One basic approach to enhancing contrast in UTE imaging is subtracting two images acquired at distinct echo times (TEs). In the dual-echo UTE imaging technique with echo subtraction, bone contrast is acquired by subtracting a second echo image from a first echo image which is equivalent to T2 bandpass filtering (19). Signals from long T2 tissues experience minimal decay by the time of the second echo, while the signal from bone undergoes significant decay by the time of the second echo. As a result, long T2 tissues show a high signal in the second echo, while bone shows a signal void. Subtraction of the second echo image from the first echo image leads to suppression of long T2 signals, leaving bone signal minimally unaffected, creating high contrast for cortical bone. Rescaled subtraction (46), where the first UTE free induction decay (FID) image is scaled down prior to subtraction to lower signal from long-T2 tissues in the first compared to the second echo, works more efficiently in creating high positive contrast for short-T2 species, especially cortical bone, which has a much lower mobile proton density than surrounding muscle or fat. Figure 1 shows an example of 3D dual-echo UTE imaging with rescaled subtraction applied to the tibia of a healthy volunteer. Conventional 3D UTE imaging provides a relatively high signal but negative contrast for the tibia ( Figure 1A ). Regular echo subtraction presents a positive contrast between bone and muscle, but a negative contrast between cortical bone and fat, as fat also has a short T2* ( Figure 1C ). The contrast between bone and fat/muscle increases using the rescaled subtraction technique ( Figures 1D–F ). However, subtraction techniques are sensitive to patient motion, which can cause misalignment between the source images and result in artifacts.

Preparation pulses can be employed to selectively suppress signals from long T2 components, improving contrast by allowing better visualization of short T2 tissues (47–49). In UTE imaging with long T2 saturation, saturation pulses are used to suppress the signals from long T2 tissues, such as muscle and bone marrow fat, which typically produce higher signals than bone. For example, a 90° pulse with a relatively long duration and a low amplitude can flip the longitudinal magnetization of long T2 tissues into the transverse plane, where a large spoiling gradient can subsequently dephase the transverse magnetization (48). In comparison, bone magnetization is barely excited by this long saturation pulse as the decay rate of bone exceeds the excitation rate. Therefore, a long 90° pulse can be used with a large spoiling gradient to suppress long T2 tissues, leaving bone to be subsequently detected by UTE data acquisition. T2 selective RF excitation (TELEX) can be used to increase bone contrast (47). Dual-band long-T2 suppression pulses further improve the suppression of signals from muscle and fat (49). However, residual signals from muscle and marrow fat due to B1 and B0 inhomogeneities may still compromise bone contrast.

Off-resonance saturation with subtraction can generate contrast for short T2 components by utilizing the broader absorption line shape of short T2 tissues, such as bone, compared to long T2 tissues like muscle or fat, making them more sensitive to off-resonance RF radiation (50). UTE imaging with off-resonance saturation contrast (UTE-OSC) employs a high-power saturation pulse placed a few kHz off the water peak to preferentially saturate signals from bone, leaving long T2 muscle and fat signals largely unaffected (50). Subtraction of UTE images with off-resonance saturation from basic UTE images can effectively suppress signals from muscle and fat, creating high bone contrast.

One limitation of saturation techniques that utilize hard RF pulses is their sensitivity to B0 and B1 inhomogeneities, making them less robust compared to adiabatic inversion (18). The adiabatic inversion recovery UTE (IR-UTE) contrast mechanism employs a long adiabatic inversion pulse to invert the longitudinal magnetizations of long-T2 water (e.g., muscle) and long-T2 fat (18, 21, 24, 51). The duration of the adiabatic inversion pulse is much longer than bone T2* (18). As a result, the longitudinal magnetizations of muscle and marrow fat are fully inverted, while the bone magnetization is not inverted but largely saturated by the long adiabatic inversion pulse (51). The UTE data acquisition starts at an inversion time (TI) adjusted so that the inverted long T2 magnetizations approach the null points, leaving the uninverted bone magnetization being selectively detected by UTE data acquisition. The adiabatic inversion pulse has a relatively broad spectral bandwidth, thereby insensitive to B1 and B0 inhomogeneities. The IR-UTE technique allows uniform inversion of long T2 magnetizations, providing robust high contrast imaging of bone (18, 21). Figure 2 shows representative IR-UTE images of cortical bone in the forearm, which is depicted with excellent image contrast but invisible with conventional clinical FSE sequences.

