Tendon evaluation with ultrashort Echo Time (UTE) MRI: A Systematic Review

Abstract

Tendon disease ranks among the leading reasons patients consult their general practitioners, comprising approximately one-third of musculoskeletal appointments. Magnetic resonance imaging (MRI) is regarded as the gold standard for assessing tendons. Due to their short transverse relaxation time (T2), Tendons show up as a signal void in conventional MRI scans, which employ sequences with echo times (TEs) around several milliseconds. Ultrashort echo time (UTE) sequences utilize TEs that are 100–1,000 times shorter than those used in conventional sequences. This enables the direct visualization of tendons and assessment of their relaxation times, which is the basis for quantitative MRI. A systematic review was conducted on publications after 1990 in Google Scholar and PubMed databases. The search terms “ultrashort echo time,” “tendon,” and “UTE” were used to identify studies related to this investigation. This review discussed the current knowledge in quantitative UTE-MRI imaging of tendons. Quantitative UTE-T1, UTE-T2*, UTE-MT, and UTE-T1ρ techniques were described, and their reported applications in the literature were summarized in this review. We also discussed the advantages and challenges of these techniques and how these quantitative biomarkers may change in response to tendon pathology.

1. Introduction

Tendon disease ranks among the leading reasons patients consult their general practitioners, comprising approximately 30% of musculoskeletal appointments [1]. Tendinopathy is more prevalent in athletes and affects various regions like the ankle, shoulder, elbow, and knee. Multiple studies on athletes have yielded epidemiological findings indicating the prevalence of patellar tendinopathy at 15%, Achilles tendinopathy at 12.5%, and rotator cuff tendinopathy at 23% [2-4]. In both the general population and among workers, a 2-4% prevalence has been noted for upper extremity tendinopathy, with higher rates observed in older adults [5-8].

Normal tendons are composed of fibroblasts, called tenocytes, and are surrounded by extracellular matrix (ECM). ECM is predominantly made of tightly packed collagen fibers oriented along the primary loading direction. Collagen accounts for around 60-85% of their dry weight. Approximately 1-5% of the dry weight of tendons is composed of proteoglycans. Tendinopathy involves various biochemical alterations, including aberrations in the composition of the collagenous matrix and a notably elevated turnover rate of the collagenous matrix[12-14]. Elevated matrix metalloproteinases (MMP) activity and increased degradation of collagen signify heightened matrix breakdown. Conversely, elevated collagen gene expression and a low pentosidine level, indicating a relatively young collagenous matrix, suggest an upturn in matrix synthesis[13]. Tendinopathic tendons are also associated with increased amounts of proteoglycans (glycosaminoglycans), such that the large proteoglycans bind water leading to tissue swelling [15,16].

Magnetic resonance imaging (MRI) is widely regarded as the definitive imaging modality for assessing tendons, offering exceptional differentiation of soft tissues [9]. Tendons are comprised of highly ordered collagen fiber structures, leading to strong dipole-dipole interactions and, thereby, short apparent transverse relaxation times (T2*s). As a result, tendons cannot be visualized directly using conventional MRI sequences [10]. Specifically, conventional MRI sequences utilize echo times (TEs) of several milliseconds (ms), which is longer than tendons’ T2*. Although there is clinical value in seeing the outline of tendons, there is no possibility of quantification or detecting subtle changes in the structure of tendons using conventional sequences [11].

Ultrashort echo time (UTE) sequences are a class of clinically compatible sequences capable of providing TE values less than 0.1 ms [17]. These sequences make it possible to image short T2 tissues, such as tendons, directly with a high signal.

Qualitative analysis has been the primary approach for evaluating the musculoskeletal (MSK) system using conventional sequences in MRI for a considerable period. Quantifying the signal enhancement in tendons after intravenous administration of gadolinium has been suggested for detecting tendon diseases [72]. However, contrast administration is associated with discomfort, health risks and cost [73]. With increased research in MSK MRI, noninvasive quantitative imaging has become particularly interesting due to its potential to provide earlier disease detection, rigorous comparative investigations, and more robust monitoring tool for disease progress and treatment[18,19]. Figure 1 shows a representative Achilles tendon imaged at 3T using conventional T₁-weighted CUBE (three-dimensional fast spin echo sequence in General Electric (GE) scanner) and three-dimensional (3D) UTE sequences. The tendon demonstrated no detectable signals on the conventional MR image (Figure 1a), while it showed a high signal when imaged with 3D-UTE (Figure 1b).

Figure 1:

A representative Achilles tendon imaged at 3T (a) using a clinical T₁-weighted CUBE sequence (TR= 926 ms, TE=18 ms) where the tendon (blue arrow) demonstrated no detectable signals and using (b) a UTE MRI (TR= 500 ms, TE=0.032ms) sequence where the tendon (blue arrow) demonstrated a high signal required for potential quantitative assessments.

In this review study, we summarize the quantitative studies conducted on tendons utilizing UTE sequences, including T2*, T1, magnetization transfer (MT), and T1ρ sequences. By examining the outcomes of these investigations, we aim to provide a summarized yet comprehensive overview of the current state of knowledge regarding the application of UTE sequences for assessing tendon health and pathology.

