Semi-automated detection of sEEG electrodes in computed tomography scans

Details

Supervisors

Macq, Benoît ; Kupers, Ron

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en informatique, à finalité spécialisée

Abstract

Stereoelectroencephalography (sEEG) is a critical procedure in the pre-surgical evaluation of patients with drug-resistant epilepsy, allowing the identification of the epileptogenic zone (EZ) through intracranial recordings. However, localizing the implanted electrode contacts in post-operative computed tomography (CT) scans remains a time-consuming and error-prone task, especially when dealing with hundreds of contacts per patient. This thesis proposes an algorithm for the accurate and efficient localization and labeling of sEEG electrode contacts in CT scans. This work leverages mathematical and computational approaches that have not been previously explored in the context of sEEG electrode localization. Mathematical morphology is used to extract punctual information about the contacts and detect intersecting electrodes. Notions of graph theory and graph search procedures are then applied to retrieve optimal linear models for the electrodes. A final post-processing step is performed to account for electrode bending and ensure coherent spacing between the contacts identified. The pipeline is validated on clinical CT data from patients undergoing sEEG at Rigshospitalet, Copenhagen. Results demonstrate promising accuracy and high robustness to CT artifacts and electrode bending.