A single inversion pulse can reduce the signal from fat and long T2* components (such as muscle) by up to 80% (21). However, using dual inversion pulses allows for the complete nulling of both, providing more effective signal suppression (52–54). The double adiabatic inversion recovery UTE sequence (double-IR-UTE) employs two identical adiabatic inversion pulses (duration of ~6 ms) with the same center frequency to sequentially invert the longitudinal magnetizations of long T2 species, followed by multispoke UTE data acquisition (55). The two adiabatic inversion pulses are applied with pre-defined inversion times TI1, which is the time between the centers of the two adiabatic inversion pulses, and TI2, which is the time from the center of the second adiabatic inversion pulse to the center spoke of the multispoke acquisition. Robust long T2 suppression can be achieved by timing the center spoke at the null point. Long T2 transverse magnetizations acquired before the null point are of opposite polarity to those acquired after the nulling point, leading to cancellation in the regridding process during image reconstruction and, therefore, efficient suppression of long T2 signals from muscle and marrow fat. Bone magnetization is not inverted but saturated by the two long adiabatic inversion pulses, recovers after the second TI2, and is subsequently detected by UTE data acquisition. The advantage of double-IR-UTE is the robust suppression of long T2 tissues with a broad range of T1s, such as fat and muscle, which can be nulled simultaneously using specific combinations of TI1 and TI2. The double-IR-UTE sequence is insensitive to inhomogeneities in the B1 and B0 fields due to the use of adiabatic inversion pulses with relatively broad spectral bandwidths. Figure 3 shows double-IR-UTE imaging of the knee joint in a healthy volunteer, which shows high signal from short- and ultrashort-T2 species, such as the patellar tendon and cortical bone.

UTE data acquisition can be combined with relaxation-parameter contrast (56). UTE with relaxation parameter contrast and subtraction exploits the sensitivity of bone proton magnetization to both T2 and RF pulse duration. Excitation pulse parameters are selected to determine the extent of concurrent relaxation and excitation. The RF pulse duration and amplitude can be changed to adjust the relaxation dependence of bone contrast. To selectively detect signals from magnetization within a specific range of T2 values, two RF pulse durations are chosen so that the sensitivity transition between them brackets the range of interest. Two UTE datasets with similar imaging parameters but different RF excitation pulses are acquired. Bone contrast is created by subtraction of the two UTE images, as shown in Figure 4 .

The 3D DURANDE UTE sequence and bone-selective image reconstruction have been proposed for rapid bone imaging (57). This technique acquires two dual-echo UTE datasets following short and long RF pulses, with encoding gradients varying continuously along the entire pulse train to halve the total imaging time. The DURANDE UTE sequence employs two rectangular RF pulses (RF1 and RF2), differing in duration and amplitude but having the same pulse area applied alternately in successive TR periods along the entire pulse train. Two echoes at a short TE and a long TE are collected from the beginning of the gradient ramp-up within each TR. As a result, four echoes are produced and combined via a view-sharing approach to generate two independent k-space datasets during image reconstruction. Accelerated UTE bone imaging can be achieved by using the sparsity of bone voxels in the corresponding subtraction images.

In STAIR-UTE, 3D IR-UTE data are acquired with a short TR and a high flip angle within specific absorption rate (SAR) limits for clinical imaging (58–60). The short TR and TI combination is selected to achieve robust suppression of long-T2 muscle and marrow fat regardless of their different T1 values. Multiple spokes are acquired for efficient volumetric imaging of cortical and trabecular bone (60). The STAIR-UTE sequence is more efficient than other UTE or ZTE techniques, such as the spectral presaturation with IR UTE (SPIR-UTE), in selective imaging of trabecular bone (61). Figure 5 shows STAIR-UTE images of the spine and SPIR-UTE images of the fingers. The SPIR-UTE images showed T₂* values of 2.42 ± 0.56 for the capitate, which is much longer than the T₂* of 0.31 ± 0.01 ms for the trabecular bone of the spine measured on STAIR-UTE images (60, 61), or the T₂* value of ~0.3 ms for the cortical bone (62). The much longer T₂* values suggest that SPIR-UTE imaging of the trabecular bone is subject to significant long-T2 signal contamination. In comparison, STAIR-UTE-measured T₂* values for the trabecular bone are close to those measured for cortical bone, suggesting that bone marrow fat is completely suppressed and only signal from trabeculae is selectively detected in STAIR-UTE imaging (60).

