Use and applicability of magnetic resonance elastography of the lumbar spine in adults: a scoping review

Abstract

Background

Magnetic Resonance Elastography (MRE) is a non-invasive imaging technique that quantifies tissue stiffness by analyzing shear wave propagation. While MRE is widely used in hepatic imaging, its application in the lumbar spine remains an emerging field. Understanding the repeatability and reproducibility of MRE measurements in the lumbar spine is crucial for its clinical implementation. This scoping review aims to summarize current evidence on the use and applicability of MRE for assessing lumbar spine structures, including intervertebral discs and paraspinal muscles.

Methods

A systematic literature search was conducted in MEDLINE (PubMed), CINAHL, Embase, and The Cochrane Library. Studies investigating MRE of the lumbar spine in adult populations were included. Key aspects such as MRE acquisition methods, repeatability and reproducibility of measurements, and study heterogeneity were assessed. Extracted data were categorized based on study design, imaging techniques, and primary outcomes related to lumbar stiffness assessment.

Results

This review identified 11 relevant studies. These studies demonstrated the capability of MRE to characterize shear stiffness in the lumbar intervertebral discs and paravertebral muscles, in both resting states, across various muscle conditions, and under different interventions such as physical activity and therapeutic taping. The review documents the heterogeneous methodological approaches of the studies, highlighting the innovative but varied approaches to this field. Due to this, diverse findings were reported, some of which were contradictory.

Conclusion

The current evidence of MRE of the lumbar spine is promising though limited due to heterogeneous study methodologies. Future research should focus on larger, multicenter studies with standardized protocols. Despite the current limitations in evidence, MRE holds potential for non-invasive lumbar spine assessment and further research validation.

Introduction

Worldwide, low back pain (LBP) is a common condition affecting people of all ages [1,2,3,4] and is one the leading causes of disability (years lived with disability) [1]. It is estimated that more than 800 million people globally will have LBP by 2050 [3]. Diagnostic imaging such as Magnetic Resonance Imaging (MRI) or radiography may be indicated in case of no improvement following relevant non-surgical treatment [5, 6] or if malignancy or facture is suspected [6, 7].

MRI of the lumbar spine is usually applied for acquiring anatomic images with a high soft tissue contrast, such as the intervertebral discs or nerves [5,6,7,8,9]. However, MRI also allows for special acquisition techniques such as Magnetic Resonance Elastography (MRE). MRE quantifies the mechanical properties of tissue and works on the principle that propagating mechanical waves move more rapidly in stiff rather than soft tissue. When subjecting the body to shear waves through external vibration, their propagation can be encoded using motion-encoding gradients. The acquired wave data is processed through an inversion algorithm, which primarily measures the shear modulus reflecting the tissue’s resistance to deformation. Thus, the shear wave speed or storage modulus can be calculated using a specific algorithm. Ultimately, this process leads to the generation of quantitative maps of tissue elasticity and viscosity (Fig. 1). When tissues are pathological, biomechanical changes may occur which may alter their rigidity. Thus, by examining and characterizing the tissues stress-strain dynamics, valuable information may be obtained on the mechanical differences between normal and abnormal tissue [10]. A recent narrative review [11] emphasized the importance of MRE to understand the microstructural integrity and degeneration of the intervertebral disc (IVD). However, the complexity of the lumbar region also encompasses an intricate network of muscles, ligaments, and other soft tissues. This understanding of the lumbar spine complexity is the rationale behind this scoping review.

Fig. 1

Magnetic Resonance Elastography of the lumbar spine. Axial T2-weighted MRI (A & C) with corresponding MR elastograms (B & D) of the lumbar spine. DS: Dural sac, ES: Erector spinae, FO: Fish oil capsule (actuators placement), IVD: Intervertebral disc, MM: Multididus, PM: Psoas major, QL: Quadratus lumborum, VEP: Vertebral endplate. SWS: Shear Wave Speed in m/s

Scoping reviews are particularly relevant to map key concepts, identify research gaps, and elucidate areas of research heterogeneity. Current studies on MRE of the lumbar spine [12,13,14,15,16,17,18] have applied different experimental designs, and their documented findings vary, with some cases being contradictory. For instance, two studies explored the relationship between MRE-derived shear stiffness and Pfirrmann score. One study found that stiffness increased with a higher Pfirrmann score, while the other study found the opposite relationship. There is thus a need for a broader perspective including the potential clinical implications of using MRE to differentiate between normal and abnormal lumbar tissue.

The purpose of this scoping review was to examine and outline current evidence on the use and applicability of MRE of the lumbar spine with a special focus on identifying normal and abnormal lumbar tissue in adults.

Methods

This scoping review follows the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) checklist³¹. A protocol was developed to guide this scoping review, detailing the intended approach, methods, and procedures. The protocol has not been published but is available from the corresponding author upon reasonable request.

Search strategy

A three-step search strategy was applied to search for relevant literature in relevant databases [19]. The first step used the index terms and words from titles and abstracts in the first selected articles to form further search terms. The introductory search terms were used to search the PubMed database in collaboration with a research librarian to identify more precise and extensive search terms to identify all relevant studies. The second step used the identified keywords and index terms adjusted to each of the selected databases. The third step covered an assessment and screening of the reference lists of each selected full text article for additional relevant literature.

