Probing muscle recovery following downhill running using precise mapping of MRI T2 relaxation times

Postexercise recovery rate is a vital component of designing personalized training protocols and rehabilitation plans. Tracking exercise‐induced muscle damage and recovery requires sensitive tools that can probe the muscles' state and composition noninvasively.


INTRODUCTION
Postexercise recovery rate is a vital component of planning training protocols and essential for maintaining high-level performance and preventing overreaching, overtraining, and injuries. 1 During recovery, while muscles strive to restore homeostasis, 2 they temporarily lose a certain level of muscle function and capacity. 3 Sufficient recovery will lead to favorable adaptations including muscle remodeling, improved function of the skeletal muscle, and enhanced performance. 2 The rate of recovery following exercise-induced muscle damage (EIMD) has been studied in relation to training experience, 4 age, 5 and gender, 6 while many studies explored efficient ways to facilitate recovery and to shed light on the individual recovery mechanisms. [7][8][9] These studies notwithstanding, the variability in recovery rate between and within participants remains uncertain. More in-depth and accurate tools for tracking muscle state are therefore needed to increase the sensitivity to early/pre-symptomatic muscle damage and facilitate muscle recovery after acute damage. The severity of EIMD is influenced by training intensity, 10 duration, 11 and the type of muscle contraction (e.g., concentric, eccentric). 12 Eccentric contractions produce force by muscle lengthening, 13 causing relatively high mechanical stress on the involved structures and higher disruptions to the sarcolemma and the extracellular matrix. 14 Accordingly, eccentric protocols are used widely in studies of muscle damage and recovery. 3,15,16 Physiologically, aerobic-based EIMD involve a cascade of microstructural processes. [16][17][18] During excessive eccentric contractions, high mechanical stress is placed on the involved tissues, causing myofibers tears, damage to structural proteins in the extracellular matrix, as well as to myofibrils and cell membranes. EIMD also leads to high permeability of the capillaries, osmotic shift, and leakage of muscle proteins to the blood flow. 13,[19][20][21] Additionally, as part of the inflammatory reactions, a variety of immune cells are recruited to the site of injury, resulting in local muscle edema, 17 while phagocytic white cells such as neutrophils accumulate in the muscle as early as 1-2 h following exercise. These have a critical role in healing and removal of necrotic debris, 9 yet they can potentially trigger further cell damage by releasing oxygen free radicals, lysosomal proteases, and elastases. 22 This represents a secondary injury process, which can involve collateral damage to healthy cells that were not injured during the initial stimulation. 23 Assessment of muscle recovery following EIMD is usually done by collecting blood markers, 24 using pain perception questionnaires, 25 and measuring stride length and force production. 16,26 MRI is another effective tool in providing region-specific assessment of muscle state, thus allowing us to identify the location of the injured area. MRI's transverse (T 2 ) relaxation time constant is highly sensitive to the tissues' water content, biochemistry, and microarchitecture at the cellular level. Specifically, this parameter increases as the water content increases, 27 as free water has a longer T 2 compared with bound water, 28,29 making it sensitive to microstructural changes related to EIMD.
The link between MRI measurements and postexercise cellular processes like mechanical stress, osmotic shifts, inflammation, permeability increments, edema, and secondary injury is not definitive. At the macrostructural level, however, muscle injury leads to reduction in the amount of bound water 30 and to a higher water content within the injured area, 31 thereby elevating the local T 2 relaxation time. 32 Muscle recovery aims to restore homeostasis and return to baseline fluid profile and T 2 relaxation values. 33 T 2 can therefore serve as an indirect biomarker of muscle damage and recovery.
Several studies used MRI's T 2 relaxation time as a marker for muscle damage. Aboodarda et al. 34 compared the intensity of T 2 -weighted spin-echo images of the thigh before and after eccentric knee extensions, to estimate muscle damage. In studies that use quantitative T 2 values, this relaxation time is mostly estimated by fitting multi-echo spin-echo (MESE) data to the exponentially decaying signal model. For example, this method was used by Takahashi et al. 32 on subjects performing quadriceps centric contractions, by Maeo et al. 35 after downhill running (DHR), and on rats after DHR. 30,36,37 Quantitative T 2 was also used to probe muscle state in professional athletes following the completion of triathlons, using exponential curve fitting of gradient and spin-echo protocol data. 38 This protocol offers relatively short acquisition time at a tradeoff of compromised image quality and weaker encoding of the T 2 relaxation time constant. 39 Human studies reported relatively minor changes in T 2 , ranging approximately 5 ms and below in the lower limbs. 20,32,35,38 DHR studies show even smaller changes of <2 ms, which were not significant in most tested muscles. 35,38 It is therefore essential to use high-precision techniques for estimating changes in T 2 following exercise. It was previously shown that the echo modulation curve (EMC) algorithm offers high precision, [40][41][42] which can be used to unravel subtle pathological changes that are visually undiscernible 43,44 and allow the investigation of second-order mechanisms that affect the signal decay curves in T 2 -weighted protocols (e.g., magnetization transfer and molecular diffusion). 45,46 In this study we harness the EMC quantitative T 2 mapping technique 47 to investigate muscle recovery following intensive DHR exercise on a cohort of 24 amateur male athletes. To estimate muscle state, T 2 values were measured at different regions in the thigh and compared across three time points: at baseline (BL; i.e., before the exercise protocol, 1 h after, and 48 h after DHR). Further investigation was performed regarding the correlation between the distribution of T 2 values and the participants' training routines and age.