Past research has focused on high resolution imaging of marrow to indirectly detect trabecular microstructure (5, 63, 64). Two major challenges exist: the high susceptibility at the marrow/bone interface and the multiple fat peaks, both of which significantly reduce T₂*, leading to low marrow signal (misclassified as bone) and overestimation of trabecular volume. UTE is insensitive to T₂* shortening. However, UTE employs non-Cartesian radial sampling, which is sensitive to chemical shift artifacts (65). UTE imaging on the fat peak resolves this issue (66). Bone is off-resonance in fat-centered imaging, but it has a much lower signal than marrow, and the off-resonance artifact is negligible. Figure 6 shows UTE and clinical GRE imaging of a trabecular bone sample from a 65-year-old male donor. UTE on the water peak shows strong chemical shift artifacts, which are significantly reduced in UTE imaging on the fat peak. Trabecular bone thickness is overestimated at longer TEs (e.g., TE = 1.1, 2.2, 3.3, or 4.4 ms) or with the clinical GRE sequence. UTE imaging on the fat peak is expected to perform even better in older osteoporotic or diabetic patients who typically have a higher fat fraction in the marrow.

ZTE employs a short rectangular pulse excitation followed by readout gradient flat-top sampling to minimize the effective TE (29). A small flip angle (1-2°) is typically used to minimize the dead-time gap, which causes a spherical void in the center of k-space. A variety of approaches have been developed to address this k-space gap and associated low frequency artifacts in the reconstructed images (33). The repetition time (TR) is minimized to speed up data acquisition. Higher receiver bandwidths (62.5-83.3 kHz) are recommended to mitigate chemical shift artifacts. Bias field correction, contrast inversion, and background segmentation are employed for CT-like bone contrast (34–36). The principal difference between ZTE and UTE sequences is the temporal order of setting the spatial encoding gradient and RF excitation (33). UTE offers the freedom to adjust TE, a feature not possible in ZTE imaging. UTE also allows high flip angles, a significant advantage in direct bone imaging using the STAIR contrast mechanism (60). On the other hand, the ZTE sequence acquires k-space data after the readout gradients are fully ramped up, avoiding fidelity issues introduced by gradient ramping (33). ZTE has a shorter effective TE and can detect signal from shorter T2 species. ZTE can be applied in many of the same applications and with many of the same magnetization preparation methods as UTE.

Many other UTE-type sequences have been developed for bone imaging. These sequences can be combined with each of the above contrast mechanisms for high-contrast imaging of bone. For example, adiabatic inversion recovery-based preparations can be combined with ZTE (29–36), vTE (40), AWSOS (43), RHE (44), and PETRA (37–39) sequences for high contrast imaging of cortical bone and other short-T₂ tissues, respectively. On-resonance long-T₂ suppression or off-resonance short-T₂ saturation can be applied to SWIFT, PETRA, WASPI, RHE, and ZTE sequences to create short-T₂ contrast. For example, SWIFT with off-resonance saturation has been used to image the interface between cartilage and subchondral bone (67). A systematic study of the above contrast mechanisms combined with ZTE, vTE, WASPI, SWIFT, AWSOS, PETRA, RHE, and Looping Star sequences remains to be investigated, and their SNR and CNR efficiency remains to be compared.

T1 relaxation is a fundamental MR tissue property and describes how fast the longitudinal magnetization recovers to the steady state. Many T1 measurement techniques have been combined with UTE acquisitions to provide accurate T1 measurements of bone, such as saturation recovery UTE (18), inversion recovery UTE (68), UTE with variable repetition time (UTE-VTR) (69), and UTE with variable flip angle (UTE-VFA) methods (70). The UTE-VTR method is sensitive to B1 inhomogeneity. The actual flip angle imaging (AFI) method has been widely used for 3D B1 mapping (71). By combining UTE and AFI techniques, it is possible to use a pair of interleaved UTE acquisitions with a short TR (e.g., 20 ms) and a longer TR (e.g., 100 ms) to produce accurate B1 mapping for bone (69). Furthermore, combining UTE-VTR and UTE-AFI (UTE-AFI-VTR) provides accurate T1 mapping for bone with B1 correction. A short T1 of ~250 ms was reported for cortical bone (69).

