phase-contrast MR techniques with oscillating motion-sensitive gradients within the sequence.75 Essentially, investigators are able to map the response of tissue to a mechanical perturbation. The mechanical perturbation can occur with the muscle relaxed or during active muscle contraction. The outcomes measured include wavelength and shear modulus (ratio of shear stress to shear strain). The longer the wavelength (mechanical waves traveling faster) of the tissue response, the stiffer the material being perturbed.76 Thus, in theory, one should be able to measure muscle stiffness or elasticity in vivo. Moreover, this approach can be used to define normal muscle elasticity and to quantify changes in muscle in response to altered use patterns, trauma, or a disease process. For example, the MRE wavelength for spastic muscle should be longer than that for normal muscle, and the wavelength should be shortened by successful therapeutic interventions. This approach clearly is preferable to more invasive approaches for measuring in vivo force, such as the surgical implantation of buckle transducers.77
Most of the work with MRE in humans has been done at the Mayo Clinic and Foundation (for a more specific review of the technique, see Manduca et al78). Jenkyn et al79 demonstrated that MRE is able to measure muscle tension in both the active (isometric contraction) and the passive states. Much of the work was performed with leg muscles, but other studies79,80 have shown that similar results can be obtained with upper-limb muscles, such as the biceps brachii muscle.75 These studies have been extended by Basford et al81 and are addressed later in this article.
Active contraction of muscle fibers apparently can be detected with MRE. In a recent study, Heers et al80 used MRE to measure the active resistance of subjects to imposed plantar flexion and dorsiflexion moments. The same subjects then underwent EMG recording of the same muscles as those evaluated with MRE, but outside of the scanner. Surface EMG recordings of the TA, LG, and MG muscles were obtained, and fine-wire electrode recordings of the soleus muscle were obtained. The correlation between MRE wavelengths and electromyographic activity for the respective muscles was very high (R2=.82–.90). Because EMG recording currently is considered the gold standard for measuring muscle activity, these findings suggest that MRE is a valid approach for the noninvasive assessment of muscle properties in the context of muscle activity.
The MRE approach does not sample the entire muscle; therefore, the high correlation with electromyographic activity could be an artifact. Thus, the issue of whether whole-muscle effects (instead of subregion effects) can be tested with MRE needs further research.
Ultrasonography
In the United States, imaging with ultrasonography has not been used extensively for the peripheral neuromuscular system; however, it has been used extensively in Japan and Europe. In particular, Fukunaga and colleagues82–84 have used ultrasonography to great advantage in examining musculotendinous structure and movement. This approach does not require a special room, as MR imaging does, but requires only an ultrasonography transducer, sophisticated analysis software, and, most importantly, skillful application of the transducer. The technique is completely noninvasive, is carried out in vivo, and is essentially real-time. This approach may have great potential for clinical and research uses. The in vivo measurements appear to correlate well with in vitro measurements, suggesting that the technique has validity.85 Ultrasonography can help to determine changes in tendon stiffness attributable to an exercise program; such findings ultimately have both clinical importance and research importance. Kubo et al86 used ultrasonography in conjunction with electromyographic activity measurements to determine mechanical and functional changes attributable to an exercise regimen. The results provided important movement control information while at the same time suggesting the mechanism for functional changes by providing in vivo mechanical information. Recently, Bojsen-Moller et al87 observed different shear forces for the soleus muscle and the MG muscle during isometric plantar flexion when the knee angle was varied from extension to flexion. Differences observed between active trials and passive trials suggested that the differences were attributable to different force outputs of the muscles. These data reinforce the idea that there is a technique capable of differentiating the function of the triceps surae muscle into constituent muscles, suggesting that this muscle should no longer be viewed as a single functional unit.
Clinically Relevant Uses of Imaging
Choice of Exercise
If MR imaging were readily available in a clinic, one could potentially use the technology to determine whether the exercise chosen was appropriate to activate the targeted muscles. For example, Takeda et al88 used pre- and post-exercise T2-weighted images to determine the most effective exercise for strength ening the supraspinatus muscle. These authors hypothesized that a particular exercise and position would best recruit the supraspinatus muscle. The literature on this topic, based on the use of EMG recording, was equivocal. What they discovered was that the position of the subject was critical to best activate the supraspinatus muscle. This example illustrates the potential therapeutic use of MR imaging for selecting an exercise protocol.