2. Materials and Methods

This review conducted between April and June 2023 aimed to provide a thorough description of the current quantitative UTE applications on tendons. The literature search was performed in PubMed, Web of Science, and Google Scholar databases using the following keywords: “Ultrashort echo time”, “tendons”, and “UTE” published after the year 1990. The results were screened first automatically and then through title and abstract reading to exclude duplicated records, non-English written reports, non-quantitative studies, animal studies, and review articles. Study selection followed the PRISMA 2020 guidelines, as summarized in Figure 2.

Figure 2:

Study selection flowchart based on PRISMA 2020 guidelines.

3. Results

All reviewed quantitative UTE-MRI-based investigations performed on tendons are summarized in Table 1. The reviewed studies were categorized based on the employed UTE-MRI technique including UTE-T2*, UTE-T1, UTE-MT, and UTE-T1ρ, which are described in detail in the following sections.

Table 1.

Summarized UTE-MRI-based investigations performed on tendons.

Author/year	Quantitative biomarker	Tendon	Subject and sample size	Findings/Goals
Robson et al. (2004)[20]	T2*	Achilles (AT)	Healthy controls (4) and Achilles tendinopathy patients (4)	Feasibility study
Du et al. (2010)[21]	T1ρ	Achilles	Healthy volunteers (5) and cadaveric ankle specimens (6)	Development of 2D UTE-T1ρ sequence
Hodgson et al. (2011)+[22]	MT	Achilles	Healthy volunteers (8) and psoriatic arthritis patients (1)	Demonstrate MT effects in Achilles tendon at >1kHz off-resonance
Syha et al. (2011)[23]	MT, T2*	Achilles	Healthy volunteers (8) and Achilles tendinopathy patient (1)	Feasibility study- MTR calculation
Wright et al. (2012)[24]	T1	Achilles	Healthy volunteers (8) and patients with psoriatic arthritis and Achilles tendinopathy (7)	Calculation of T1, comparison of saturation recovery UTE (SR-UTE) and variable flip angle UTE (VFA-UTE) against gold standard inversion recovery spin echo (IR-SE)
Juras et al. (2012)+[25]	T2* bicomponent	Achilles	Healthy volunteers (10) and chronic Achilles tendinopathy patients (5)	Compare T2* in healthy and pathologic AT, at two field strengths (3T and 7T)
Juras et al. (2013)[26]	T2* bicomponent	Achilles	Healthy volunteers (10) and patients with painful tendon (10)	Compare mono- and bi-exponential T2* analysis in healthy and degenerated AT
Du et al. (2013)[27]	T1 ρ	Achilles	Ex-vivo Achilles tendon samples (6)	To demonstrate magic angle effect in T1rho imaging
Grosse et al. (2013)[28]	MT	Achilles	Healthy volunteers (17) and Achilles tendinopathy patients (13)	Compare MTR in healthy and pathological AT and correlation between clinical symptoms and MTR values
Chang et al. (2014)[29]	T2*	Achilles	Ex-vivo Achilles tendons (6)	Compare T2* of AT at baseline in air and after immersion into saline, Fomblin, or perfluorooctyl bromide (PFOB)
Chang et al. (2014)[30]	T2*	Achilles	Ex-vivo Achilles tendons (4)	Compare T2* of AT before and after freeze-thaw cycles.
Syha et al. (2014)[31]	MT	Achilles	Healthy male volunteers (7)	To investigate short-term exercise-induced effects of hydration state of the Achilles tendon by means of MTR and tendon volume
Grosse et al. (2015)[32]	T1, T2*, MT	Achilles	Healthy volunteers (21) and Achilles tendinopathy patients (17)	Compare T1, T2* and MTR between healthy volunteers and tendinopathy patients
Grosse et al. (2015)[33]	T1, T2*	Achilles	Healthy volunteers (21) and healthy recreational long-distance runners (18)	Compare T1 and T2* of AT of heathy volunteers and healthy recreational long-distance runners
Chang et al. (2015)[34]	MT	Multiple tendons (tibialis, flexor digitorum longus, …)	Ex-vivo tendon samples (15)	To determine if MTR could differentiate between tendons under different states of tensile load and to compare these changes between normal versus degenerated tendons.
Chang et al. (2015)[35]	T2* bicomponent	Multiple tendons (tibialis anterior, tibialis posterior, flexor digitorum longus, …)	Ex-vivo tendon samples (10)	To determine if the application of tensile force alters the single- or bi-component T2* values of human tendons. Also, to determine if single- or bi-component T2* values differ when measured with 2D-UTE, 3D-UTE, or 3D-UTE-Cones sequences
Gärdin et al. (2016)[36]	T2*	Achilles	Healthy controls (10) and chronic mid portion Achilles tendinosis patients (20)	To investigate if the T2 * of Achilles tendons can discriminate between chronic Achilles tendinosis and healthy controls; to correlate with clinical score; to evaluate its short-term repeatability; and to estimate minimal detectable change
Kijowski et al. (2017)[19]	T2* bicomponent	Patellar(PT)	Healthy controls (10) and patellar tendinopathy patients (11)	To compare UTE-T2* parameters of PT between healthy volunteers and patients with patellar tendinopathy
Chaudhari et al. (2017)[37]	T2	Patellar	Healthy volunteers (11)	To develop a radial, UTE double-echo steady-state (DESS) sequence for rapid, signal-to-noise ratio (SNR)-efficient, and high-isotropic-resolution morphological knee imaging.
Qiao et al. (2017)[38]	T2*	Achilles	Healthy volunteers (10) and patients with diseased Achilles tendons (11)	To compare T2* value of healthy and diseased AT with a 3D-UTE sequence and analyze the correlation between T2* value and clinical scores
Jerban et al. (2017)[39]	T1, MT, T2*	Peroneus longus and brevis	Ex-vivo tendon samples (6)	To employ UTE-MRI for fatigue fracture detection in fibula cortical bone
Ma et al. (2018)[40]	T1ρ	Patellar	Ex-vivo knee samples (4) and healthy volunteers (6)	To use UTE sequences to measure T2*, T1 and magnetization transfer ratio (MTR) variations of tendon samples under static tensile loads
Chen et al. (2018)++[41]	T2*	Achilles	Healthy volunteers (5)	Feasibility of UTE imaging and T2* measurement at low fields
Chen et al. (2018)[42]	T1, T2*, MT	Achilles	Ex-vivo ankle specimens (5)	To examine the normal MR morphology of the cadaveric Achilles tendon and enthesis at 3 T using novel three-dimensional ultrashort echo time (3D-UTE) Cones sequences, and at 11.7 T using conventional MRI sequences.