Inclusion criteria

Inclusion criteria for this scoping review are summarized in Table 1.

Table 1 Inclusion criteria with rationales

Information databases

The following databases were searched up to 18th December 2023: MEDLINE (PubMed), CINAHL Complete (EBSCO), The Cochrane Library and Embase (Elsevier). Relevant thesauruses were applied for each database.

Furthermore, a block search was conducted for each of the two domains, i.e., “Magnetic Resonance Elastography” and “lumbar spine”. An exemplary search is shown in Table 2. No direct contact with authors was made to identify additional sources.

Table 2 Block search in medline (Pubmed), with chosen search terms and the combined block search

Screening and selection

Citations were organized using Mendeley Reference Manager (version 2.97.0) and duplicate entries were removed. Two independent reviewers screened titles and abstracts in accordance with the inclusion criteria. Full-text articles complying with the inclusion criteria, including those found through reference lists, were retrieved, and carefully evaluated. Any disagreements were resolved through discussion to reach consensus, and a third reviewer was involved if necessary to achieve consensus.

Data extraction

Appropriate descriptive data from the selected articles were extracted and schematized in a data extraction instrument. The data extraction instrument was used in an iterative process to ensure that additional relevant data was included and continually updated throughout the data extraction process. Data extraction was performed by two independent reviewers in parallel and subsequently compared for consistency. Any discrepancies were resolved through discussion and consensus was obtained. In addition, if any review was detected, its reference list was scrutinized to find potentially relevant studies that might have been missed in our original search. There was no explicit process for contacting original authors for data confirmation.

Data were extracted based on the following categories:

• Article characteristics: Publication year, first author and country of origin.

• Study details: Purpose, anatomical region, and participant demographics, e.g., gender and age.

• Target tissue: Muscles, intervertebral discs, etc.

• Technical details: Hardware, technical setup, and methods of data acquisition and patient positioning.

• Outcomes: MRE outcomes, e.g., shear stiffness and main findings.

Presentation of results

Study and MRE characteristics related to the usage of MRE of the lumbar spine are summarised in text and in tables. The main results of the included articles are related to the applicability of MRE in different lumbar spine tissues and presented in a narrative form including tables and figures. The terminology used to describe stiffness in the included studies varied, using terms like ‘shear stiffness,’ ‘shear modulus,’ and ‘complex modulus’. To ensure clarity and maintain simplicity, ‘shear stiffness’ has been consistently adopted.

Results

To provide an overview of the findings, we have included an infographic summarizing the key results of this study. See Fig. 2.

Fig. 2

Overview of main results from the 11 included studies

Inclusion of studies

The database searches conducted in April 2023 identified 112 records of which 9 full-text articles investigating the use and applicability of MRE of the lumbar spine [12, 14,15,16,17,18, 20,21,22] were included. A new search was conducted in December 2023 and one new relevant article was retrieved [23]. An additional relevant study was identified through expert recommendation, which had not appeared in the initial search results. After evaluating its relevance, the study was included in the review. Figure 3 illustrates the inclusion process.

Fig. 3

PRISMA flowchart of study selection and inclusion process

Study characteristics

Study characteristics are shown in Table 3. The 11 studies included were published between 2015 and 2023, with sample sizes ranging from three to 80. Ten included only asymptomatic participants [12, 15,16,17,18, 20,21,22, 24], while one did not specify this [14]. Five studies included only men [15, 17, 21, 22], and six included both genders [12, 14, 16, 18, 23, 24]. Participant ages ranged from 19 to 71 years. Weight-data was inconsistently reported; two studies [14, 21] provided mean weight, and one provided BMI [23], while the rest omitted weight details.

Table 3 Study characteristics and main results

Three studies focused on the IVD [16, 18, 24], five on the psoas major muscle (PM) [15, 17, 20,21,22], and four on the paraspinal muscles (PSM) [12, 14, 20, 23]. Eight studies conducted MRE with participants in supine position [14,15,16,17, 21,22,23], while two used prone position [12, 20], while one study did not specify positioning [18]. Ten studies assessed tissues at rest [12, 15,16,17,18, 20,21,22,23,24], while one study also included stretching and contraction [14]. Additionally, one study evaluated the effects of physical activity on biomechanical properties [21].

MRE measures

All studies measured and reported shear stiffness [12, 14,15,16,17,18, 20,21,22,23,24], and one study also included phase angle [18]. The three studies focusing on the IVD compared shear stiffness to different stages of disc degeneration using the Pfirrmann score [16, 18, 24]. Additional measurements in the reviewed studies include noise effect measurements such as signal-to-noise ratio (SNR) and octahedral shear strain-based measure of SNR (OSS-SNR) [18, 24]. One study measured T2 values and the apparent diffusion coefficient (ADC) [21]. Five studies performed a repeatability analysis [12, 16, 22,23,24], while another five conducted a reproducibility analysis [18, 20, 22,23,24]. Another study examined the Mean Oscillation Amplitude (MAV) [17]. A detailed overview of MRE measures is presented in Table 3.

