Comparison of Spin-Echo T1- and T2-Weighted and Gradient-Echo T1-Weighted Images at 3T in Evaluating Very Preterm Neonates at Term-Equivalent Age

Discussion

Myelination, a dynamic process in the developing brain, may be the most important indicator of brain maturation.⁵ This fact prompted us to determine which anatomic structures had high rates of myelination at 3T. On the basis of both SE-T2WI and GRE-T1WI, we found these sites to be the DSCP, ICP, and the lateral lemniscus, each of which had myelination in 90%–100% of term-equivalent patients, according to both observers. Other sites, such as the corpus callosum splenium, DSCP, medial lemniscus, pyramidal decussation, PLIC, superior cerebellar peduncle, and spinal tract/nucleus of cranial nerve V also exhibited nearly as high rates of myelination but were favored by 1 sequence (either SE-T2WI or GRE-T1WI). We also found that SE-T2WI showed slightly higher overall rates of myelination and that GRE-T1WI clearly outperformed SE-T1WI in myelination evaluation at 3T.

Assessment of normal myelination is one of the most important questions to be answered by neonatal imaging, for which previous studies at lower field strengths have used SE-T1WI to assess the degree of myelination.^4,5,9,10 Because the myelination patterns at 3T and on GRE-T1WI are largely unknown, we assessed myelination in term-equivalent infants to not only determine their myelination patterns but to also compare 3 commonly used sequences at 3T in a head-to-head manner. To our knowledge, there are no previous studies comparing the myelination patterns on GRE-T1WI with either SE-T1WI or SE-T2WI in neonates or premature infants. On the basis of our results, GRE-T1WI was superior to SE-T1WI both in detecting rates of “myelination” of each structure and regarding interobserver agreement. We acknowledge that it has not been proved that such structures are hyperintense as a result of successful myelination.

Our results were similar in some aspects to those in a previous study of 12 term neonates performed at a lower field strength, in which structures such as the DSCP, ICP, lateral lemniscus, medial lemniscus, pyramidal decussation, PLIC, and the superior cerebellar peduncle also demonstrated higher rates of myelination on GRE-T1WI than in our study.⁴ However, our study may contradict prior data that T1-weighted precedes T2-weighted images in detecting myelination; in particular, our study found a higher rate of myelination within the optic tracts, perirolandic cortex, and spinal tract/nucleus of cranial nerve V on SE-T2WI compared with GRE-T1WI at 3T. In a previous study of very preterm infants imaged at term-equivalent age at 1T, the authors compared 3 different sequences (SE-T1WI, inversion recovery, and SE-T2WI) and found a higher percentages of myelination on SE-T1WI relative to SE-T2WI in most white matter structures; the large majority of patients exhibited myelination in the DSCP, ICP, lateral lemniscus, medial lemniscus, PLIC, and superior cerebellar peduncle.⁵ Such results are similar to our findings on SE-T2WI and GRE-T1WI, with the addition of a high rate of myelination of the corpus callosum splenium and spinal tract/nucleus of cranial nerve V on SE-T2WI, and the cranial nerve V fascicle and pyramidal decussation on GRE-T1WI (each >80%).

Additionally, our study at 3T found a slightly greater extent of myelination overall on SE-T2WI than on GRE-T1WI, which may be due to several factors. First, the use of a 3T field strength likely played a role, in which there is theoretically a 3-fold increase in SNR from 1T that augments the visualization of abnormalities on SE-T2WI.^{11⇓⇓–14} Second, SE-T2WI at 3T may display greater background white matter hyperintensity in premature patients, thus creating greater contrast of certain structures to background white matter. Third, the difference in rates of myelination detection between SE-T2WI and GRE-T1WI is slight and could even be accounted for by interobserver differences.

We acknowledge that it has yet to be proved that T1-bright signal within structures on GRE-T1WI represents true myelination. The potential drawback of SE-T1WI at 3T is that the gray-white matter contrast on SE-T1WI is lessened compared with lower field strengths; this occurs due to prolongation of the T1 times of tissue to the point at which T1 times can be closer together.^15,16 Because the overall contrast between various anatomic landmarks of the brain is likely diminished on SE-T1WI at 3T, GRE-T1WI has been proposed as an alternative to obtain T1WI at 3T. Our results suggest that GRE-T1WI could potentially replace SE-T1WI as a sequence to evaluate myelination at 3T.¹⁷ However, this necessitates future evaluation with long-term follow-up MR imaging and clinical data to determine whether such T1-bright structures truly reflect active myelination or are precursors of myelination. In this regard, a preliminary study using diffusion tensor imaging in preterm infants at term-equivalent age had noted higher fractional anisotropy in such regions but still questioned whether those findings reflected accelerated myelination or maturation versus activity focally induced in affected pathways.¹⁸ Thus, future studies could use diffusion tensor imaging to determine whether T1-hyperintense and T2-hypointense regions at 3T truly reflect myelination.^{18⇓⇓–21}

This study had several limitations, including the inherent limitations of a retrospective study. Qualitative analysis based on subjective scoring can also be considered a limitation, in the absence of a consensus review, though subjectivity is inherent in regular daily practice. We note that 1 goal of this study was to evaluate for such “subjectivity” at 3T to ascertain whether and in what structures myelination could be reproducibly identified and found that, at a minimum, the DSCP, ICP, and lateral lemniscus were reliably identified as myelinated by both observers on both SE-T2WI and GRE-T1WI. Technical limitations also exist, such as motion between sequences or artifacts such as pulsation, which could alter the visualization of smaller structures. Also, other limitations include the lack of a control group (such as term infants) with which to compare myelination patterns, the relatively small number of patients, and the lack of a longitudinal follow-up imaging with time. Despite the fact that the patients within this study demonstrated normal neurologic development at least up to 9 months, we acknowledge that minor neurologic deficits and developmental delays could become manifest at later ages.