Ma et al. (2018)[43]	MT	Achilles	Ex-vivo tendon specimens (3)	To accelerate the quantitative UTE imaging using a time-efficient 3D multispoke Cones sequence with MT (3D-UTE-Cones-MT) and signal modeling.
Zhu et al. (2018)[44]	MT	Rotator cuff	Ex-vivo tendon specimens (20)	To assess the sensitivity of MT modelling to magic angle effect
Chen et al. (2018)[45]	MT, T2*	Achilles	Healthy volunteers (7) and psoriatic arthritis patients (9)	To use 3D-UTE-Cones sequence for quantification and comparison of T2* and magnetization transfer ratio (MTR) of Achilles tendon and its enthesis of healthy volunteers and psoriatic arthritis patients using a 3T scanner.
Ma et al. (2018)[46]	T1	Patellar	Healthy volunteers (16)	To measure T1 relaxations for the major tissues in whole knee joints using 3D-UTE Cones AFI-VFA(actual flip angle imaging-variable flip angle) method
Sharafi et al. (2019)[47]	T1ρ bicomponent	Patellar and Achilles	Healthy volunteers (10)	To develop a 3D T1ρ prepared zero echo time (ZTE)-based pointwise encoding time reduction with radial acquisition (3D-T1ρ-PETRA) sequence
Jerban et al. (2019)[48]	MT, T1	Anterior tibialis and posterior tibialis	Healthy young (26) and elderly women (22)	To determine if UTE-MT-modeling is a quantitative technique sensitive to the age-related changes of tendons.
Guo et al. (2019)[49]	MT, T1, T1ρ, T2*	Supraspinatus	Ex-vivo samples (36)	To assess if quantitative MR (T2*, T1, MT, T1ρ) can detect abnormalities in tendon samples treated with collagenase
Liu et al. (2019)[50]	T2* bicomponent	Patellar	Healthy volunteers (5) and ex-vivo samples (6)	To assess the technical feasibility of three-dimension multi-echo fat saturated ultrashort echo time cones (3D FS-UTE-Cones) acquisitions for single- and bicomponent T2⁎ analysis in PT
Krämer et al. (2019)[51]	T1, T2*	Quadriceps and patellar	Healthy volunteers (5)	To quantify both T1 and T2* in the quadriceps and patellar tendons of healthy volunteers and visualize the results based on 3D segmentation by using bivariate histogram analysis
Wan et al. (2020)[52]	T1, T1ρ, MT	Patellar	Ex-vivo knee samples (9)	To investigate if 3D UTE Cones sequence can be accelerated using a longer spiral sampling window
Papp et al. (2020)[53]	T2*	Patellar	Patellar tendinopathy patients (5)	To compare fractional order and exponential fitting in UTE imaging of patellar tendon
Wu et al. (2020)[54]	T1ρ, T2*	Patellar	Ex-vivo tendon samples (5)	Cones-AdiabT1ρ shows a much-reduced magic angle effect compared to Cones-CW-T1ρ and Cones-T2*
Xie et al. (2020)[55]	T2*	Rotator cuff (RCT)	Arthroscopic rotator cuff repair patients (25)	To evaluate and characterize the healing process of the repaired rotator cuff based on longitudinal changes in UTE-T2* values, clinical outcomes, and repair status in patients after arthroscopic rotator cuff repair (ARCR)
Jang et al. (2020)[56]	T2* bicomponent	Patellar	Ex-vivo knee joints (6) and healthy volunteers (4)	To assess the feasibility of using a single scan ramped hybrid encoding (RHE) method for faster bicomponent T2* analysis of the human knee joint
Ashir et al. (2020)[57]	MT, T2	Rotator cuff	Healthy controls (10) and symptomatic tendinopathy patients (14)	To assess the bilateral RCTs in shoulders of patients with unilateral symptomatic tendinopathy and control subjects using a UTE-MT sequence
Breda et al. (2020)[58]	T2*	Patellar	Athletes with patellar tendinopathy (65)	To assess regional variability in binding states of water in PT
Breda et al. (2021)[59]	T2*	Patellar	Athletes with patellar tendinopathy (76)	To assess the association between PT T2 * relaxation times and symptom severity/clinical outcome.
Malmgaard-Clausen et al. (2021)[60]	T2*	Patellar	Healthy controls (25) and early tendinopathy patients (65)	To investigate the difference in T2 * between patients with early tendinopathy and healthy controls, and to investigate the relationship between T2 * and clinical outcomes, tendon size, and mechanical properties.
Agergaard et al. (2021)[61]	T2*	Patellar	Chronic patellar tendinopathy patients (44)	The purpose was to investigate if the load magnitude influenced
Anjum et al. (2021)[62]	T2*	Achilles	Healthy volunteers (6)	To examine multi-component relaxation modelling in multi-echo ultra-short echo time (UTE) data of human AT
Malmgaard-Clausen et al. (2021)[63]	T2*	Achilles	Early Achilles tendinopathy patients (69)	Randomized controlled trial using UTE-T2* values as one of the secondary outcome measures
Loegering et al. (2021)[64]	T2* bicomponent	Achilles	Asymptomatic young (13) and asymptomatic old volunteers (13)	To investigate the relationship between UTE findings and imaging assessment of sub-clinical tendinopathy
Agergaard et al. (2021)[65]	T2*	Patellar	Chronic patellar tendinopathy patients (15)	To evaluate test–retest reproducibility of UTE MRI T2* mapping of tendinopathic patellar tendons and to evaluate the intra- and inter-observer reproducibility of the measurement
Fang et al. (2022)[66]	MT, T2*	Achilles	Healthy controls (5) and Runners (32)	To evaluate UTE-MT sequence in the detection of changes in Achilles tendons of amateur marathon runners before and after long-distance running
Jerban et al. (2022)[67]	MT, T1	Anterior tibialis, posterior tibialis, and Achilles tendon	Osteoporosis (31) vs Osteopenia (13) vs Normal (30)	To determine whether UTE-MT-modeling could detect differences in tendon quality across osteopenia and osteoporosis patients.
Okuda et al. (2022)+++[68]	T2*	Patellar	Healthy volunteers (12)	To quantify and differentiate the UTE-T2* values of normal knee ligaments and tendons using a 1.5-T MRI scanner
Moazamian et al. (2023)[69]	MT, T1	Achilles	Healthy young (27) and psoriatic arthritis patients (26)	To investigate the use of UTE-MRI techniques (T1 and MT modeling) for imaging of the Achilles tendons and entheses in patients with psoriatic arthritis (PsA) compared with asymptomatic volunteers
Jerban et al. (2023) [70]	T1, T1ρ	Achilles	Ex-vivo samples (11)	To investigate the relationship between quantitative UTE- adiabatic T1ρ measures and the biomechanical properties of Achilles tendons and entheses.

