Automated Identification of Stereoelectroencephalography Contacts and Measurement of Factors Influencing Accuracy of Frame Stereotaxy

Objective:
Stereoelectroencephalography (SEEG) is an established invasive diagnostic technique for use in patients with drug-resistant focal epilepsy evaluated before resective epilepsy surgery. The factors that influence the accuracy of electrode implantation are not fully understood. Adequate accuracy prevents the risk of major surgery complications. Precise knowledge of the anatomical positions of individual electrode contacts is crucial for the interpretation of SEEG recordings and subsequent surgery.
Methods:
We developed an image processing pipeline to localize implanted electrodes and detect individual contact positions using computed tomography (CT), as a substitute for time-consuming manual labeling. The algorithm automates measurement of parameters of the electrodes implanted in the skull (bone thickness, implantation angle and depth) for use in modeling of predictive factors that influence implantation accuracy.
Results:
Fifty-four patients evaluated by SEEG were analyzed. A total of 662 SEEG electrodes with 8,745 contacts were stereotactically inserted. The automated detector localized all contacts with better accuracy than manual labeling (p < 0.001). The retrospective implantation accuracy of the target point was 2.4 ± 1.1 mm. A multifactorial analysis determined that almost 58% of the total error was attributable to measurable factors. The remaining 42% was attributable to random error.
Conclusion:
SEEG contacts can be reliably marked by our proposed method. The trajectory of electrodes can be parametrically analyzed to predict and validate implantation accuracy using a multifactorial model.
Significance:
This novel, automated image processing technique is a potentially clinically important, assistive tool for increasing the yield, efficiency, and safety of SEEG.