Visualization of Subdural Electrodes

We read with interest the technical note of Schulze-Bonhage et al on the visualization of subdural electrodes by curvilinear reformatting of MR imaging in the March issue of the AJNR (1).

We agree with the authors that there are obviously several approaches to localizing subdural electrodes, each of which has its specific advantages and limitations. In their note they also refer to the method used by our group (described in detail by Winkler et al [2]). We fail to understand, however, how they can describe our method as “time-consuming” and estimate that it would require at least “half a day” for data processing as compared with only 30 minutes for their approach. We would like to correct this misrepresentation by giving a more realistic picture of the time frame of our method.

We first use a preimplantation MR imaging study to render a 3D reconstruction of the cortical surface anatomy. Image registration is then performed with a postoperative CT scan to localize the implanted subdural electrodes. Preoperative MR imaging is an integral part of the routine diagnostic evaluation in all patients being considered for epilepsy surgery. After electrode implantation, we use a CT scan to superimpose the electrodes onto the reformatted 3D MR image. By contrast, Schulze-Bonhage et al use postoperative MR imaging. Two advantages of the postoperative CT scan are that it is more readily available in most institutions and clearly less time consuming than postoperative MR imaging. In addition, the CT scan is less prone to movement artifacts, and the localization of electrodes is more precise, because fewer artifacts and no image distortion interfere with the data processing.

The data processing of preoperative MR imaging and postoperative CT in our approach consists of the following steps:

1. Interactive coregistration of MR imaging and CT (5 minutes).

2. Creation of a new combined data set from both studies (2 minutes).

3. Interactive segmentation of the skull in the region of interest (5–15 minutes).

4. Volume rendering of the desired view (>1 minute per view).

Thus, the entire procedure for our data processing lasts less than half an hour, by no means the “half a day” estimated by Schulze-Bonhage et al (1).Our protocol has proved itself highly reliable for electrode localization and has been routinely used since 1998 in all invasive epilepsy surgery candidates in the University of Munich Epilepsy Program.

• Copyright © American Society of Neuroradiology