⁺7T

⁺⁺0.35T

⁺⁺⁺1.5T

3.1. UTE T2*

UTE T2* measurement of tendons has been widely investigated by acquiring UTE images at varying TEs. UTE T2* has been the most widely studied sequence for quantitative tendon imaging. Robson et al. were among the first to report T2* relaxation values using UTE sequences for Achilles tendons in vivo. They observed that two-dimensional (2D) UTE sequences provided anatomical detail which was not apparent with conventional sequences [20]. Juras et al. used a 3D-UTE sequence for bi-component T2* quantification of the Achilles tendon at 7T[25]and 3T[26] field strengths. They found that bi-component exponential analysis was superior to single-component exponential analysis for diagnosing Achilles tendinopathy. Single-component T2* and the short-component T2* (from bi-component analysis) values for patients (n=10) were significantly higher than those in controls. Kijowski et al. in their study on patellar tendinopathy patients (n=11), found similar results and concluded that the fast relaxing component fraction (i.e., short component) and its relaxation time were more sensitive in detecting tendon disease[19]. Liu et al. employed a fat-saturated (FS) multi-echo 3D-UTE-Cones sequence to assess the relaxation times and fractions of both short and long components within tendons. Specifically, Cones refers to the spiral trajectories used to fill the k-space during this UTE acquisition. They demonstrated a predominance of short components in tendons, similar to previous studies [50]. Jang et al. suggested using a rapid single scan ramped hybrid encoding for bi-component T2* mapping in knee joint which provided for shorter scan times (9 minutes). This technique captures all the required echoes in a single acquisition, reducing motion sensitivity [56]. A recent study by Loegering et al. on 13 young and 13 old volunteers using 3D-UTE multi-echo with radial acquisition attempted to compare their slow and fast relaxing components. They found that older tendons exhibited a significantly higher short-component T2* and a smaller short-component fraction than young tendons, highlighting the sensitivity of the sequence for detecting early and sub-clinical tendinopathy[64].