Study population
Twenty-four healthy active men volunteered to participate in the study. Participants' age and weight were 34.8 ± 9 years old and 76 ± 11 kg. All participants trained at least 4 times a week and regularly competed in long-distance running and triathlon events. Before joining the study, all participants signed an informed consent, which was approved by Shaare Zedek Medical Center (Internal Review Board #0345-19) and Tel Aviv University Ethical committee (#0000400-4) and is registered at https://clinicaltrials.gov/ (NCT04025723).

Experimental design
Participants visited our research center on three separate occasions ( Figure 1). During their initial visit (D 0 ), blood samples were collected. Participants also completed a health and physical activity questionnaire, a BL MRI scan, and a graded exercise test to determine maximum heart rate (HR max ) and volitional exhaustion (i.e., maximal O 2 consumption [VO 2 max]). Subsequent visits (D 1 , D 2 ) were scheduled 2 days apart. During D 1 , a 60-min DHR protocol was performed followed by the second MRI scan,

F I G U R E 1
Schematic illustration of the experimental protocol. Participants arrived at three different days beginning with a screening process and an MRI scan on day zero (D 0 ), followed by a downhill running (DHR) exercise and additional MRI scans at subsequent visits. See Section 2 for a detailed description of the entire protocol. 1 h/48 h, 1 h/48 h after DHR protocol; BL, baseline. preformed 1 h after DHR. At D 2 , 48 h following DHR, participants underwent the third and last MRI scan.
Participants were instructed to refrain from consuming alcohol and caffeine for 12 h before visits, and fast for 2 hours before each visit. In addition, participants were asked to avoid unfamiliar or high-intensity exercises for 24 h before D 0 , D 1 visits, and between D 1 and D 2 visits in accordance with standard exercise protocols. [48][49][50]

Biochemical analysis of blood samples
Blood samples were collected at the beginning of D 0 visit in 10-mL Vacutainer tubes, and serum samples were kept at room temperature for 1 h before being centrifuged at 1300 g for 10 min. Samples were aliquoted into 1.8-mL microcentrifuge tubes and frozen at −80 • C until analyzed. Creatine kinase and lactic dehydrogenase were subsequently estimated using a Roche clinical chemistry and immunochemistry analyzer (Cobas c111; Roche Diagnostics).

Graded exercise test
On D 0 , after MRI scan and blood test, participants underwent a graded exercise test on a motorized treadmill (Saturn 100/300, h/p/cosmos; Nussdorf-Traunstein, Germany) with an individualized protocol. During the exercise, the treadmill speed was increased every minute for the first 5-6 min until reaching a comfortable speed, at which time participants could easily run for an hour. Next, the treadmill grade was increased by 2% every minute until participants reached volitional exhaustion, defined as the point at which participants were unable to continue exercising despite verbal encouragement. During the graded protocol, breath-by-breath analysis (Quark Cardiopulmonary Exercise Testing, Cosmed) was used to collect ventilatory and metabolic measurements while the participants breathed through an oro-nasal facemask (7450 Series; Hans Rudolph). Participants' heart rate was continuously monitored using a chest strap (Garmin, model Acc; HRM-Dual). VO 2 max was determined as the highest 30-s average O 2 uptake achieved during exercise, whereas HR max was determined as the highest recorded HR during the test.

