How accurate are prenatal tractography results? A postnatal in vivo follow-up study using diffusion tensor imaging

Abstract

Prenatal detection of abnormal white matter tracts might serve as a structural marker for altered neurodevelopment. As a result of many technical and patient-related challenges, the accuracy of prenatal tractography remains unknown. We hypothesized that characteristics of prenatal tractography of the corpus callosum and corticospinal tracts derived from fetal diffusion tensor imaging (DTI) data are accurate and predictive of the integrity of these tracts postnatally. We compared callosal and corticospinal tracts of 12 subjects with paired prenatal (age: 23–35 gestational weeks) and postnatal (age: 1 day to 2 years) DTI examinations (b values of 0 s/mm2 and 700 s/mm2, 16 gradient encoding directions) using deterministic tractography. Evaluation for the presence of callosal segments and corticospinal tracts showed moderate degrees of accuracy (67–75%) for the four segments of the corpus callosum and moderate to high degrees of accuracy (75–92%) for the corticospinal tracts. Positive predictive values for segments of the corpus callosum ranged from 50% to 100% and for the corticospinal tracts, 89% to 100%. Negative predictive values for segments of the corpus callosum ranged from 25% to 80% and for the corticospinal tracts, 33% to 50%. The results suggest that when the tracts are not well characterized on the fetal MR, predictions about the postnatal tracts are difficult to make. However, accounting for brain maturation, prenatal visualization of the main projection and commissural tracts can be clinically used as an important predictive tool in the context of image interpretation for the assessment of fetal brain malformations.