Myelin is a lipid-protein bilayer that surrounds the axonal fibers of neurons (1). It plays an essential role in normal brain function by facilitating the rapid conduction of action potentials in the axon (2). Many neurological and degenerative diseases, such as multiple sclerosis (MS), are characterized by myelin damage and loss (3, 4). Thus, evaluations of demyelination and remyelination are essential for the accurate diagnosis and treatment monitoring of these diseases. Conventional magnetic resonance imaging (MRI) techniques, such as T₁- and T₂-weighted fast spin echo (T₁w- and T₂w-FSE), provide high soft-tissue contrast and are routinely used in the diagnosis of MS and many other neurological diseases. However, these clinical sequences are unable to distinguish demyelination lesions, such as those observed in MS, from the ones caused by inflammation, edema, axonal loss, or gliosis (5–7). Moreover, it is also difficult for clinical MRI to detect myelin damage in brains that appear to be normal (i.e., where there are no apparent lesions) (8). Consequently, developing myelin-specific imaging techniques is of critical importance to improved evaluation of neurological diseases in clinical practice.

In the last two decades, many myelin water imaging (MWI) techniques have been developed for the specific evaluation of myelin changes. Myelin water (MW) is the water component tightly bound or trapped in the myelin bilayer (9–14). It has much shorter T₁ and T₂ relaxation times than intracellular/extracellular water. Myelin water fraction (MWF), defined as the signal ratio of MW to total water, is the biomarker used to quantify MW content changes (15–18).

State-of-the-art MWI techniques like multi-echo spin echo T₂ relaxometry (15), multicompartment analysis of T₂* decay (16), and multicomponent-driven equilibrium single-pulse observation of T₁ and T₂ (mcDESPOT) (17) have been developed to quantify MWF and have been applied for the assessment of MS. The high correlation between the MRI-measured MWF and histologically quantified myelin content has been demonstrated in brain sample studies (19, 20). Moreover, all these techniques are able to detect significant MWF decreases in demyelinated lesions in patients with MS in comparison to the normal white matter (NMW) in healthy individuals (15–17, 21, 22). Despite the success of these myelin-specific techniques, studies have found that these multicompartment modeling techniques are sensitive to system flaws such as B₁ and B₀ inhomogeneities (23–27). Different data post-processing strategies may also produce different results (28–31). Consequently, these techniques suffer from limited accuracy and robustness in terms of MWF estimation, shortcomings that slow down their clinical translation.

Another promising technique that has been developed for selective imaging of MW, known as direct visualization of short transverse relaxation time component (ViSTa), is based on the T₁ difference between MW and long T₂ intracellular/extracellular water components (32). The T₁ relaxation times for the long T₂ components in white matter and gray matter range from 750 to 1,000 ms and from 1,300 to 1,800 ms, respectively, at 3T (32–34). In contrast, the T₁ relaxation times for MW components are typically shorter than 600 ms (17, 18, 35, 36). This technique employs a double inversion recovery (DIR) preparation to robustly suppress all the long T₁ water components in the brain. Moreover, not only is ViSTa less sensitive to B₁ and B₀ inhomogeneities, but it does not require any complicated modeling to quantify MWF. That being said, ViSTa's scan time is too long for clinical use for whole brain coverage (∼3 min per slice).

Most recently, Ma et al. have developed a short TR adiabatic inversion recovery (STAIR) technique in combination with a 3D Cones acquisition (STAIR-Cones) for time-efficient selective MWI in the whole brain (37). This technique uses a short TR to suppress long T₁ water components with a variety of T₁s. Similar to ViSTa, the STAIR sequence is relatively insensitive to B₁ and B₀ inhomogeneities because it uses an adiabatic full passage (AFP) pulse for signal inversion; however, because the 3D Cones sequence is not a clinically available sequence (it is primarily available in GE research scanners), the STAIR-Cones sequence is currently limited in its potential for clinical translation.

In this study, we proposed a new clinically translational MWI sequence for whole-brain MWI and quantification, which is a combination of the STAIR technique and echo-planar imaging (EPI) readout. Unlike the research-dedicated Cones sequence, the EPI sequence is a routinely used clinical sequence that is available to all vendors, posing the proposed STAIR-EPI as a technique with greater potential in clinical translation than STAIR-Cones. Moreover, similar to the ViSTa and STAIR-Cones techniques, the proposed STAIR-EPI does not require complicated post-processing such as solving the ill-conditioned problems for those multicompartment modeling techniques (15–17). We investigated the feasibility of the proposed STAIR-EPI technique for the quantification of MWF and compared its values in the lesions and normal-appearing white matter (NAWM) of seven MS patients against the values of NWM of seven healthy volunteers on a clinical 3 T MRI scanner.

Figure 3 shows the representative STAIR-EPI and PD-EPI images as well as corresponding aMWF maps from a 31-year-old male healthy volunteer. Much higher MW signal intensities in the STAIR-EPI images are found in white matter regions than those in grey matter regions. The aMWF maps also demonstrate a higher aMWF in the white matter region than in the grey matter region.

Figure 4 shows representative T₂-FLAIR, PD-EPI, and STAIR-EPI images as well as aMWF maps from three MS patients. The hyperintense lesions in T₂-FLAIR show low signal intensities in the STAIR-EPI images and also have lower aMWF values than NAWM regions, demonstrating a decrease in MW content for these MS lesions.