MRE techniques

Seven of the 11 studies conducted their experiments on 3T MRI machines [12, 15,16,17, 21, 22, 24], while four studies used 1.5T MRI systems [14, 18, 20, 23]. Five of the studies used a custom built MRE device of which four were a loudspeaker-based actuator with pneumatic transmissions (LS-P) [15, 17, 21, 22], and one was a loudspeaker-based actuator with rigid transmission (LS-R) [18]. The remaining six studies used a commercially available LS-P system from Resoundant [12, 14, 16, 20, 23, 24]. Regarding the passive actuators, six studies used the commercially available vibration pad developed by Resoundant [12, 14, 16, 20, 23, 24], one used a wooden plate [18], and four studies had 3D printed their own actuators [15, 17, 21, 22]. Vibration frequencies differed between 50 and 120 Hz; ten studies made use of a single vibration frequency [12, 14,15,16,17, 20,21,22,23,24] and one made use of multiple frequencies, also known as multifrequency MRE (MMRE) [18]. Different variations of spin-echo (SE) [18, 24] and gradient echo (GE) [12, 14,15,16,17, 20,21,22,23,24] sequences were employed for MRE image acquisition. Regarding the Region-of-Interest (ROI), six studies manually drew the ROI [14, 18, 20, 22,23,24], one study used a custom algorithm [16] and four did not specify this [12, 15, 17, 21].

Regarding post-processing algorithms, one study used the Multifrequency Dual Elasto-Visco (MDEV) inversion algorithm [18], five used Local Frequency Estimate (LFE) algorithm [15, 17, 20,21,22], and two used a Principal Frequency analysis (PFA) algorithm [16, 24]. The remaining studies did not specify this [12, 14, 23]. A detailed overview can be found in Table 4.

Table 4 Experimental setup and data processing

Applicability of MRE for lumbar spine tissue

Intervertebral discs

The study by Streitberger et al. [18] aimed to develop and evaluate MRE for the human spine. Initial assessments were conducted on ex vivo bovine discs, revealing pronounced viscous-fluid behavior within the nucleus pulposus (NP). Clamping the ex vivo bovine discs resulted in a significant reduction in the mechanical parameters of the NP, indicating marked softening of the NP tissue under compression. This alteration was highly significant. Sixteen asymptomatic subjects, with a mean age of 31 years (25–50 years) were investigated in vivo. OSS-SNR measurements showed that shear wave images used for inversion were reliable. Moreover, a significant correlation was observed between decreased shear stiffness and increased disc degeneration measured by the Pfirrmann score [25]. The reproducibility of MRE measurements was also established, with a variability of ~ 11% for shear stiffness and phase angle. Lastly, the researchers observed that the wavelengths of the shear waves decreased with increasing excitation frequency.

Walter et al. [16] undertook a study to investigate the repeatability of MRE shear stiffness of the IVD and its association with age and degeneration. The study included 47 asymptomatic subjects between 20 and 71 years. A consistent MRE measurement was observed with no significant variations between morning and afternoon evaluations. Differences in shear stiffness were identified across various IVD levels. The study also found a significant increase in MRE-derived shear stiffness as the Pfirrmann score increased in both the annulus fibrosus (AF) and NP regions. However, the connection between age and MRE-derived stiffness exhibited only weak correlations.

Co et al. [24] contributed to the methodological development of IVD MRE by validating a SE-EPI sequence against the traditional GE sequence. The study included 28 healthy subjects (age range 19–55 years) and demonstrated high reproducibility and repeatability of SE-EPI-derived shear stiffness in both NP and AF regions, and a good correlation with the GE sequence. Shear stiffness values measured with GE were lower than those obtained with SE-EPI, with differences of approximately 3.2 kPa in the NP and 21.4 kPa in the AF region. The study also investigated stiffness measurements in 2D and 3D scans, revealing that 2D-based scans yielded significantly higher stiffness values. Notably, SE-EPI reduced scan times by at least 51% compared to GE. OSS-SNR values were similar between SE-EPI and GE sequences. Lastly, the study found a low observer-bias for ROI delineation, corresponding to less than 4% of the mean measurement.

Posterior paraspinal muscles

Creze et al. [14] included seven volunteers with a mean age of 25.6 ± 6 years and explored the feasibility of applying MRE to the PSM and examined the elasticity characteristics of the two primary PSMs, multifidus and erector spinae. The analyses found no significant difference in shear stiffness between the left and right muscles. However, variations in stiffness patterns and shear stiffness were apparent, especially when comparing relaxed muscles to stretched and contracted states. Significant differences in shear stiffness were also observed between the multifidus and erector spinae muscles at rest and during stretching, but not during contraction.

Wang et al. [12] investigated the influence of Kinesio Tape (KT) on the PSM using MRE. They included 66 asymptomatic volunteers with a mean age of 28.4 ± 8.4 years. They found that the superficial muscle depth exhibited a significant decrease in shear stiffness with KT compared to the non-taped side; this pattern was consistent for both 5 cm and 2.5 cm ROIs. No significant changes were found for the deeper muscle layers. The study found no significant difference in shear stiffness between genders and no significant correlation between taping direction and changes in stiffness.

