Feasibility of diffusion tensor tractography of brachial plexus injuries at 1.5 T

Objective: The objective of this study was to examine the feasibility of diffusion tensor imaging and diffusion tensor tractography (DTT) at 1.5 T for the detection of nerve root avulsions in patients with brachial plexus injuries (BPI). Materials and Methods: We performed a 1.5-T magnetic resonance imaging on 28 patients (mean [SD] age, 25 [9.1]) with BPI using the following imaging protocol: (a) magnetic resonance myelography (MRM), (b) magnetic resonance neurography, and (c) diffusion tensor imaging. A reproducible tractography approach was developed to assess the myeloradicular continuity, which consists of multiple regions of interests placed on each hemicord, including the ventral and dorsal rootlets from C4 to T2 nerve roots. Two independent observers blindly evaluated DTT and MRM studies. The degree of agreement between DTT and MRM findings was estimated on a per-root basis on the 140 nerve roots (C5-T1) on the injured side by calculation of the κ coefficient (K value) and the Bland-Altman plot analysis. The diagnostic accuracy of DTT was assessed by comparing it with the MRM findings of the 140 nerve roots on the injured side on a per-root basis. Results: Diffusion tensor tractography allowed a complete visualization of the C5-T1 intact nerve roots on the normal side in 100% of studies.Complete nerve root avulsions were recognized on DTT either as a total loss of fibers or as a very short segment of incoherent fibers in apparent continuity with the spinal cord.The MRM identified 88 intact nerve roots (62.9%), 44 completely avulsed nerve roots (31.4%), and 8 partially avulsed nerve roots (5.7%). The DTT and MRM were concordant in 127 of the 140 nerve roots (90.7%) and exhibited an excellent overall agreement (K value, 80.8). The brachial plexus DTT had an 88.1% sensitivity, 98.1% positive predictive value, 98.8% specificity, 92.6 negative predictive value, and a 94.5% overall accuracy for detecting the presence of a nerve root avulsion. The κ coefficients for the interobserver reliability of DTT and MRM were 0.85 and 0.80, respectively. Conclusions: Our results suggest that cervical nerve root avulsions can be successfully visualized at 1.5 T in patients with BPI despite the anatomical complexity and susceptibility and motion artifacts. We propose that DTT is a reliable and reproducible method for the investigation of BPI because it provides a successful anatomical and functional display of neural structures that are not otherwise attainable with conventional studies.