Table 1 summarizes the mean and standard deviation (SD) values of the aMWF measurements from all participants. The aMWF value of NWM in healthy volunteers is 10 ± 1.3%, while the aMWF values of NAWM and MS lesions in MS patients are 8.6 ± 1.2% and 3.6 ± 1.3%, respectively. Significant difference is achieved between these three groups (p < 0.001).

Figure 5 shows the paired comparison of measured aMWF values on each of the two groups (i.e., NWM vs. NAWM, NWM vs.MS lesions, and NAWM vs. MS lesions). The results indicate a notable distinction of aMWF measurements between MS lesions against both NAWM and NWM (p < 0.001). A significant difference in aMWF measurement is also observed between NAWM and NWM (p < 0.001). These results demonstrate the feasibility of the STAIR-EPI technique in the detection of demyelination in MS.

In this study, we developed a new translational STAIR-EPI sequence for selective MWI and aMWF quantification in the whole brain. In MS patients, the hyperintense MS lesions in T₂-FLAIR showed low signal intensities in the STAIR-EPI images, indicating high sensitivity in the detection of demyelination for the STAIR-EPI technique. Moreover, the measured aMWF values of MS lesions (3.6 ± 1.3%) and NAWM (8.6 ± 1.2%) in MS patients were significantly lower than those of NWM (10 ± 1.3%) in healthy volunteers. This study reveals the effectiveness of the STAIR-EPI technique in detecting myelin loss in both MS lesions and NAWM in MS patients, a capability that could subsequently be applied in clinical MRI scanners from all vendors. The implementation of the proposed STAIR-EPI sequence is relatively easy and requires neither specialized involvement from scanner vendors nor from application specialists.

EPI is one of the fastest MRI sequences and has already made significant contributions to clinical diagnosis and scientific investigation for various parts of the body including the brain, abdomen, and pelvis (41). With single-shot EPI, a complete set of spatial-encoding data can be gathered following a single RF excitation. However, single-shot EPI suffers from strong imaging distortion induced by eddy currents and B₀ inhomogeneity due to the low bandwidth in the phase-encoding direction. Multi-shot EPI strategy mitigates the imaging distortion because of its increased bandwidth in the phase-encoding direction. In this study, the STAIR technique was combined with the multi-shot EPI acquisition scheme to achieve relatively high-quality MWI. The image quality improved with more shots, but more shots also led to increased scan time (42). We found that eight shots provided a reasonable level of image quality within an acceptable 15-minute scan time.

One of the major advantages of the STAIR-EPI technique is its simplicity, given that it requires neither complex sequence implementation nor complicated post-processing in comparison to conventional techniques (24, 28–30, 37). Additionally, the STAIR preparation is relatively insensitive to B₀ and B₁ field inhomogeneities because it uses the AFP pulse for magnetization inversion (43). These benefits make the STAIR-EPI technique well-suited for clinical practice. Both DIR and STAIR are effective techniques for selective imaging of short T₁ MW and signal suppression of long T₁ intracellular and extracellular water in the brain (32, 37). However, the STAIR technique has an improved scan efficiency over the DIR technique because of its much shorter TR.

Previous research on the MWF quantification in NWM has presented a range of values using different methodologies (15–18, 23, 26, 44–49). For example, multicomponent T₂ decay analysis estimated MWFs ranging from 9% to 15.6% (15, 44–48), whereas multicomponent T₂* decay analysis yielded an MWF range of 6.9% to 14.4% (9, 16, 23, 26, 49). Labadie et al. employed multicomponent T₁ modeling and determined an MWF of 8.3% (18). Ma et al., employing the 3D STAIR-Cones technique, found an MWF value of 9.2% (37). Moreover, a range of MWF values has also been reported for MS lesions in previous MWI studies (15, 37, 48, 49). The multicomponent T₂ analysis indicated MWFs ranging from 1.7% to 6.4% (15, 48), while a multicomponent T₂* analysis determined an MWF of ∼0% (49). For the most recent STAIR-Cones study, Ma et al. reported an MWF value of 4.5% for MS lesions (37). In this study, the mean MWF values of NWM (i.e., 10 ± 1.3%) and MS lesions (i.e., 3.6 ± 1.3%) are consistent with those reported in previous research.

As reported in previous studies, the multicomponent T₂ relaxometry sequence took between 25 and 38 min to complete (45, 46), whereas the multicomponent T₂* relaxometry sequence took between 20 and 30 min (16, 26). In comparison, the total scan time for the proposed STAIR-EPI technique is around 16 min, which is relatively shorter than the typical multicomponent T₂ and T₂* relaxometry techniques.

There were some limitations in this study. First, only seven MS patients were scanned in this technical feasibility study. We plan to recruit more patients to investigate demyelination or remyelination in the future. Second, as proof of concept, a relatively large NEX (i.e., 30) was used in the STAIR-EPI scan to achieve a high SNR performance, significantly prolonging the scan time for whole brain coverage. The scan time could be reduced by using a lower NEX value (e.g., 10 or less). Moreover, the recent development of the denoising technique via deep learning could significantly increase the image SNR, thereby facilitating sufficient SNR improvement with a much reduced NEX (e.g., 5) (50, 51).

The STAIR-EPI technique detects demyelination in MS, facilitating easy clinical translation for the whole brain MWI.