Chevalier et al. [23] investigated the reliability of MRE for assessment of the multifidus and the erector spinae muscles. The study included 17 asymptomatic participants with a mean age of 28 ± 4 years. The study revealed that MRE provided shear stiffness measurements of the lumbar muscles with varying degrees of reproducibility, ranging from fair to excellent, and demonstrated excellent interobserver agreement. However, the study highlighted potential limitations within the MRE protocol, particularly in assessment of the erector spinae muscle where the inter-day reproducibility was poor. Challenges included technical issues such as the need for repositioning of the actuator during setup and data quality affected by asymmetrical wave amplitudes and artifacts. The study observed a trend of decreasing stiffness from the surface of the multifidus muscles, and a significant difference in stiffness when changing the positioning of the actuator. No significant differences between multifidus and erector spinae were found.

Psoas major muscle

The study by Numano et al. [15] introduced an MRE method utilizing a GE type multi-echo sequence, which eliminates the need for motion-encoding gradients (MEG) thus preventing prolongation of the echo time (TE). The study included phantoms and three asymptomatic volunteers. The study found that the readout lobes caused an MEG-like effect, and that later TE images had a higher sensitivity to vibrations. The direction of frequency encoding was observed to affect the wave image patterns, with distinct patterns seen for left-right and head-feet readouts. Additionally, the study explored the impact of magnetic susceptibility on the images, revealing that while artifacts were present on the magnitude image at later TEs, the wave image remained unaffected. In the PM, this MRE method was able to measure the mean shear stiffness. Notably, a heightened sensitivity to vibration in later TE was advantageous for perceiving waves in deeper tissues.

In a later study by Numano et al. [17], an actuator with three individually controlled chambers was used to investigate the PM. The study included seven asymptomatic volunteers aged 20–25 years. This study unveiled that the MAV exhibited higher values in the anterior-posterior than in the left-right direction. The three-chamber actuator further highlighted differences in propagating shear waves based on the actuator location on the body relative to the muscle that was investigated. The actuator location significantly influenced MAV results with each position (right, center, left) resulting in unique MAV patterns for the different muscles. No statistically significant differences in shear stiffness were found between ipsilateral and contralateral activation of the actuators.

Habe et al. [22] introduced a refined actuator for PM-MRE and evaluated its measurement precision and patient comfort. The study included nine asymptomatic volunteers aged 23.1 ± 2.1 years. The participants found this actuator sufficiently comfortable, leading to a low discomfort score. The study examined the repeatability of measurements and found that the percent repeatability coefficient (%RC) was 20.7% for the right PM and 18.9% for the left PM, which is around the ≤ 19% required by the Quantitative Imaging Biomarker Alliance (QIBA) for liver MRE. Moreover, with three different operators, the measurements showcased exceptional reproducibility, evident from the near-perfect Intra-Class Correlation (ICC) values. The study also provided shear stiffness of the PM for both sides.

Hsieh et al. [20] studied the reliability of MRE, particularly investigating whether MRE might encounter higher technical failure rates and reduced image quality in PM compared to PSM. The study included phantoms and 80 asymptomatic volunteers with a mean age of 38.6 ± 11.2 years. In the phantom study, a mean stiffness value of approximately 3.97 kPa was observed and a low coefficient of variation. The technical failure rate was low with no failures reported, and the image quality was rated at 15 (max score). However, when translating the phantom study to a living environment, disparities emerged between genders and muscle type. Females demonstrated a significantly greater tissue stiffness in the PM than males, and the technical failure rate was noticeably elevated for PM (60-66.3%) over PSM (0-2.5%). Moreover, the image quality scores revealed that PSM images had superior quality compared to PM. Low reproducibility was noted for both muscle groups. Of note, most scans had limited areas within the 95% confidence stiffness map.

Habe et al. [21] aimed to analyze the time-course alterations in the physical properties of the PM after exercise. The study included nine asymptomatic volunteers aged 23.4 ± 5.5 years. The shear stiffness of the PM was reduced up to about 30 min after exercise, only to be gradually elevated and increased significantly by about 100 min after exercise compared to pre-exercise. Contrarily, both T2 values and ADCs increased significantly up to about 65 min after exercise but then settled at initial values. The statistical analyses in the study were able to show significant shifts in the shear stiffness, T2 values, and ADCs over various post-exercise durations. While the exercised side showed variations of up to 20%, the non-exercised side remained more stable with a maximum change of 10%. By day two and seven, all metrics reverted to pre-exercise levels.

Discussion

The results from the studies in this review on the use and applicability of MRE of the lumbar spine showed that MRE is a promising non-invasive imaging technique for reliable assessment of shear stiffness of paraspinal muscles and intervertebral discs. However, the studies also highlighted some of the challenges to make this technique useful in larger settings and clinical practice.

Sources of error in MRE imaging

The process of estimating material parameters of tissue presents its own challenges, particularly due to the assumptions made by the inversion algorithms used to interpret the wave data. These algorithms are predicated on certain ideal conditions such as the presence of purely shear waves and tissue homogeneity. However, the real physiological environment may differ significantly, leading to several potential sources of estimation errors. The effect of boundary conditions, such as tissue heterogeneity may introduce significant estimation errors. An inversion algorithm that does not account for the presence of boundaries may incorrectly interpret the wave speed data, leading to erroneous conclusions about the inherent material properties of the tissue [26].