Chang et al. conducted two distinct ex vivo experiments to explore the influence of tendon loading on T2* measurements. Results revealed that when subjected to lower loads (10N), no significant alterations were detected in the T2* values [35]. However, when higher loads ranging from 15N to 30N were applied, a notable load-dependent decrease in T2* was observed [39]. T2* values of tendons in ex vivo experiments do not seem to be significantly affected by immersion in fluids such as saline or perfluoropolyether (e.g., Fomblin, Ausimont, NJ, USA) [29], as well as by repeated freeze-thaw cycles [30]. Therefore, ex vivo findings are expected to apply to in vivo studies. Grosse et al. demonstrated the comparable impact of physical activity, specifically long-distance running, on UTE T2* values in the Achilles tendon of a group of long-distance runners (n=18)[33]. In their respective studies, Gardin et al. n=20)[36] and Qiao et al.(n=10)[38] found higher UTE-T2* values for diseased Achilles tendons when compared to healthy controls. A larger study by Malmgaard-Clausen et al. on 65 early tendinopathy patients (Achilles, patellar) found significantly higher UTE-T2* values in pathological tendons[60].

Xie et al. validated UTE T2* sequences against established clinical studies by following up on patients with rotator cuff repair (n=25) longitudinally across 24 months. The clinical outcomes in patients showed significant correlations with UTE-T2* values at six months and 12 months, highlighting the potential of using UTE-T2* mapping to track the tendon-healing process noninvasively [55]. They observed an increased T2* from 3 to 6 months and then a decreased T2* at 12 months post-injury, likely implying different healing phases. Agergaard demonstrated substantial/high intra and inter-observer reproducibility for UTE-T2* values in patellar tendons, proving the feasibility of the use of UTE-T2* mapping in research and clinical practice [65]. In another large study by Breda et al.UTE-T2* relaxation times in patellar tendon decreased in 76 patients undergoing exercise therapy [59]. The baseline higher T2* in the injured tendons could be explained by more free water and disrupted tendon fibers, which might be restored by the exercise therapy. Figure 3 shows the UTE T2* pixel map and fitting curve in a representative cadaveric Achilles tendon imaged in the sagittal plane using UTE-T2* cones sequence.

Figure 3.

(a) UTE-T2* pixel map generated for Achilles tendon from single component T2* decay fitting performed on each pixel. (b) UTE-T2* fitting curve using six different echo times (TEs = 0.032-30 ms) within an exemplary region of interest (indicated in while box in (a)).

3.2. UTE T1

Obtaining T1 measurements of tendons using UTE images can be accomplished by acquiring images with varying repetition time (TR) or flip angle (FA) to be fitted in exponential models. However, accurately determining the FA for tissues with short T2* can be challenging due to B1 field inhomogeneities. Estimating the actual flip angle (AFA) using a calculated B1 has been recommended for accurate T1 measurement [46].

Wright et al. observed that spondyloarthritis patients with Achilles tendinopathy exhibited elevated T1 values compared to healthy individuals [24]. A histological and biochemically correlational study by Guo et al. found higher T1 values in rotator cuff tendon specimens treated with collagenase enzyme, which is known to degrade the collagen fibers[49]. T1 relaxation times are reported to be less sensitive to slight pathological tendon alterations compared to T2* and MTR values[32]. In general, T1 of tendons is reported to be insignificantly affected by loading, age, and diseases like osteoporosis/osteopenia. [39,48,67]. Figure 4 shows the UTE T1 pixel map and fitting curve in a representative cadaveric Achilles tendon imaged in the sagittal plane using UTE-T1 sequence. In a recent study, Moazamian et al. reported higher UTE-T1 values in Achilles tendon and its enthesis of patients with psoriatic arthritis (n=26) compared with asymptomatic young subjects (n=27) [69].

Figure 4.

(a) UTE-T1 pixel map generated for Achilles tendon from single component T1 recovery fitting performed on each pixel. (b) UTE-T1 measured for an exemplary region of the tendon, indicated in while box in (a), using VFA method using four different acquisitions with differing flip angles (FAs=5-30°).

3.3. UTE MT

Magnetization Transfer (MT) imaging in MRI involves selectively saturating the magnetization of bound protons or macromolecules using an off-resonance radiofrequency pulse. The macromolecular spins have a much broader absorption line shape in frequency domain than the liquid spins, making them much more sensitive to an appropriately placed off-resonance saturation pulse. This preferential saturation of the macromolecular spins can be transferred to the surroundings liquid spins. The transferred magnetization causes signal attenuation in the surrounding water protons, resulting in contrast differences and providing information about tissue composition and macromolecular content in MT images [74,75].