Bone water exists as pore water residing in the macroscopic pores and as loosely bound water attached to the organic matrix (72). UTE sequences can detect pore water with a longer T₂* of ~3 ms and loosely bound water with an ultrashort T₂* of ~0.3 ms (62, 73–76). IR-UTE or STAIR-UTE allows partial inversion and nulling of pore water with longer T2*, leaving bound water with ultrashort T2* to be selectively imaged (21, 60, 77). Figure 7 shows single- and bi-component fitting of UTE and IR-UTE images of a bovine cortical bone sample (77). Excellent bi-component fitting was achieved to show the existence of two distinct water components: bound water with a short T₂* of 0.26 ms (72.4% by volume) and pore water with a longer T₂* of 1.56 ms (27.6%). The IR-UTE images show a single component with T₂* ~0.31 ms, suggesting that pore water is efficiently suppressed and bound water selectively imaged (77).

There is another group of protons, collagen backbone protons, which have extremely short T2* relaxation times and are invisible with UTE sequences. UTE magnetization transfer (UTE-MT) modeling can measure collagen backbone proton fraction and exchange rates between water and collagen protons (73–75, 78–83). Figure 8 shows UTE-MT imaging of a bovine bone sample. Excellent two-pool MT modeling and MT parameters mapping were achieved using a Gaussian lineshape (79). The lower half of this bone sample shows increased variations in UTE image signal intensity and MT parameters, suggesting an abnormality that needs further investigation.

UTE sequences can be used to map TWPD, BWPD, PWPD, and MMPD (20–23, 25, 76, 84–91). TWPD can be estimated by comparing the UTE MRI signal of bone with an external reference with known proton density (20, 21, 76, 84–88). BWPD can be measured with IR-UTE or STAIR-UTE, which efficiently suppresses pore water (21, 23, 60). PWPD can be quantified by subtracting bound water from total water. MMPD can be quantified by combining total water proton density with macromolecular fraction (MMF) (91). Figure 9 shows 3D mapping of TWPD, BWPD, PWPD, and MMPD for tibial midshaft of a 35-year-old healthy female, a 76-year-old female with osteopenia, and a 57-year-old female with OP, respectively (91). The OP patient has higher PWPD but lower MMF and MMPD, consistent with increased porosity and loss of mineral/collagen.

Susceptibility is an important material property. QSM techniques can estimate calcium and iron accumulation in the brain (92). Bone susceptibility is more challenging to measure due to the lack of signal. UTE can detect phase evolution in cortical and trabecular bone. The phase changes with increasing TEs can be used to evaluate bone susceptibility using various algorithms such as Morphology Enabled Dipole Inversion (MEDI) (93). UTE with QSM (UTE-QSM) provides information about bone susceptibility, which is indirectly related to bone mineral (93–98). Figure 10 shows UTE-QSM and µCT-measured volumetric BMD (vBMD) of a human bone sample, with an excellent linear correlation between QSM and vBMD (n=9). UTE-QSM can reliably evaluate vBMD in cortical bone. Similar results are also observed for trabecular bone.

There is a close association between bone perfusion and bone remodeling and fracture repair (99–102). Increased cortical bone turnover and inflammation are also associated with increased blood flow (99). There is a strong correlation between bone perfusion and BMD (101, 102). However, the nature of bone makes it difficult to investigate perfusion. The techniques applicable to many soft tissues are difficult or impossible to apply to bone. For example, dynamic contrast-enhanced MRI (DCE-MRI) can be used to study perfusion in various tissues and organs. The technique employs fast T1-weighted images to capture signal changes induced by exogenous intravascular nondiffusible gadolinium-based contrast agents as a function of time. Conventional DCE-MRI can study perfusion in the marrow of trabecular bone (103), but cannot study perfusion in cortical bone due to the lack of detectable signal. Dynamic UTE imaging has been developed to evaluate perfusion in cortical bone (104, 105). A recent study reported dynamic 2D UTE imaging of the tibial midshaft of a 38-year-old healthy volunteer and found ~20% signal enhancement after intravenous gadolinium contrast injection (105). Kinetic analysis demonstrated a K^tran of 0.23 ± 0.09 min^-1 and K_ep of 0.58 ± 0.11 min^-1 for the tibial midshaft of this volunteer. DCE-UTE can potentially be used to evaluate bone remodeling and fracture recovery.