The importance of transparency and detail in MRE studies

To properly compare data cross studies, transparency, and detail in MRE experiments are important. Generally, the included studies were thorough in their reporting of various aspects such as type of MRI scanner, MRE equipment, vibration frequency, and MRI sequence type, but other relevant aspects were not clearly accounted for in all studies. The MRE guidelines committee published an article in 2021 [26] on recommendations for ‘good practice’ in MRE publications. It mentions several elements that are recommended to be reported, but some are not found or clearly defined in the studies included in this review. For example, it is recommended that the choice of inversion algorithm is clearly defined to allow others to reproduce the study. In this review, three studies did not clearly describe which inversion algorithm they used for processing the MRE images [12, 14, 23]. Other studies have examined the use of different inversion algorithms and found that these have an impact on the measured stiffness. One study evaluated the performance of various inversion algorithms by comparing them to a known ground truth using a phantom containing different inclusions [27]. Results showed that while the algorithms were consistent with the ground truth for softer inclusions, discrepancies were more pronounced for stiffer inclusions. Moreover, there were variations in the values obtained from in vivo tests on kidneys and brains across the different algorithms. Such inconsistencies emphasize the challenge in comparing data across studies that use different algorithms or not state which algorithm is used. Another example is the choice of ROI. The guidelines committee states that a given study should clearly set out the criteria for drawing the ROI. Detailing of ROI delineation varied between studies in this review. Some studies included a detailed description of the criteria for delineating their ROI [14, 16], which makes it easier to replicate. Other studies simply referred to a figure showing a picture of their ROI [15, 17], which leaves it up to the reader to interpret the criteria. Furthermore, some studies did not clarify whether the participants were positioned on their back or stomach during scanning. Although none of the included studies have investigated the significance of the patient’s positioning on the measured stiffness in the lower back, is reasonable to assume that positioning may have impacted the measured stiffness.

Comparative analysis of MRE methods and findings in the lumbar spine

The results section of this review revealed various acquisition and processing methods used across studies, showcasing the adaptability of MRE in the context of the lumbar spine, although it complicates direct comparison of data. For example, two studies examined the applicability of MRE to the IVD. One studied 16 asymptomatic subjects with an average age of 31 years (25–50 years) and recorded lumbar IVD stiffness at L3/4 = 5.46 ± 1.14 kPa and L4/5 = 6.71 ± 1.51 kPa [18]. In contrast, the other study with 47 subjects ranging between 20 and 71 years found L3/4 = 14.66 ± 2.03 and L4/5 = 14.32 ± 2.32 [16]. While age might explain the difference, the studies also varied in factors such as MRE-system, frequency, and inversion techniques. For example, Walter et al. [16] used a frequency of 80 Hz, while Streitberger et al. [18] used multifrequency with a mean of 60 Hz. A higher frequency may be associated with a higher measured shear stiffness due to the dispersion of the shear modulus [28]. Both studies confirmed the ability of MRE to measure IVD shear stiffness and differentiate different levels of disc degeneration, but they presented opposite results. Walter et al. [16] reported a stiffness increase with a higher degree of degeneration. Contrary, Streitberger et al. [18] reported a decrease in IVD stiffness with increasing degeneration. This observation aligned with their ex-vivo findings in bovine disks which demonstrated a lower magnitude of the complex shear modulus after increasing the load via clamping. Given the distinct experimental setups and image analysis techniques, varied outcomes are expected. Streitberger et al. [18] noted that their chosen inversion algorithm (MDEV) could notably underestimate shear stiffness in stiff tissue. The inversion algorithm selected in Walter’s [16] study, the PFA, also had its limitations. Primarily, it restricts each ROI to a single value, lacking the capability to present a standard deviation of stiffness across a region. Consequently, this limitation hinders the assessment of tissue homogeneity, which could serve as an indicator of degeneration. Moreover, the PFA method does not support the generation of stiffness maps, thereby impeding the visualization of spatial variations in tissue stiffness.

Three studies examined MRE on PSM [12, 14, 20] with stiffness ranging from 1.60 ± 0.14 kPa [14] to 7.03 ± 1.00 kPa [12] These studies had different experimental setups, thus reasons for the variation between studies could be e.g., differences in vibration frequency (100 [14, 20] and 120 hz [12]) and field strength (1.5T [14, 20] and 3T [12]). These findings suggest that the variations in shear stiffness observed in MRE measurements of PSM might be attributable either to the sensitivity of the MRE technique to specific technical factors such as vibration frequency and field strength or to inherent properties of the PSM themselves. Another important factor contributing to PSM stiffness is the histopathological state of the tissue. Myofascial trigger points, commonly found in patients with lumbar myofascial pain, are characterized by entrapment of water molecules due to chemical interactions with glycosaminoglycans. This process increases signal intensity in T2-weighted MRI and leads to elevated stiffness in MRE due to heightened intramuscular pressure [29]. Given that MRE quantifies mechanical properties of tissues, these histological changes could contribute to some of the variations observed in lumbar muscle stiffness measurements [30]. In addition to these muscle-related factors, the role of the thoracolumbar fascia (TLF) in lumbar stiffness should also be considered [31, 32]. The TLF is a dense connective tissue structure that encases and stabilizes the paraspinal muscles [33]. Studies suggest that in patients with chronic LBP, fibrotic changes in the TLF may lead to increased axial stiffness, independent of muscle properties [31]. While the included studies have primarily focused on assessing the stiffness of paraspinal muscles, the potential influence of TLF stiffness remains underexplored.