DHR protocol
DHR was performed on a treadmill and consisted of a 5-min warm-up at 0% grade followed by a speed increase every 30 s until 65% of HR max was achieved. Following the warm-up, the slope of the treadmill was reduced to −10% (minus 10%) for 55 min with the running speed adjusted throughout the test so that each participant's HR remained at 65% HR max . All participants completed the exercise with the target HR successfully maintained during the exercise. The average running speed was 8.9 ± 2.2 km/h.

MRI scans
MRI scans were performed on a Magnetom Prisma 3T Siemens scanner at the imaging center of Tel Aviv University. A 6-channel receiver coil was placed on the distal quarter of the right thigh, in between the Trochanter major and the Patella, and 13 axial slices were scanned. Quantitative T 2 -mapping data were collected using a 2D MESE pulse sequence for a series of increasing TE values. A long echo train (echo train length = 14) was selected, as multislice MESE protocols decay slower than the theoretical exponential decay because of stimulated and indirect echoes, which cause an elongation of T 2 relaxation curve. Experimental parameters included TR/TE = 3200/10 ms, echo spacing = 10 ms, echo train length = 14, acquisition bandwidth = 200 Hz/pixel, matrix size = 160 × 160, FOV = 220 × 220 mm 2 , slice thickness = 4 mm, slice gap = 12 mm, and 2-times GRAPPA acceleration. Figure 2 illustrates the MRI data processing pipeline. T 2 values were estimated using the EMC algorithm. 45 In this technique, a dictionary of theoretical T 2 decay curves is computed by repeating Bloch simulations of the prospective MESE protocol for a range of T 2 relaxation values and transmit-field (B + 1 ) inhomogeneity levels. The resulting dictionary contains a series of signal decay curves, each associated with a unique [B + 1 , T 2 ] value pair. After data acquisition, the experimental signal from each voxel is matched to the dictionary of simulated curves, yielding a unique T 2 value through L 2 norm minimization between experimental and theoretical curves. All fitting procedures were programmed in house using C++ and MATLAB (The MathWorks).

MRI data processing
After all maps were calculated, image registration was performed between the three time points using ANTS. 51 This process used affine transformation of the T 2 -weighted images from the second TE from each scan session, to track and repair the potential shifting of slices between time points along the longitudinal (z) axis of the leg.

F I G U R E 2
Flowchart of the MRI data-processing pipeline. T 2 maps were generated per voxel, followed by automatic removal of fat regions. Registration was then performed between scans from different time points (D 0 , D 1 , D 2 ) followed by segmentation of specific muscles on each of the registered images. Finally, the T 2 mean value and SD were calculated for each segmented region of interest.
Manual segmentation of selected muscles was performed for each subject and at every time point on three representative slices located at the middle part of the thigh and selected by a physiologist out of the 13 slices that were scanned. Segmentation was performed on the T 2 -weighted MESE images from the second TE. Two pairs of muscles were segmented: vastus lateralis (VL) and rector femoris (RF) constituting the anterior compartment of the thigh primary mover muscles, and biceps femoris (BF) and semitendinosus (ST) constituting the posterior compartment of the thigh muscles. This resulted in an overall number of 288 regions of interest (ROIs) (24 subjects × 3 slices × 4 ROIs). An experienced physiologist validated the segmentation. Finally, the mean and SD of T 2 values were calculated for each ROI, along with the corresponding number of voxels. Voxels with T 2 > 50 ms were tagged as not-muscle voxels and were not included in the mean and SD calculation. 52 This operation removed a few thin regions of intramuscular adipose tissue, particularly at the fascia lata.

Statistical analysis
Statistical analysis was performed using MATLAB R2020b statistical toolbox. A preprocessing step was performed to test the distribution pattern of T 2 values using the Shapiro-Wilk test with a confidence level of 0.05. This step verified that the mean and SD of T 2 values were normally distributed, allowing us to use a two-tailed paired sample t-test to investigate whether they changed across time points. Next, we tested the statistical correlation between quantitative MRI values vis-à-vis training experience, age, and weekly training routine (distance and duration). This was done using Pearson's linear correlation with p < 0.05 indicating statistical significance. For multiple comparisons correction, the Benjamini-Hochberg method was applied. 53,54