Bone components (water, collagen, mineral) and microstructure (cortical porosity, trabecular structure) can be qualified by many other UTE-type sequences such as ZTE (25–36), PETRA (37–39), vTE (40), WASPI (41), and SWIFT (42). For example, WASPI has been used to image bone water and the solid matrix of bone (106). SWIFT has been shown to be able to identify the presence and extent of dental caries and fine structures of the teeth, including cracks and accessory canals (107). Furthermore, solid-state 31P MRI can be achieved with UTE-type sequences by focusing on the 31P peak (108). 31P UTE MRI can map phosphorus content, assess bone mineral density, and differentiate between mature and newly remodeled bone (108, 109).

UTE MRI can be used to measure pore water concentration in cortical bone (23, 76, 84–88). A recent study showed a high correlation (R² = 0.72; P < 0.0001) between μCT porosity and pore water concentration in 32 cadaveric human cortical bone samples ( Figure 11 ) (77). Water residing in the microscopic pores of cortical bone is expected to behave more like “free” water with much longer T2* relaxation time than water bound to the organic matrix. Therefore, separating pore water from bound water is easy, allowing accurate pore water mapping without requiring ultrahigh spatial resolution to resolve the small pores. This is confirmed by the high correlation with an R² of 0.72 between μCT porosity and pore water concentration in cortical bone. μCT porosity is consistently lower than pore water content assessed by UTE MRI. Pore water content in cortical bone is also significantly correlated with its mechanical properties (110–112). In another study, UTE MRI, μCT, and histomorphometry were performed on tibial samples from 11 donors. UTE-measured pore water content showed significant correlations (R²>0.25) with histomorphometry-based lacunae and small Haversian canals, which are below the detectable range of μCT at 9 μm. The μCT-based porosity showed strong correlations with histomorphometric porosity and pore size when considering all pores or only large pores (R>0.70, P<0.01). Correlations were poor when considering only small pores in histomorphometric analyses (R<0.3) (88). Therefore, pore water in smaller pores can be detected by UTE MRI but not by μCT imaging.

UTE MRI can map bound water in cortical and trabecular bone (21, 23, 60). Bound water is a surrogate of bone organic matrix density and negatively correlates with bone mineral density, as shown in Figure 12 (113). It is also reported that bound water in human cortical bone decreases with age, although osteonal remodeling throughout life with only modest changes in tissue mineral density or ash fraction with age after skeletal maturation (114). Bound water and bone density are directly correlated with human cortical bone’s material strength (72, 84, 111, 112).

UTE-MT can indirectly assess collagen backbone protons, providing information about cortical porosity and mechanical properties (73, 81–83). A recent study reported a moderate to strong negative correlation between UTE magnetization transfer ratio (MTR) and μCT porosity (R² = 0.46–0.51), while a moderate positive correlation was observed between MTR and yield stress (R² = 0.25–0.30) and failure stress (R² = 0.31–0.35).A weak positive correlation (R² = 0.09–0.12) between MTR and Young’s modulus at all off-resonance saturation frequencies was also observed (115). UTE-MT measured MTR provides quantitative information on cortical bone and is sensitive to μCT porosity and biomechanical function. MMF derived from UTE-MT imaging can assess mechanical failures after bone stress injury, which is difficult to evaluate using other techniques (83). In another study (73), fibular samples (n=14) were subject to cyclic loading using a 4-point bending setup ( Figure 13 ). Loading was applied to reduce bone stiffness by 20%. Then, bone samples were imaged with UTE MRI and μCT before and after loading. MMF from two-pool UTE-MT modeling decreased by 12% on average, while μCT porosity measured at 6 μm voxel size showed no significant change. A representative sample is shown in Figure 13 , with averaged MMF decreasing from 63% to 55% (p=0.0001), but no detectable changes in μCT porosity (73).