Five studies investigated various aspects of the use and applicability of MRE on the PM. The measured stiffness of the PM averaged between 0.94 ± 0.09 to 2.33 kPa ± 1.12 between the studies [15, 17, 20,21,22]. Four studies were largely comparable regarding subjects, acquisition, and processing methods, while the last differed in several areas such as vibration frequency (50 Hz [15, 17, 21, 22] and 100 Hz [20]), field strength (3T [15, 17, 21, 22] and 1.5T [20]), and MRE system (custom made [15, 17, 21, 22] vs. commercial [20]). Contrary to the findings for the PSM, the similarities in stiffness measurements across different setups may indicate that MRE is a robust technique for measuring the mechanical properties of the PM. This could be due to the intrinsic properties of the PSM, their anatomical location, or the way they respond to different MRE frequencies and field strengths.

Although IVD degeneration is known to increase with age, Walter et al. [16] only found a weak correlation between subject age and MRE-derived shear stiffness, indicating that changes in shear stiffness due to degeneration may be independent of age. However, further research is needed due to the limited sample size and skewed age distribution. Another study by Hsieh et al. [20] found that females had significantly higher PM muscle stiffness than males. Hsieh mentioned that these results were inconsistent with previous ultrasonography studies, possibly due to the high technical failure rate and low reproducibility of MRE, especially in the PM.

The mechanical behavior of the IVD is known to vary over time due to changes in tissue hydration [34, 35]. Therefore, it is logical to assume that changes in the phase angle, measuring the relationship between tissue viscosity and elasticity could also occur during the day. Walter et al. [16] investigated the effect of diurnal changes on the measured shear stiffness and found no significant difference between morning and afternoon MRE-derived shear stiffness. Streitberger et al. [18] also investigated this by conducting MRE on one subject at seven different days and at different time points and found that the phase angle and shear stiffness varied by about 11%, or about 18% when calculating the ratio between interquartile range and median. The literature has defined values below 30% as indicative for consistent measurements [36]. Thus, it is seemingly not necessary to account for diurnal changes when conducting MRE measurements in the IVD.

Technical considerations in MRE execution

The findings from Numano et al. [17] highlight the importance of actuator placement. Their study showed that MAV in the PM was highest when the actuator was placed centrally on the lumbar spine, whereas MAV in the erector spinae was highest when the actuator was positioned directly over the muscle. This discrepancy likely arises from differences in anatomical positioning and mechanical coupling to the spine. PM, being a deep muscle anchored to the vertebral bodies and transverse processes, receives more efficient vibration transmission through the vertebral column rather than direct muscle stimulation. In contrast, erector spinae, a more superficial muscle group, absorbs vibrations most effectively when the actuator is directly placed over it. Overall, these findings demonstrate the importance of considering the placement of the vibration pad to ensure effective transmission of vibrations into the target tissue.

Other technical considerations are related to the anisotropy of muscles which may cause shear wave displacements to be induced primarily perpendicular to the muscle fibers [37, 38]. Additionally, research has shown that shear wave velocity is influenced by the direction of wave propagation against the muscle fibers [39, 40]. This is particularly relevant for lumbar spine MRE, where different muscle groups and disc structures may introduce heterogeneous propagation patterns.

A related factor is the choice of sequence, as investigated by Co et al. [24]. Their study demonstrated that SE-EPI sequences reduced scan times by at least 51% compared to GE (90s vs. 270s), potentially minimizing patient movement, vibration-induced fatigue, or tissue stiffness changes during prolonged scans. Importantly, their analysis of OSS-SNR values showed no significant difference between SE-EPI and GE, and a good correlation between stiffness values.

Another notable aspect from Co et al. [24] is the difference in shear stiffness values between 2D and 3D acquisitions, where 2D-derived stiffness values were significantly higher; around 0.7 kPa in NP and at least 26 kPa in AF. The significantly higher value in the AF may be attributed to structural and compositional differences in the region. The AF is anisotropic and considerably stiffer than the NP, leading to the propagation of more complex waves. While 3D acquisitions provide volumetric data that better capture wave propagation in multiple planes, offering more robust and physiologically accurate stiffness measurements, they also come with increased acquisition time, which can be restrictive in a clinical setting. 2D acquisitions, while faster and less susceptible to motion artifacts, rely on assumptions about through-plane wave propagation, which can lead to an overestimation of stiffness values. The lack of full volumetric coverage may also limit the ability to accurately assess anisotropic properties, particularly in regions like the AF, where wave behaviour is highly directional. The implementation of SE-EPI sequences, as explored by Co et al., offers a potential solution by significantly reducing scan times while still enabling multi-slice acquisitions, thereby improving the feasibility of 3D MRE in a clinical setting.