UTE-MT ratio (UTE-MTR) is a common MT-related index in the literature, defined as MTR = (S_OFF - S_ON) / S_OFF. Where S_OFF is the signal intensity in the image acquired without the magnetization transfer pulse preparation. S_ON is the signal intensity in the image acquired with the off-resonance magnetization transfer pulse preparation.

Syha et al. used MTR as a feasible method for quantifying interactions between bulk water and macromolecules in tendons. His study also found lower MTR values in a tendinopathic tendon compared with 16 healthy volunteers; however, only one pathological sample was studied [23]. Further MRI research by the group found a lower mean MTR for pathological Achilles tendons (14 patients) compared with healthy controls[28]. The MTR obtained from pulses at 2kHz demonstrated the highest test performance (sensitivity and specificity) to distinguish the slightly- and severely pathologically altered tendons from the healthy ones[32].

Syha et al. examined the effect of ankle-straining activity like cross-country running on the Achilles tendon (n=7) on the MTR. They found a significant increase in MTR in most tendon regions, which was attributed to the depletion of free water molecules after running[31]. Reduction of the MTR values after long-distance running was observed in other investigations [66]. In his experiment, Chang et al. used a 2D UTE MT sequence (MT power = 1200, frequency offsets-1.5, 3, and 5kHz) and a 3D-UTE-MT sequence (MT power 540, frequency offset 1.5kHz) for comparing MTR before and under tensile loading of tendons. An increase in MTR was observed by all sequences on the application of tension[34]. Later, the same group showed a non-significant increase in tendon MTR values under static tensile loads in another study when 3D-UTE-Cones was used[39]. These studies indicate that UTE-MT sequences can detect changes in the hydration state of tendons[34,39,66]

Using MT pulses with higher power levels and larger frequency offsets may enhance the induced saturation on proton pools with extremely short T2, as found in macromolecules such as collagen. MTR values are functions of the MT pulse power level and the frequency offset, which makes MTR hardly reproducible between different studies. Hodgson et al. utilized MT modeling performed on 2D UTE MT sequences with four MT pulse powers (250, 500, 1000, 1400) and off-resonance frequencies ranging from 2 to 100 kHz to determine “bound proton fraction” which is assumed to represent the macromolecular fraction (MMF) of tissues. The tendon in the healthy subjects in this study demonstrated a consistent MMF higher than that of a psoriatic arthritis [22].

In recent years, UTE-MT modeling has been proposed to provide multiple parameters with much higher reproducibility levels, including MMF, macromolecular relaxation time (T2mm), and exchange rates. Ma et al. proposed a multi-spoke 3D-UTE Cones MT sequence and a modified rectangular pulse model to accelerate volumetric quantitative imaging of short T2 tissues like tendon and cortical bone[43]. MMF obtained with the 3D-UTE Cones MT technique is less sensitive to the magic angle effect and more sensitive to rotator cuff tendon degeneration when compared to T2 using Carr-Purcell-Meiboom-Gill (CPMG) sequence[44]. Employing the same approach, a comparison of tendons in healthy individuals (n=7) and patients with psoriatic arthritis(n=9) demonstrated considerably elevated MTR values in the latter group that were not discernible using the conventional GRE-MTR sequence[45]. MT imaging has also been employed to distinguish variations between tendons and enthesis, whereby the latter exhibits lower MTR and MMF and higher Young's modulus and stiffness[42].

MMF acts as a surrogate measure for collagen levels, as shown in an ex-vivo study conducted by Guo et al. (2019). The study involved the treatment of 36 supraspinatus tendons with collagenase, revealing a significant correlation between MMF values and collagen content [49].

Jerban et al. in his study observed lower MMF values for tibialis tendons in older women (n=22) compared to the young cohort (n=26) [48].

Ashir et al. also discovered that MMF was lower in tendons with rotator cuff tendinopathy (n=14) compared with asymptomatic subjects. Figure 5 shows the UTE-based MMF pixel map and MT-modeling curves in the rotator cuff tendon of a representative healthy subject (first row) compared with a representative patient with rotator cuff tendinopathy [57]. Notably, the contralateral asymptomatic tendons in patients also had a lower MMF than controls, which aligns with previous reports indicating a higher risk of tears in the opposite shoulder [57]. Later, the MMF in lower leg tendons (Achilles, tibialis anterior, and tibialis posterior) of individuals with osteoporosis was compared to healthy controls, and it was found that the MMF was lower in the osteoporosis patients than in the healthy control group [48]. In a recent study, Moazamian et al. reported lower MMF values in Achilles tendon and its enthesis of patients with psoriatic arthritis (n=26) compared with asymptomatic young subjects (n=27). [69]

Figure 5:

Generated macromolecular proton fraction (MMF) pixel maps and fitting curves for healthy control (first row) and symptomatic tendinopathy patient (second row) using 3 MT powers (300°,550° and 750°) and five offset frequencies (2,5,10,20,50 kHz). Obviously, the symptomatic patient demonstrated lower MMF value. This figure was previously presented by Ashir et al. [57] Reprinting permission is granted through the Rightslink system.

