Evaluating, improving and applying Cortico-Cortical Evoked Potentials in Stereoelectroencephalography

Abstract

Stereoelectroencephalography (SEEG) and other invasive epilepsy evaluations aim to both understand brain networks and to classify the epileptogenic zone for resective surgery in patients not responding to medical therapy. Intracranial electrical stimulation in epilepsy is a complex field, fraught with difficulty in interpreting brain signals. The use of low frequency stimulation paradigms which elicit Cortico-Cortical Evoked (CCEP) responses for brain mapping is highly novel, but expanding as a field. The core aim of this PhD is to improve the methods of quantifying connectivity between distant brain regions in response to electrical stimulation. At present, there are precious few studies which examine the methodology employed in CCEP stimulation. Analysis and comparison of signal processing techniques employed to elucidate brain connectivity are practically non-existent in this field. Three key points crucial to brain connectivity estimation were identified: (1) Creating safe and efficacious stimulation paradigms (2) Understanding the area activated during stimulation and (3) Quantifying responses to stimulation in a comparable way. This research contains a tour of all of the major steps involved in CCEP stimulation and processing, separated into six major sections. The first chapter outlines the introductory literature, as well as the research was undertaken to design a safer, less intense stimulus paradigm used in the studies. This was in large part adapted from the author’s published review paper, detailing the key considerations in CCEP stimulation processing. It highlights the more modest CCEP stimulation parameters that were chosen, based on the review of all published CCEP studies. The second chapter details the work undertaken to correctly localise electrode locations, and view multiple types of neuroimaging in register with electrode positions. This culminated in a novel CCEP connectivity visualiser, which was able to view other neuroimaging modalities simultaneously with CCEP connectivity results. Lastly, the process of how to integrate clinical information from the SEEG evaluation was used to plan CCEP stimulation locations to avoid evoking epileptic seizures is described. The third chapter explores the methods available to quantify the volume of tissue activated by low frequency stimulation. Spurred in large part by recent research which showed that a significant portion of CCEP responses are volume conducted. A novel Evaluating, improving and applying Cortico-Cortical Evoked Potentials in Stereo-EEG method was created to measure stimulation artefact to quantify the volume conducted component. Using this method, it has been shown that stimulation artefact amplitude was correlated with early response amplitude close to the stimulating electrode. This concluded that stimulation artefact is a possible way to quantify the volume conducted component of CCEP responses. The fourth chapter contains a wide-ranging overview of post-processing considerations. A demonstration is given of a novel adaptation of a filtering method which will not introduce artefact around strong stimulation artefact locations. Also provided is the detail of a novel method of baseline correction, which has the ability to output a comparable statistical result for each pulse train and stimulation site. An overview of promising similarity metrics to estimate changes between the evoked response potentials and baseline data segments is also shown. This culminated in a novel method to compare these dissimilarity metrics, against those used in CCEPs currently. This found that root mean square, currently employed in CCEP studies, is well suited to define connectivity. The research also found that several metrics were particularly sensitive to epileptiform SEEG patterns. Lastly, novel visualisation software, developed during this research was shown, along with how it can be used to quickly evaluate connectivity estimates is demonstrated. The fifth chapter applies methods developed in chapter 4 to undertake the largest CCEP connectivity study of the human insula to date. A thorough literature review, in conjunction with the results and connectivity visualisations of each insula gyrus, was able to demonstrate novel findings of the insula. Particularly with regards to its role in vision and the integration of other sensory inputs. The results also highlight the importance of sampling the insula in invasive epilepsy investigations, as insula epilepsies can often mimic semiology from all lobes of the brain. The sixth chapter employed a handful of the best performing metrics from chapter 4 to evaluate connectivity between seizure propagation locations. Using a novel approach, it was shown that the autoregressive metric is well suited to evaluate connectivity in epileptic regions. It was also found that epileptogenic zone locations were highly interconnected, but were not affected from stimulation of locations involved later in seizure spread, in line with other published literature. Chapter seven contains an overview and highlights the novel contributions made in this thesis. Lastly chapter eight recommends which parameters to use in clinical practice.