Other relevant considerations which were not addressed in the included studies could be the choice of frequency, amplitude, shape and size of the actuator, field strength, physiological conditions of the patient such as local body fat (in the area between the actuator and the examined area), positioning (prone vs. supine), and use of materials to ensure optimal contact between the actuator and the patient.

Evaluation of the repeatability and reproducibility of MRE

The included studies that investigated repeatability and reproducibility of MRE used different terminologies, sometimes also not aligned with the conventional terminological definitions [26]. Repeatability is typically defined as the ability to achieve consistent results under identical conditions. Reproducibility refers to the ability to achieve consistent results under varying conditions. In the context of MRE, ‘conditions’ may include the operator conducting the scan, the type of MRE equipment, and other aspects related to the experimental setup [26]. In the study by Habe et al. [22], both repeatability and reproducibility were addressed. For repeatability, the same individuals were scanned multiple times by the same operator without changes. In their reproducibility test, subjects were scanned by different operators, clearly distinguishing the two types of tests. The study by Streitberger et al. [18] indicated that reproducibility was assessed by scanning a subject multiple times at different times, which seems more akin to a repeatability test. Similarly, the study by Hsieh et al. [20] described their reproducibility test as merely repeating a scan five times. Such differences in approaches and definitions may lead to misunderstandings and potential misinterpretation of results, highlighting the importance of precise definitions and distinctions between repeatability and reproducibility for research purposes. With that in mind, Habe et al. [22] reported a %RC similar to that required by the Quantitative Imaging Biomarker Alliance (QIBA) for liver MREs, and their reproducibility study demonstrated almost perfect reproducibility (ICC > 0.90) with different operators in the PM. Interestingly, the study by Hsieh et al. [20] reported a low reproducibility (CV > 20%) in the PM in five repeated scans. The studies differed in several areas such as MRE equipment (custom [22] vs. commercial [20]), frequency (50 [22] vs. 100 [20]) and positioning (supine [22] vs. prone [20]). Hsieh at al. [20] also reported a low reproducibility for the PSM, whereas chevalier et al. [23] reported fair to excellent reproducibility. Though these studies had similar experimental setups, they differed in other areas such as ROI placement (two separate for erector spinae [23] and multifidus vs. one [20]), and orientation (coronal [23] vs. axial [20]). Using a coronal orientation also allowed larger ROIs. These differences may indicate the importance of ROI selection and slice orientation.

Regarding repeatability, Walter et al. [16] found no significant differences between morning and afternoon MRE-derived shear stiffness in the IVD, and Streitberger et al. [18] found a variability of about 11% in complex modulus and phase angle, indicating good reproducibility. The findings from Co et al. [24] reinforce these results, demonstrating that SE-EPI derived stiffness generated highly producible and repeatable stiffness measurements, with minimal observer bias. Overall, MRE showed promising repeatability and reproducibility in certain contexts, but performance seemingly varied based on specific applications.

Clinical value and future development of MRE

MRE has the potential to become a valuable tool in the assessment of lumbar spine disorders, particularly in differentiating normal and pathological tissue properties [16, 18]. Compared to conventional MRI, MRE provides a quantitative assessment of tissue stiffness, which may aid in diagnosing conditions such as IVD degeneration, chronic LBP, and myofascial disorders. Additionally, MRE could be useful for monitoring treatment effects in both rehabilitation and surgical settings.

Postoperatively, MRE may help evaluate biomechanical changes in the PSM and IVD, providing insight into muscle atrophy, fibrosis, and changes in tissue integrity following surgical interventions. For instance, a reduction in muscle stiffness following surgery may indicate functional recovery, whereas an increase in stiffness could suggest fibrotic changes or impaired mobility in the operated region. Conventional MRI primarily provides qualitative assessments of muscle changes, which comes with some limitations. For example, studies on MRI in muscle disorders highlight that disease activity, as well as muscle damage, are difficult to assess since clinical evaluation of muscle strength is partly subjective, and reliable biomarkers of disease activity are lacking [41]. Furthermore, while muscle disease activity is often associated with high T2 signal, the interpretation of what constitutes a “high” signal is subjective [41]. The integration of MRE could offer a more precise assessment, potentially overcoming some of these limitations by providing a direct, quantitative measurement of biomechanical tissue properties.

Despite its promise, several challenges remain before MRE can be fully integrated into routine clinical practice. The lack of standardized scanning protocols and post-processing algorithms limits comparability between studies. Further research is needed to establish normative stiffness values and determine the clinical relevance of MRE-derived biomarker. Additionally, large-scale, multicenter studies including symptomatic patient populations are required to validate the diagnostic utility of MRE.

Although MRE scanning itself is not particularly time-consuming, the patient setup process remains a challenge. Proper placement of the actuator on the back, securing it with a belt, and positioning the patient—all while ensuring the actuator remains in place—can be cumbersome. This preparation phase adds to the overall scanning time, reducing patient throughput, which is a crucial factor in a clinical setting. Future studies should explore optimizing the setup, including improved actuator designs. For instance, a system where the actuator is pre-mounted on an adjustable support integrated into the scanner table could streamline the process, allowing patients to simply position themselves without requiring extensive manual adjustments, making MRE more feasible for clinical implementation.

