Quantitative susceptibility mapping with source separation in normal brain development of newborns.

Abstract

Background and purpose: Quantitative susceptibility mapping is an emerging method for characterizing tissue composition and studying myelination and iron deposition. However, accurate assessment of myelin and iron content in the newborn brain using this method is challenging because these two susceptibility sources of opposite signs (myelin, negative; iron, positive) occupy the same voxel, with minimal and comparable content in both sources. In this study, susceptibilities were measured in the normal newborn brain using susceptibility source separation.

Materials and methods: Sixty-nine healthy newborns without clinical indications were prospectively recruited for MRI. All newborns underwent gradient echo imaging for quantitative susceptibility mapping. Positive (paramagnetic) and negative (diamagnetic) susceptibility sources were separated using additional information from R2* with linear modeling performed for the newborn brain. Average susceptibility maps were generated by normalizing all susceptibility maps to an atlas space. Mean regional susceptibility measurements were obtained in the cortical GM, WM, deep GM, caudate nucleus, putamen, globus pallidus, thalamus, and the four brain lobes.

Results: A total of sixty-five healthy newborns (mean postmenstrual age, 42.8 ± 2.3 weeks; 34 females) were studied. The negative susceptibility maps visually demonstrated high signals in the thalamus, brainstem and potentially myelinated WM regions, whereas the positive susceptibility maps depicted high signals in the GM compared to all WM regions, including both myelinated and unmyelinated WM. The WM exhibited significantly lower mean positive susceptibility and significantly higher mean negative susceptibility than cortical GM and deep GM. Within the deep GM, the thalamus showed a significantly lower mean negative susceptibility than the other nuclei, and the putamen and globus pallidus showed significant associations with newborn age in positive and/or negative susceptibility. Among the four brain lobes, the occipital lobe showed a significantly higher mean positive susceptibility and a significantly lower mean negative susceptibility than the frontal lobe.

Conclusions: This study demonstrates regional variations and temporal changes in positive and negative susceptibilities of the newborn brain, potentially associated with myelination and iron deposition patterns in normal brain development. It suggests that quantitative susceptibility mapping with source separation may be used for early identification of delayed myelination or iron deficiency.

Abbreviations: CGM = cortical gray matter; DGM = deep gray matter; PMA = postmenstrual age; QSM = quantitative susceptibility mapping.