Neuromuscular Dysfunction
Castro et al89 used T1-weighted MR imaging to calculate the loss of muscle cross-sectional area in humans 6 months after complete SCI. There were significant reductions in the cross-sectional area of the triceps surae muscle but not the TA muscle. This differential atrophy from denervation was seen easily with MR imaging but would not have been detected easily with other techniques. Moreover, this finding is clinically important because therapists will need to take into account an imbalance in force production between muscles and tailor treatment plans to strengthen the weaker of the muscles.
Using T1-weighted images, Modlesky et al90 documented muscle loss in the thighs of patients with SCI and reported that muscle was the main target of nonfat tissue loss. In addition, T1-weighted images have been able to show that as muscle mass is decreased with SCI, intramuscular fat is increased91; these data indicate that patients after SCI not only have less force capability but also may be more susceptible to type II diabetes, a finding that appears to be related to intramuscular fat content.92
Hillegass and Dudley44 determined the cross-sectional area of the quadriceps muscle after SCI by using T2 time mapping sequences. Using electrical stimulation to activate motor units, they also determined that the distribution of T2 time increases differed from that of subjects who were healthy because the pixels with increased T2 times were more clustered. In addition, although large portions of the muscle were activated by the stimulation, relatively little torque was produced, perhaps because patients with SCI have small muscle fibers. These types of noninvasive in vivo studies allow a physical therapist the potential to assess the baseline level of muscle properties and the changes in muscle function that therapy may effect in patients with SCI.
In addition to allowing examination of the distribution of muscle unit clusters in patients with neurological conditions, MR imaging may be a clinically useful way of measuring changes in mechanical properties in patients with neurological conditions, such as spasticity. Using MRE, Basford et al81 examined the elastic properties of muscles in both control subjects and patients with neurological dysfunction. These investigators applied a 150-Hz shear wave by means of a nonmagnetic mechanotransducer. They sampled the responses of various leg muscles (TA, LG, MG, and soleus muscles) to this mechanical perturbation with the leg at rest, during resistance to dorsiflexion moments, and during resistance to plantar flexion moments in control subjects and in a small convenience sample of patients with lower-limb neuromuscular dysfunction. The main findings were that the wavelength and the shear modulus were both greater in the patient group, suggesting that the muscles in the patients were more stiff than those in the control subjects.
The value of this noninvasive in vivo approach is that it may provide clinically useful ways of measuring spasticity and the responses of spastic muscles to specific therapeutic approaches. One note of caution is that although noninvasive in vivo measurements were being made, the subjects were not performing a normal functional task, such as walking. However, the information gleaned from this type of study represents a major step forward in defining outcomes and the usefulness of therapeutic interventions. In addition, it may be possible to push the limits of the technique to determine whether regions within a muscle respond differently or if different muscles respond differently. For example, the wavelength and the shear modulus of the MG muscle were greater than those of the LG muscle. Finally, a constant issue in assessing spasticity is whether passive or active processes are being evaluated. The combination of MRE and EMG recording may allow this issue to be understood better. In particular, the addition of MRE may allow clinicians to determine which specific muscles or portions of muscles are spastic in order to guide a physical therapist intervention or to guide a physician who needs to inject botulinum toxin accurately.
The mapping of T2 times may be an excellent tool for investigating the distribution of muscle activity following nervous system lesions, limb immobilization, and therapeutic interventions. The work of Akima et al48 offers great promise for examining such adaptations. T2 time mapping allows the examination of adaptations even within subregions of muscles.38–40 Indeed, as shown in the review by Meyer and Prior,25 clumping of territories of muscle activation can be shown in patients who have experienced a peripheral nerve lesion. This finding is similar to that of Hillegass and Dudley,44 who used electrical stimulation in patients with SCI. Finally, the work of Vandenborne et al35 suggests that cellular processes detected by MRS could be recorded simultaneously with T2 time mapping with some spatial restrictions.
Metabolic Diseases
Magnetic resonance spectroscopy may be used for patients demonstrating metabolic dysfunction, such as patients requiring dialysis93 or patients who have diabetes, for whom information about exercise capacity would be crucial to treatment planning by a physical therapist. Magnetic resonance spectroscopy data can be used to estimate the capacity of patients to carry out particular types of exercise and possibly to help determine whether they have reached a new capacity during the progression of rehabilitation.