3.4. UTE T1ρ

T1ρ relaxation occurs after the application of a long-duration on-resonance RF pulse to “spinlock” the longitudinal magnetization vector into a rotated frame. T1ρ is hypothesized to be sensitive to low-frequency motional interactions between macromolecules, like collagen and proteoglycan, and the associated water molecules, which makes it helpful in studying tissues like tendons and cartilage [76]. UTE T1ρ sequences can visualize and quantify the T1ρ of tendons and, more recently have been used to quantify the slow and fast relaxing water components. T1ρ has a strong potential to be an early noninvasive biomarker of proteoglycan loss and early tissue degeneration [77]. T1ρ is thought to be more sensitive than T2/T2* in identifying early tissue degeneration [78,79].

Du et al. first used a 2D UTE-based T1ρ sequence to determine the T1ρ relaxation time and dispersion in the Achilles tendon. The mean T1ρ of the tendon increased by around four times when spin-lock frequency was increased from 250 to 1000Hz [21]. Further experiments by the same group on tendons demonstrated a strong ‘magic angle’ effect on UTE-T1ρ values, which increased from 5.5 ± 2.2 ms at 0° to 40 ± 5 ms at 55° orientation angle relative to the main magnetic field (B0, parallel to the scanner bore [27]. The magic angle sensitivity has been a concern when a continuous wave pulse sequence has been used. Ma et al. developed a sequence that used an adiabatic spin-locking pulse, which showed a much reduced magic angle effect in tendons compared to the previously used continuous wave T1ρ sequence [40,54]. The adiabatic UTE-T1ρ has been recently shown to be significantly correlated with the mechanical properties of Achilles tendon and its enthesis [70]. Figure 6 shows the adiabatic UTE-T1ρ pixel map and fitting curve in a representative cadaveric Achilles tendon imaged in the sagittal plane using UTE-Adiab-T1ρ sequence. However, the in vivo adiabatic UTE-T1ρ application and its sensitivity to tendinopathy is yet to be investigated.

Figure 6.

(a) UTE- Adiabatic-T1ρ pixel map generated for Achilles tendon from single component T1ρ decay fitting performed on each pixel. (b) UTE- Adiabatic-T1ρ measured using four different spin-lock time (TSL=0 – 48 ms) within an exemplary region of interest (indicated in while box in (a)).

Sharafi et al. developed a 3D-T1ρ zero echo time (ZTE) based sequence with pointwise encoding time reduction with radial acquisition (PETRA) for biexponential relaxation mapping of knee tissues including patellar tendon and Achilles tendon. They were able to successfully demonstrate the feasibility of the sequence for bicomponent T1ρ assessment of these tissues [47].

4. Discussion

In this study, we reviewed various studies utilizing UTE sequences for tendon imaging, examining their results, advantages, and drawbacks. This review aimed to provide insights into the potential applications of quantitative UTE sequences in tendon imaging. Although most of the discussed quantitative UTE-MRI techniques are still in the validation phases, they showed significant potential to be translated into clinical investigations for compositional and ultrastructural assessment of tendons.

Table 2 summarizes the principle, acquisition protocol, potential sensitivity to compositional changes in tendons, and advantages and disadvantages of the reviewed UTE techniques. UTE-T2* and UTE-T1 can be considered as the least sophisticated techniques (no pulse preparation, shorter scan time, and simple exponential models) to be translated to clinical investigations. However, these techniques are not sensitive to a specific compositional change of the tendons. Nevertheless, tendon injuries are expected to lead to an increase in T1 and T2* caused by inflammation, higher water content, and fiber disruption. UTE-T1ρ and UTE-MT usually need longer scan time and result in higher specific absorption rate (SAR) caused by the pulse preparation (spin-lock and off-resonance saturation) in these techniques. UTE-T1ρ and UTE-MT are likely sensitive to macromolecular changes in tendons. Tendon injuries are expected to lead to an increase in T1ρ while a decrease in MT measures caused by inflammation, higher water content, and fiber disruption. It should be noted that all the reviewed MRI techniques in this study are highly likely correlated to each other. Therefore, recommending a unique technique for a specific tendon composition or condition is challenging and requires future well-designed experiments and comparisons with ground truth measures from histology and electron microscopy.

Table 2:

UTE techniques principles and potentials

Quantitative UTE MRI	Principle	Acquisition protocol	Scan time	Potentially sensitive to	Advantage/Disadvantage
UTE-T1	Spin lattice relaxation time: differential recovery of the longitudinal magnetization	- Variable flip angle UTE - Inversion recovery preparation with UTE - Variable repetition time UTE	Moderate	Nonspecific (water, collagen, or proteoglycan)	Widely available
UTE-T1ρ	Spin lattice relaxation time in the rotating frame	Spin-lock preparation with varying spin-lock time acquired with UTE	Moderate	Macromolecules like collagen and proteoglycans.	- Relatively high specific absorption rate (SAR) - Low sensitivity to the magic angle effect when adiabatic preparations are used
UTE-T2*	Apparent spin-spin relaxation time: decay of transverse magnetization caused T2 relaxation and magnetic field inhomogeneity	UTE and non-UTE acquisitions with differing TEs	Short	Nonspecific (water, collagen, or proteoglycan)	- Widely available - Susceptible to the magic angle effect.
UTE-MT modeling	Magnetization exchange between free water protons and protons in macromolecules	Off-resonance saturation pulse preparation (different power and frequency offsets) with UTE	High	Macromolecules like collagen and proteoglycans.	-Less sensitive to magic angle effect.

