Nancy Katherine Rollins MD, Presenter: Nothing to Disclose

Lina Chalak MD, Abstract Co-Author: Nothing to Disclose

Michael Craig Morriss MD, Abstract Co-Author: Nothing to Disclose

Youngseob Seo PhD, Abstract Co-Author: Nothing to Disclose

Zhiyue Jerry Wang PhD, Abstract Co-Author: Nothing to Disclose

The scalar metrics available with diffusion tensor imaging (DTI) include  fractional anisotropy (FA), axial (AD) and radial diffusivity (RD). We assessed the sensitivity, specificity, and accuracy of these measures relative to conventional MRI (cMRI) in the characterization of abnormal (AbWM) and normal appearing white matter (NAWM) in the neonatal brain.

There were 56 eligible patients (32M; 24F) who were 35-48 weeks post conception age imaged from 2009 –2011. Indications for MRI were seizures in 18; HIE sp/cooling in 14; abnormal HUS in 7; neonatal respiratory failure s/p ECMO in 6; and misc in 11. Exclusion criteria were congenital abnormalities, hydrocephalus, or significant intracranial hemorrhage . DTI parameters (3T) were b=700; 2mm3; 30 directions. Tensor data was spatially normalized using TBSS modified for multi-subject analysis of neonates. Custom program written in IDL was used to interrogate the internal capsules-posterior limbs; para-sagittal and peri-Rolandic white matter (WM); inferior and superior longitudinal fasciculus; and the callosal genu and splenium. Blinded to results of DTI, 2 experienced pediatric neuroradiologists assessed cMRI for T1, T2, or diffusion abnormalities in these locations. SISA was used to calculate sensitivities and specificities and predictive accuracies of DTI compared to cMRI; Kappa measures of agreement are at 95% confidence intervals. This study was IRB approved.

AbWM was seen in the selected regions of WM in 7/56 patients; acute stroke and diffuse WM injury were associated with both increased and decreased FA. Eight patients had abnormal FA, AD, and/or RD in >1 region of NAWM. Using FA, the sensitivity of DTI varied from 33-100%; specificity from 94-98%; concordance 91-96%; kappa 0.4-.7 9. Using FA+AD, the sensitivity of DTI was 75-100%; specificity was 92-98%; concordance 91-98%; kappa 0.63-.91. The addition of RD to FA+AD did not improve the overall accuracy of DTI.  

The use of the individual diffusivities increased the sensitivity of DTI to abnormal WM and suggested abnormal WM in some patients with NAWM. Pathologic alterations in FA were due more to changes in axial than radial diffusivity. Clinical follow up is ongoing.

Although time-consuming, determination of individual diffusivities enhances the sensitivity of DTI to white matter injury beyond that provided by FA alone.

Rollins, N, Chalak, L, Morriss, M, Seo, Y, Wang, Z, Comparison of DTI and Conventional MRI in Neonates: Is Determination of the Individual Tensor Metrics Worth the Trouble?.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11008428.html