RESULTS
A summary of the study population characteristics, including weekly training routine and training experience, is presented in Table 1. While a wide variation existed in the weekly training duration (8 ± 4 h) and weekly accumulated running distance (47 ± 27 km), the variability of VO 2 max, which constitutes the gold standard of fitness level, was relatively low (48.5 ± 6.3 mL/kg/min). Biochemical analysis of EIMD blood markers at baseline (i.e., before DHR protocol) produced 142 ± 32 U/L for lactic dehydrogenase and 224 ± 122 U/L for creatine kinase, corresponding to values at rest. 55,56 Data for five MRI scans (3 subjects) were excluded from the analysis due to extensive motion artifacts. Remaining data points were included in the statistical analysis. An example of segmented muscles is shown in Figure 3 overlaid on a T 2 -weighted axial image of the thigh (TE = 20 ms). Shaded areas denote the four selected ROIs: VL, RF, BF, and ST. Representative T 2 maps for 1 participant are shown in Figure 4 for the three time points: at baseline (BL), 1 h after, and 48 h after the DHR protocol. Table 2 lists the mean T 2 values for each of the assayed muscles (VL, RF, BF, and ST). Statistically significant changes were found in the mean T 2 values for the primary movers (VL and RF muscles) between baseline and 1 h after DHR protocol. This change remained significant also 48 h after the exercise. Differences became significant for the other two muscles, ST and BF, at the third time point-presumably because of a secondary effect associated with a systemic response to the DHR protocol.
Analysis of intra-ROI SD of T 2 values is presented in Table 3. SD values decreased immediately following Characteristics of the study population.

Mean SD Min Max
Age (

F I G U R E 3
Muscle regions investigated in this study overlaid on a T 2 -weighted axial image (TE = 20 ms) of the right thigh. Shaded areas correspond to vastus lateralis muscle (VL, red), rector femoris muscle (RF, blue), biceps femoris (BF, aqua), and semitendinosus (ST, green). exercise for all tested muscles (BL vs. 1 h), reflecting an overall decrease in the tissue's heterogeneity. Muscle heterogeneity was also statistically different between BL and 48 h following DHR. However, there was no significant change between 1 h and 48 h after DHR, indicating that changes in tissue heterogeneity occur at short time scales and do not fully recover after 48 h. To further support these results, these analyses (of both mean and SD of T 2 values) were repeated for two other slices, producing similar results (see Supporting Information).
Baseline T 2 values from the combined ROI (containing all four muscles) were analyzed for correlation with participants' age, the number of training hours per week (3.5-20 h), weekly running distance (10-125 km), and years of experience (0.25-35 years). Statistically significant correlation was found between mean T 2 values at BL and weekly running duration (R = −0.61, p = 0.003), and between SD of T 2 values at BL and weekly running duration (R = −0.59, p = 0.004). These results remained significant after multiple comparisons correction (eight comparisons); see Supporting Information for more details. Differences in T 2 values before and after DHR were also correlated with training habits, producing no significant correlation.