Quantitative UTE biomarkers, like UTE T2* and continuous wave UTE T1ρ, exhibit a notable susceptibility to magic angle effects, especially in tendons, owing to their collagen fibers' dense and parallel alignment. Figure 7 shows the variations in these values depending on the orientation angle of the tendons inside the scanner bore could be significant and surpass the alterations caused by the actual pathology. This figure was previously presented by We et al. [46]. Among the discussed UTE techniques, UTE-Adiabatic T1ρ and MT modeling represent novel and promising techniques for addressing the magic angle effect in tendons. In the light of this fact, more tendon-specific studies are expected to use UTE-Adiab-T1ρ and UTE-MT modeling in future investigations.

Figure 7:

Angular dependency of UTE Cones Adiab T1ρ, CW- T1ρ and Cones-T2*. T1ρ is less sensitive to the magic angle effect than T2*. Among T1ρ sequences, Adiabatic-T1ρ exhibits a decreased magic angle impact compared to the previously employed CW-T1ρ. This figure was previously presented by We et al. [54]. Reprinting permission is granted through the Rightslink system.

In non-Cartesian UTE imaging, fat signals exhibit a radial shift due to off-resonance sampling of fat. Since many UTE sequences employ non-Cartesian trajectories such as radial or cones, it becomes crucial to employ fat suppression techniques to minimize the interference caused by fat signals. This is particularly important in regions such as the Achilles and patellar tendons, which are adjacent to fat pads and susceptible to fat signal contamination. Conventional fat saturation is still the most widely used fat suppression technique in UTE, but the fat sat pulse attenuates the short T2 tendon signals. In their research, Chen et al. investigated the impact of fat saturation on UTE T2*, T1ρ, and MT sequences[80]. They tested fat-saturated and non-fat-saturated modules on whole knee cadaveric specimens at 3T. The study revealed strong correlations and agreement between the two techniques, with only around a 10% difference observed for the relaxation times and MT values (MTR and MMF) of the quadriceps tendon.

A recent study by Ma et al. introduced a new approach utilizing a composite soft-hard RF pulse to effectively suppress fat signals in UTE imaging for tissues with short T2 values[81]. The soft component of this composite pulse is characterized by a narrow bandwidth a small negative flip angle and is centered on the fat peak. On the other hand, the hard component is a short rectangular pulse with a small positive flip angle. By employing this composite pulse, the fat magnetization is tipped down and then back with the same flip angle, reaches a state of equilibrium, leaving only the excited water magnetization. In their feasibility study, Ma et al. compared a conventional fat saturation (FatSat) model with a novel technique using the signal suppression ratio (SSR) as an evaluation metric. The results demonstrated that the soft-hard composite pulse approach outperformed the FatSat method by exhibiting lower signal attenuation in water imaging, effective in vivo fat suppression, improved preservation of both long and short T2 signals, significantly higher SSR for short T2 signals, and enhanced contrast between water and fat. Jang et al. applied fat sat using a novel, single-point Dixon approach, allowing better UTE contrast for tissues with short T2[82]. This could help in much more efficient fat suppression and more accurate values. Additional research to explore the influence of quantitative UTE imaging with various fat suppression techniques would be highly valuable.

In addition to hydrogen proton MRI widely available in hospitals, Sodium MRI has emerged as a novel and intriguing MRI sequence in recent years. Tendinopathy is characterized by elevated levels of water and proteoglycan concentrations, leading to an associated rise in sodium content[83,84]. Sodium MRI has been used to measure and quantify SNR, apparent tissue sodium contrast, and sodium relaxation times in Achilles tendons. As expected, increased SNR has been found in Achilles tendinopathy patients, owing to the increased GAG content [85,86]. A significant drawback, however, of this technique is the low resolution of images, which makes them prone to partial volume effects.

Employing UTE-MRI biomarkers in clinical investigation encounters a few significant challenges. First, quantitative UTE MRI protocols, especially MT for MMF analysis and T2* and T1ρ for bicomponent analyses, require long scan times (>20 minutes), making them less useful for clinical implementation and increasing the risk of motion artefacts. Second, for such long MRI scans, motion registration becomes a crucial component of the protocol for in vivo quantitative imaging to mitigate the impact of patient movement during scans. Various accelerating techniques can be utilized to accelerate the quantitative UTE protocols and significantly reduce motion-related artifacts while maintaining minimal errors. These techniques include stretching the readout trajectory [52], compressed sensing, and integration of deep convolutional neural networks (CNNs)[87]. Third, most quantitative MRI techniques may show site and vendor-dependent variation, which challenges future clinical adoption. Well-designed multi-site and multi-vendor investigations are required to promote such quantitative MRI-based techniques.