Emad Ahmadi MD, Presenter: Nothing to Disclose

Anastasia Yendiki, Abstract Co-Author: Nothing to Disclose

Louis Puybasset MD, PhD, Abstract Co-Author: Nothing to Disclose

Damien Pierre Galanaud MD, PhD, Abstract Co-Author: Research Consultant, Olea Medical

Omid Khalilzadeh MD, MPH, Abstract Co-Author: Nothing to Disclose

Lionel Velly MD, PhD, Abstract Co-Author: Nothing to Disclose

Vincent Perlbarg PHD, Abstract Co-Author: Nothing to Disclose

Rajiv Gupta PhD, MD, Abstract Co-Author: Nothing to Disclose

Quantification of injuries to white-matter (WM) bundles in diffusion magnetic resonance images (dMRI) has a great potential for prognostication of coma caused by traumatic brain injury (TBI). We studied a new method for reconstructing 18 WM bundles automatically in dMRI with the purpose of quantifying and localizing damage along each bundle. We tested this method for predicting neurologic and cognitive outcomes caused by bundle injuries in TBI-associated coma.

We studied dMRI and T1 images of 53 patients who remained comatose at least 7 days after TBI, and 17 controls. We used Freesurfer for automatic segmentation and labeling of brain substructures in T1 images. Fully automated probabilistic tractography was performed with TRACULA (Tracts Constrained by Underlying Anatomy). Up to two diffusion orientations, corresponding to crossing fiber bundles, were fit to the dMRI data at each voxel in WM. This information was combined with the structural segmentation extracted from T1 images to reconstruct 18 WM bundles for each subject. Diffusion anisotropy and diffusivity were calculated at every point along the trajectory of each bundle in each subject. These values were compared between subjects at each point along each bundle. Comparisons were made between patients and controls, and between patients with good and poor outcome. Clusterwise correction was used to correct for multiple comparisons. The injured areas of WM bundles in each patient were then extracted by comparing each patient’s anisotropy values along WM bundles with the distribution of the same values in controls.

Thirteen WM bundles showed significant difference at least in one region of neighboring points between comatose patients and controls, and 11 WM bundles showed significant difference at least in one region between patients with good and poor outcome. The figure shows the injured areas of WM bundles in a patient with poor outcome.

Our method for dMRI analysis using TRACULA allows us to extract clinically relevant information about the integrity of each WM bundle that can differentiate between patients with good and poor outcome, and might facilitate decision making for patients in coma caused by TBI.

We have studied a new method for analysis and visualization of diffusion imaging, and have shown its use in prognostication and decision making for patients with TBI-associated coma

Ahmadi, E, Yendiki, A, Puybasset, L, Galanaud, D, Khalilzadeh, O, Velly, L, Perlbarg, V, Gupta, R, Prognostication of Coma Caused by Traumatic Brain Injury Using Quantification of Damage to Individual White-matter Bundles in Diffusion Magnetic Resonance Imaging.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14013671.html