Immobilization and Musculoskeletal Injuries
Cross-sectional area has been faithfully measured on T1-weighted images of immobilized limbs.8 It has been shown by numerous investigators that immobilization and disuse reduce the cross-sectional area of muscles, something easily measured with MR imaging. Using MR images as the outcome measure, Akima and Furukawa94 found that partial meniscectomy induced atrophy in the quadriceps muscle group on the side of surgery. Neither the hamstring muscles nor the adductor muscles were found to be atrophied. Further, atrophy was noted across all of the quadriceps femoris muscles. In a 2004 study, Kawashima et al8 showed that intensive exercise and hypergravity during 20 days of bed rest prevented a loss of muscle bulk. Phase-contrast and spin-tag techniques95 also have been used to study atrophied muscle and the ability of specific therapeutic interventions to reverse this atrophy. Thus, multiple imaging procedures can assess the effects of immobilization and the strategies used to minimize the negative effects of immobilization.
Using cine–phase-contrast MR imaging, Finni et al96 were able to detect differences in the velocity of contraction among leg muscles in control subjects and patients who had undergone Achilles tendon rupture repair. By measuring the velocities of contractions of the distal ends of the MG, soleus, and flexor hallucis longus muscles, they were able to determine that there were normally significant variations in the relationship of these contraction velocities but, importantly, that a clear adaptive behavior was used by patients. The patients used the flexor hallucis longus muscle more and continued to use this activity pattern even while increasing ankle plantar-flexion torque with rehabilitation. This is yet another example of the variety of information that can potentially help physical therapists shape or modify their treatment plans. Ultrasonography also can be used to make in vivo calculations of tendon stiffness and can be used as a measure of adaptation or functional capability.97 This knowledge is important for motor control studies and, more importantly, for the clinical assessment of patients, such as those with Achilles tendon tears and other tendon injuries, allowing a clinician to plan a therapeutic regimen that will not compromise a patient's recovery.
Future Challenges
The techniques described in this article offer an exciting future for research and the evaluation of the musculoskeletal system in humans. Some of the techniques discussed in this article are quite costly relative to standard physical therapy evaluative tools. However, these approaches, particularly in combination, provide important noninvasive in vivo information about human movement and potentially important clinical information. Thus, a challenge that physical therapists face as the field of physical therapy becomes a doctoring profession is being able to “prescribe” the use of evaluative tools, such as imaging technologies, which have long been considered out of the purview of physical therapists. The simplest way to be able to overcome this challenge is to provide a clear rationale and scientific basis for why having the information from muscle imaging improves outcomes and can affect physical therapist practice.
It is doubtful that advanced imaging procedures will take place at local musculoskeletal physical therapy clinics, but now is the time to begin addressing problems from a new perspective without preconceptions. As movement scientists and clinicians, physical therapists should have the will to adopt a new perspective and become part of evaluative groups and centers that carry out comprehensive musculoskeletal evaluations of clients or patients. These centers then can provide practicing therapists with information to help develop treatment regimens and then quantitatively evaluate whether the regimens influenced the targeted tissues as planned. An early alternative to evaluative centers might be negotiating with imaging facilities to, for example, add a test to a usual anatomical MR imaging session. However, an evaluative group or center would provide the best information needed for an individual patient because of the availability of multiple evaluative tools, including imaging. Remember that no one technique is the best technique and no 2 patients are the same. The profession of physical therapy will be challenged to provide appropriate advanced training in imaging for practicing physical therapists and to incorporate new training in imaging into entry-level curricula to take full advantage of these imaging technologies.
Physical therapists should be encouraged by the fact that multiple quantitative, noninvasive, and in vivo approaches are now available for evaluating musculoskeletal function. As evidenced by the other articles in this Special Series, much of the publicity related to novel imaging approaches has focused on imaging of the brain, but the advances in studying muscle function have been quite remarkable. Physical therapists also must become involved in the evolution of new imaging technologies that will provide the most relevant and cost-effective information about muscle function. It has been said that “seeing is believing,” and with a little caution I am beginning to believe what I see.
Acknowledgments
The author acknowledges grant support from the National Center for Medical Rehabilitation Research, National Institute of Child Health and Human Development, National Institutes of Health (grant HD 32571).
Footnotes
-
For more about this PTJ Special Series on the role of neuroimaging in rehabilitation, read the editorial by Richard K Shields on page 639.
- Received June 16, 2006.
- Accepted January 10, 2007.
- Physical Therapy