DISCUSSION
In this study, we identified significant elevations in T 2 relaxation times alongside a reduction in the tissue heterogeneity (SD of T 2 ) after aerobic EIMD protocol. This reflects microstructural muscle damage and may identify the individual rate of recovery following EIMD. Changes in T 2 relaxation times can result from several sources including mechanical stress, inflammatory reactions, changes in cells' permeability, edema, and secondary injury. These mechanisms can be further separated into localized and systemic processes operating over different timeframes. Although mechanical stress occurs during the exercise, inflammatory reactions begin immediately after and may last for several days following stimulation. 57 Secondary injury emerges later, although the timeframe is not yet clear. 23 Based on measurements of T 2 values at three time points and across different muscle groups (agonists and antagonists), it was possible to distinguish between these short-term and long-term physiological processes.
Quantitative T 2 maps were estimated using the EMC algorithm. 45,47 By incorporating the exact scan settings and pulse-sequence timing diagram into the postprocessing procedure, this technique offers more accurate and precise mapping of T 2 values with excellent agreement to ground truth, 47 robustness to variations in T 1 relaxation times, and robustness to transmit (B + 1 ) and receive (B − 1 ) fields inhomogeneities. 45 This endows the EMC technique with the ability to identify subtle tissue changes, which are not detectable using conventional visual assessment of MR images 42,44 and allows us to compare different time points with high precision. Immunity to main field (B 0 ) inhomogeneity was also important in our study given that the same anatomy was positioned differently during each scan relative to the magnet and the receive coil. 45,58 Potential B 0 bias, however, was avoided by using a spin echo-based T 2 mapping protocol. Quantitative T 2 maps derived using standard exponential fitting might also reflect similar changes after exercise. Such values, however, will be highly overestimated 59 ; suffer from higher interscanner, intrascanner, and intersubject variability 42 ; and significantly lower reproducibility across scanners and scan parameters. Notably, our results were only significant on a group-wise level, which averaged out the natural interscan variability, and enables us to identify the relatively small (<2 ms) changes after DHR protocol.
A consensus exists regarding elevation of T 2 values in muscles after eccentric exercise. 21 This change is independent of the type of exercise and can last anywhere between hours and days. Maeo et al., 35 for instance, showed significant elevations of mean T 2 24 h following DHR of 45 min at −15 • slope in the vastus intermedius (+3%), lateral gastrocnemius (+6%), and soleus (+3%) muscles. Similar DHR protocols were performed on rats at −16 • slope for 90 min 37 and −15 • slope until exhaustion (3-4 h), 36 both reporting elevation in T 2 values for 72 h after running. Another study in humans that examined 10 repetitions of eccentric knee extension for maximum effort showed an increase of +4% in T 2 of the quadriceps muscles. 34 In contrast, a Bruce uphill treadmill protocol (a concentric exercise) led to elevation in T 2 for only about 30 min in the gastrocnemius (+14%), anterior tibialis (+10%) and soleus (+8%) muscles, after which values returned to BL. 20 In congruence to previous findings, the current study found a significant elevation of T 2 values between BL and 1 h after DHR in the VL and RF muscles, and additional elevation between 1 and 48 h following DHR in the VL, ST, and BF muscles. We attribute the variability between different muscle groups to the different roles of the posterior and anterior muscles during DHR. In this type of exercise, the anterior muscles are the agonists (primary movers), performing eccentric contractions, whereas the posterior muscles are the antagonists undergoing concentric contraction. The immediate elevation of T 2 after exercise in the primary movers can be associated with a direct microstructural damage induced by the exercise mechanical stress. Delayed microstructural damage (i.e., secondary injury) is a slower systemic process affecting all the engaged muscles (agonists, antagonists, and synergists), which may be the source of T 2 elevation between 1 h and 48 h after the exercise time points.
A previous study that measured T 2 values in the quadriceps 48 h after DHR 35 reported an increase of 0.6 ms in the vastus medialis, 1.1 ms in the vastus intermedius, 0.3 ms in the RF, and 0.8 ms in the VL, whereas the change in the vastus intermedius was the only significant one. Another study, which measured T 2 values 3 h after a

T A B L E 2
Mean ± SD of T 2 values measured at three time points and averaged across all subjects.

Mean T 2 Values (ms)
Change

Note:
Values are shown for four regions of interest: vastus lateralis (VL); rector femoris (RF); biceps femoris (BF); semitendinosus (ST). Also shown are the absolute differences between each of the two time points. Numbers in bold indicate statistically significant differences (p < 0.05) after Benjamini-Hochberg correction for multiple comparisons.

T A B L E 3
Intra-region-of-interest SD of T 2 values measured at three time points and averaged across all subjects.

SD of T 2 values (ms)
Change triathlon, 38 reported an increase of 1.6 ms in the RF, 1.4 ms in the BF, and 0.8 ms in the ST, although no change was statistically significant. In the current study, the mean change in T 2 between BL and 48 h was 1.6 ms in the RF, 0.9 ms in the VL, 0.4 ms in the ST, and 0.7 ms in the BF. These results are consistent with the effect size measured in previous studies, although the results in the current study are statistically significant, presumably due to the high precision of the EMC technique 42,45 compared with T 2 mapping methods used in previous studies. Moreover, we are unaware of any studies that compared DHR-related muscle damage in agonist versus antagonist muscles. A related comparison was done after 3 × 100 repetitions of rebound jumping at the maximum speed, where T 2 elevation was observed only in the agonist muscles. 60 The current study, on the other hand, suggests that all muscles are affected, while the antagonist muscles are affected at longer time scales. The SD of T 2 values was found to be a key marker of muscle state, reflecting the heterogeneity of the muscle tissue. In previous studies, elevation of T 2 SD values was found to correlate with neurodegenerative muscular diseases, manifesting abnormal increase of intramuscular fat and tissue fibrosis, 61-63 whereas a reduction in SD was found in lumbar intervertebral disc degeneration. 64,65 To the best of our knowledge, the current study is the first to analyze heterogeneity changes within muscle tissue following aerobic-based EIMD. Even though extensive exercise may lead to positive muscle development, it may also cause short-term microstructural damage and changes in the distribution of water molecules between subvoxel compartments, leading to changes in measured T 2 values. The current study results indicate that changes in T 2 SD were most notable between BL and 1-h time points, with no apparent change at the 48-h time point (Table 3). This reduction was significant in all tested muscles, suggesting the existence of a short-term systemic process, perhaps due to an increase in cellular permeability, which promotes the passage of large molecules from cells into the extracellular fluid. This reduces the amount of water bounds to the cells' surface and causes free water from the interstitial space to enter the cells, resulting in cell swelling. 30 A natural variability in recovery rate exists between subjects and should be considered when designing personalized training programs. This variability is reflected in the muscles' T 2 values and depends on various parameters such as individual physiology, age, training experience, and training routine. We found no correlation between each individual factor and changes in T 2 after exercise, presumably because they need to be analyzed in tandem to accurately predict muscle performance and recovery rate. Further studies with larger sample size should also consider other cofactors such as sleep habits, dietary habits, and occupation.