Comparison of MRE and other techniques

MRE is one of several imaging techniques used to assess tissue stiffness, with ultrasound elastography being a commonly used alternative. While both modalities provide quantitative measurements of mechanical tissue properties, they differ in underlying principles and applicability. Ultrasound elastography is widely used due to its accessibility, lower cost, and real-time imaging capabilities. It relies on either shear wave elastography, or strain elastography to estimate tissue stiffness. However, its penetration depth is limited, making it less suitable for deep-seated structures such as the PM. Additionally, ultrasound elastography is more operator-dependent, introducing potential larger variability in measurements. In contrast, MRE offers deeper tissue penetration, and is less dependent on operator skill. It provides high-resolution, quantitative stiffness maps and is particularly valuable for structures embedded within complex anatomical regions, such as the lumbar spine. However, MRE is more resource-intensive, requiring specialized hardware and longer acquisition. For these reasons, the choice of modality should be based on the clinical question at hand. In addition to MRE and ultrasound elastography, mechanical indentation devices offer an alternative method for assessing tissue stiffness, particularly in research settings. Unlike imaging-based techniques, indentation devices provide direct mechanical measurements by applying controlled force to the surface and recording displacement. This method has been used to evaluate spinal stiffness in conditions like LBP [42]. However, indentation techniques lack imaging capabilities and are primarily used for localized mechanical assessments rather than comprehensive diagnostic evaluations. In contrast, MRE is fully integrated with MRI, allowing stiffness measurements to be directly correlated with anatomical imaging.

Limitations of this review

Despite a thorough search, only 11 studies were included in this scoping review, which highlights gaps in the literature and knowledge in the field. While expanding the scope to encompass ex vivo and animal studies might have resulted in a larger volume of literature, such inclusion could have diluted the review’s focus on the application of technology in clinical settings. Based on the included studies, conclusions regarding effectiveness or applicability of MRE for the lumbar spine should be drawn with caution, given the varying outcomes reported. It is worth noting that unlike systematic reviews, scoping reviews generally do not yield results synthesized from multiple sources of evidence through a formal process of methodological appraisal to determine evidence quality [43]. A systematic review would provide deeper analysis but require a more focused question and stricter criteria, possibly missing the wide range of ongoing research and innovations. However, this scoping review sets the stage for future systematic reviews by identifying key research areas and gaps, guiding future and more focused research questions. Finally, the purpose of this review was to broadly examine the use of MRE on the lumbar spine in vivo, from a more clinical perspective. There are several technical elements beyond those mentioned in this review, which may influence the outcome of the respective measurements [26].

Perspectives

The included studies demonstrated the potential of MRE for characterizing lumbar spine tissues and their biomechanical properties. However, there are still areas that need further investigation to strengthen the evidence base, optimize techniques, and enhance the clinical applicability of MRE. The following aspects should be considered in future studies:

• 1. Standardization of experimental protocols: To facilitate comparison across studies and to improve the reproducibility and reliability of results, it is essential to standardize various aspects of MRE protocols. Some examples to consider include:

  • a. Participant positioning: Standardizing the positioning during MRE examinations may minimize variations in tissue properties and shear wave propagation.

  • b. MRE parameters: Consistency in MRE sequence parameters such vibration frequency and inversion algorithm may reduce discrepancies in measurements.

  • c. Vibration systems and placement: The efficiency and consistency of mechanical wave transmission to the target tissues depend on the vibration system and its placement on the body. Standardizing the vibration systems and placement guidelines across studies can minimize variability in wave transmission and thus enhance the reliability of MRE measurements.

  • d. ROI selection: The size, shape, and placement of the ROI may influence the calculated stiffness values. Developing standardized guidelines for ROI selection and analysis will improve comparability of data.

• 2. Larger and more diverse populations: The studies in this review included only asymptomatic participants, and some had limited sample sizes. Future studies should consider larger and more diverse populations, including symptomatic participants, to better understand the relationship between MRE-derived measurements and clinical symptoms or conditions.

• 3. Longitudinal studies: Longitudinal research designs would enable the assessment of changes in MRE-derived measurements over time and their relationship with the progression of lumbar spine conditions, such as degeneration or chronic LBP.

• 4. Comparison with other imaging techniques: The diagnostic accuracy and clinical usefulness of MRE should be compared with other imaging techniques, such as T1- or T2-weighted MRI, to establish the added value of MRE in the assessment of lumbar spine conditions.

• 5. Evaluation of MRE in clinical practice: Research should focus on investigating the potential of MRE in the clinical setting, including its ability to guide patient management, predict treatment outcomes, and assess the effectiveness of therapeutic interventions.

Conclusion

The current evidence on the use and applicability of MRE of the lumbar spine is promising but limited. MRE is a promising non-invasive imaging technique for assessing tissue stiffness in the lumbar spine, with the potential to detect and monitor intervertebral disc degeneration, evaluate lumbar muscles, and assess the effects of some therapeutic interventions in muscles. The small number of studies and their heterogeneous methodologies warrant further research in this promising diagnostic technique. Future research should focus on conducting larger, multicenter studies to establish the diagnostic accuracy, reproducibility, and clinical utility of MRE of the lumbar spine. Standardized protocols should be developed to ensure consistent and reliable measurements across different centers and populations.