In recent years, many studies have shown that UTE MRI can provide markers of cortical bone porosity, morphologic structure, mineralization, and osteoid density, which are useful measures of bone health (20–23, 76, 84–91, 116–120). In a recent study, Jones et al. reported UTE MRI of 15 participants with OP and 19 without OP (117). The OP group showed elevated pore water (11.6 mol/L vs. 9.5 mol/L; P = 0.007) and total water densities (21.2 mol/L vs. 19.7 mol/L; P = 0.03), and lower cortical bone thickness (4.8 mm vs. 5.6 mm; P < 0.001) and ³¹P density (6.4 mol/L vs. 7.5 mol/L; P = 0.01) than the non-OP group, respectively. Meanwhile, there was no evidence of a difference in bone water (BW) or ³¹P-to-BW concentration ratio. Furthermore, pore and total water densities were inversely associated with DXA and HR-pQCT measured BMD (P < 0.001) (117). In another study, Jerban et al. investigated the differences in water and collagen contents in tibial cortical bone between female osteopenia (OPe) patients, osteoporosis (OPo) patients, and young participants (Young) using a clinical 3T scanner (91). They found MMF, BWPD, and MMPD were significantly lower in OPo patients than in the young group, whereas T1, TWPD, and PWPD were significantly higher in OPo patients. The largest OPo/Young average percentage differences were found in MMF (41.9%), PWPD (103.5%), and MMPD (64.0%), with PWPD significantly higher (50.7%), while BWPD significantly lower (16.4%) in OPe than the Young group on average. Meanwhile, MMF was significantly lower (27%) in OPo patients compared with OPe group (91). As a result, UTE-MRI measured TWPD, PWPD, and MMF were recommended to evaluate individuals with OPe and OPo. Manhard et al. also demonstrated the feasibility of quantitatively mapping bound and pore water in vivo in human cortical bone with practical human MR imaging constraints (84). Jacobson et al. reported a comprehensive set of UTE MRI biomarkers to assess cortical bone. They found the UTE MRI-derived porosity index and signal-intensity-based estimated BMD correlated with the HR-pQCT variables (porosity: r = 0.73, p = 0.006; BMD: r = 0.79, p = 0.002) (120).

UTE MRI has also been used to assess fracture risk. In a recent study, Nyman et al. quantified bound water concentration (C_bw) and pore water concentration (C_pw) in the radius and tibia as predictors of bone fragility (121). Maps of C_bw and C_pw were acquired from the uninjured distal third radius of 20 patients who experienced a fragility fracture of the distal radius (Fx) and 20 healthy controls (Non-Fx), and from the tibia mid-diaphysis of 30 women with clinical OP (low T-scores) and 15 women without OP (normal T-scores). They found C_bw was significantly lower (p = 0.0018) and C_pw was higher (p = 0.0022) in the Fx group than in the Non-Fx group. The area-under-the-receiver operator characteristics curve (AUC with 95% confidence intervals) was 0.73 (0.56, 0.86) for hip BMD (best predictors without MRI) and 0.86 (0.70, 0.95) for the combination of C_bw and C_pw (best predictors overall), as shown in Figure 14 . Meanwhile, C_bw was significantly lower (p = 0.0005) in women with OP (23.8 ± 4.3 ¹H mol/L) than in women without OP (29.9 ± 6.4 ¹H mol/L). They also found that it was C_bw, not C_pw, which was sensitive to bone-forming osteoporosis medications over 12 months. Their results are largely consistent with the study by Gallant et al. (122), who found the hydroxyl groups on raloxifene provided a possible explanation for the therapeutic effect of raloxifene, a Food and Drug Administration (FDA)-approved agent that is designed to treat bone loss, decrease fracture risk, and improve bone mechanical properties. The benefits of raloxifene treatment are essentially independent of bone mass changes and are mediated by an increase in matrix-bound water as measured by UTE MRI. The study suggests a cell-independent mechanism that can be utilized for novel pharmacological approaches to enhancing bone strength (122).

Bone is highly vascularized. Perfusion plays an important role in the growth and development of bone as well as in disease and healing (99–102). Reduced perfusion is observed in the trabecular bone of patients with OP (98). It is believed that decreased osseous vascularity contributes to increased fracture risk (123). Reduced perfusion occurs in synchrony with reduced BMD in vertebral trabecular bone (124). UTE can be used to evaluate bone perfusion (104, 105). There is an extensive enhancement in blood vessels due to fracture of the tibial plateau two days after injury, with specific enhancement of the periosteum distinguished from that of blood vessels, as shown in Figure 15 (104). Even without contrast enhancement, UTE can detect callus formation from a 22-year-old male with a fractured tibia examined 3 weeks after injury (125).