General considerations and study limitations
The present study has several aspects that should be considered. First, T 2 changes are not specific to one physiological process, and different processes may lead to similar changes in T 2 relaxation times. For example, damage to cell membrane will increase the amount of extracellular fluid and produce longer relaxation times, whereas an increase in membrane permeability due to osmotic pressure will also shift water to the extracellular space, leading to a similar elevation in T 2 .
Muscle tissue is known to have multi-compartment microarchitecture, whereas specific changes in T 2 may correspond to specific subvoxel compartment (e.g., bound vs. free water or intercellular vs. intracellular changes). The use of single-component analysis may therefore be improved by using multicomponent analysis, which will deliver higher sensitivity to changes in the tissue's micro-compartmentation. 66 The current study also did not consider the effect of magnetization transfer, which can bias MESE signals. According to previous investigations, magnetization-transfer interactions accumulate along the MESE echo train, resulting in small underestimation of the calculated T 2 values. 67 These are expected to have a relatively small effect on the fast-relaxing muscle signals, although further investigation is needed to understand this effect and to incorporate it into MRI studies of muscles.
Another important aspect is the absence of a readily available and reliable tool for automatic segmentation of the individual thigh muscles, 68 which led us to perform manual segmentation. Recently, automatic neural networks have been published, 69-71 demonstrating satisfactory performance. However, none of these tools have been made publicly available, whereas in-house implementation and training require significant amounts of time and data that are beyond the scope of this study. This limited our analysis to only four muscles (two agonists and two antagonists) and three representative slices. Although a global analysis would better reflect the overall muscle state, 72 we believe that it would not produce fundamentally different results. To validate this, we reproduced all results for two additional slices (above and below the one given in Tables 2 and 3) and for a combination of all three slices, consistently producing the same findings (see Supporting Information).
The finite timeframe of the experimental design was another limiting factor. Here, we followed the recovery phase at two time points, 1 h and 48 h after DHR, due to practical understanding that study participants belong to a population that does not rest for more than 48 h between training sessions. For this reason, participants were instructed not to exercise 24 h before a BL scan, even though 24 h may not be enough time for the muscle to stabilize after exercising. Routine workouts of the study participants were generally based on noneccentric exercise, which is more familiar, less intense, and is thus less likely to induce muscle damage. 73,74 One of the main conclusions of this observation is that future study designs should include longer rest periods before exercise.
Finally, all subjects in the current study were males. Reported findings may therefore not apply to female populations. This choice to focus on only one gender was done due to the small number of participants and the need for as homogeneous as possible group of participants. Males were chosen in this case, as they experience fewer hormonal variations, which can affect the recovery following exercise as previously reported in Oosthuyse and Bosch 75 and Enns and Tiidus. 76

CONCLUSIONS
This study shows that noninvasive measurements of MRI's T 2 relaxation time can provide quantitative information regarding muscle state and recovery following EIMD. Study findings indicate that both the mean and SD of T 2 values are useful in this context. These findings will help subsequent studies that are geared toward understanding exercise damage and recovery processes, as well as provide valuable information for planning personalized training plans, rehabilitation activities, and for understanding the pathophysiology of diseases that affect skeletal muscles.