Diffusion tensor imaging tractography of the optic radiation for epilepsy surgical planning: a comparison of two methods

Abstract

The optic radiation is a key white matter structure at risk during epilepsy surgery involving the temporal, parietal or occipital lobes. It shows considerable anatomical variability, cannot be delineated on clinical MRI sequences and damage may cause a disabling visual field deficit. Diffusion tensor imaging tractography allows non-invasive mapping of this pathway. Numerous methods have been published but direct comparison is difficult as patient, acquisition and analysis parameters differ. Two methods for delineating the optic radiation were applied to 6 healthy controls and 4 patients with epileptogenic lesions near the optic radiation. By comparing methods with the same datasets, many of the parameters could be controlled. The first method was previously developed to accurately identify Meyer's loop for planning anterior temporal lobe resection. The second aimed to address limitations of this method by using a more automated technique to reduce operator time and to depict the entire optic radiation. Whilst the core of the tract was common to both methods, there was significant variability between the methods. Method 1 gave a more consistent depiction of Meyer's loop with fewer spurious tracts. Method 2 gave a better depiction of the entire optic radiation, particularly in more posterior portions, but did not identify Meyer's loop in one patient. These results show that whilst tractography is a promising technique, there is significant variability depending on the method chosen even when the majority of parameters are fixed. Different methods may need to be chosen for surgical planning depending on the individual clinical situation.

Figure 1

Method 1. Seed region in Meyer's loop near the LGN (red) in the axial (a) and coronal planes (c), with a close up of Meyer's loop (b). Waypoint in the lateral wall of the lateral ventricle (blue) identified in the coronal plane (d).

Figure 2

Method 2. Seed regions in the optic tract (red) in the axial (a) and coronal planes (b). Large regions of interest (ROIs) including voxels above and below the calcarine fissure (red) lateral to the midline in the axial (a) and sagittal planes (c), used both as seed region for the identification of the LGN region and as an endpoint for the final identification of the optic radiation. Seed regions in the LGN region (pink) in axial (d) and coronal planes (e, and f), with the voxel (green) closest to the LGN where the optic tract and optic radiation overlap. Exclusion masks (green) in the midline encompassing both the thalamus and corpus callosum (h), in the thalamus in a coronal plane (i), at the level of the trigone (j) and all three shown in the axial plane (g).

Figure 3

Optic radiation identified by Method 1 only (blue), Method 2 only (red) and both methods (yellow) in a control subject, overlaid on the T1-weighted image. The images are shown in radiological convention with axial slices from inferior to superior and sagittal slices from left to right.

Fig. 4

Optic radiation identified in patient 1 by Method 1 only (blue), Method 2 only (red) and both methods (yellow), overlaid on the T1-weighted image. The arrows indicate the right inferior parietal DNET. Image conventions as Fig. 3.

Fig. 5

Optic radiation identified in patient 2 by Method 1 only (blue), Method 2 only (red) and both methods (yellow) overlaid on the T1-weighted image. The arrows indicate the left temporal cavernoma. Image conventions as Fig. 3, with the coronal images being from anterior to posterior.