From Bedside to Desktop: A Data Protocol for Normative Intracranial EEG and Abnormality Mapping

Abstract

Normative mapping is a framework used to map population-level features of health-related variables. It is widely used in neuroscience research, but the literature lacks established protocols in modalities that do not support healthy control measurements, such as intracranial electroencephalograms (icEEG). An icEEG normative map would allow researchers to learn about population-level brain activity and enable the comparison of individual data against these norms to identify abnormalities. Currently, no standardised guide exists for transforming clinical data into a normative, regional icEEG map. Papers often cite different software and numerous articles to summarise the lengthy method, making it laborious for other researchers to understand or apply the process. Our protocol seeks to fill this gap by providing a dataflow guide and key decision points that summarise existing methods. This protocol was heavily used in published works from our own lab (twelve peer-reviewed journal publications). Briefly, we take as input the icEEG recordings and neuroimaging data from people with epilepsy who are undergoing evaluation for resective surgery. As final outputs, we obtain a normative icEEG map, comprising signal properties localised to brain regions. Optionally, we can also process new subjects through the same pipeline and obtain their z-scores (or centiles) in each brain region for abnormality detection and localisation. To date, a single, cohesive dataflow pipeline for generating normative icEEG maps, along with abnormality mapping, has not been created. We envisage that this dataflow guide will not only increase understanding and application of normative mapping methods but will also improve the consistency and quality of studies in the field. Key features • Resultant normative maps can be used to test a broad range of hypotheses in the neuroscience field. • Provides a more detailed walkthrough of the methods in the normative mapping study conducted by Taylor et al. [1] and other related publications [2-12]. • Offers flexibility: readers can tailor the final output by considering key decision points included throughout the protocol. • Involves sub-pipelines, which may be useful to researchers in isolation (i.e., icEEG electrode localisation and/or interictal segment selection).

Keywords: Abnormality mapping; Computational methods; Data pipeline; EEG; Epilepsy; Neuroimaging; Neuroscience; Normative mapping.

Figure 1.. Example folder structure for database creation (pipeline C) and subsequent normative mapping (pipeline D)

Figure 2.. Example timeline of an icEEG exam for a dummy subject, demonstrating where suitable periods of interictal data would be and the temporal location of the final interictal segments.

In this example, the normative map is being constructed in the wake state (DP A1), and three segments are being selected (DP A6) under the constraint that they are 2 h away from seizures (DP A2), 4 h away from one another (DP A7), and over 24 h after implantation (quality control).

Figure 3.. Example icEEG data to aid in identifying suitable periods and segments. Each panel shows a different subject and a subset of 30 channels.

(A) A 380 s data period involving a seizure. The seizure onset, as marked by clinicians, is indicated by a green line. This line would correspond to the start of a red unsuitable box in Figure 2. (B) A 380 s data period showing non-physiological noise, indicated by the streaky appearance. Such large amplitudes are not representative of brain activity, and data exhibiting this should be discarded. (C) A clean 380 s data period, ready to proceed to segment selection. This would correspond to the suitable interictal period boxes in Figure 2. (D) A 70 s interictal segment with three examples of unsuitable channels highlighted in yellow. The segment is usable as long as these channels are noted so they can be removed downstream. (E) The power spectral densities (PSDs) of a 70 s interictal segment. Suitable channels are coloured black, showing a 1/f curve with power line noise at 60 Hz. The red line shows a channel with a broad peak around 60 Hz and an overall higher power in most frequencies (unlike the black channels), which most likely indicates a faulty channel. The blue line shows a channel that likely represents non-physiological activity due to its oscillatory pattern in the frequency domain. Both channels should be marked as unsuitable. (F) A clean 70 s interictal segment ready to be extracted and stored.

Figure 4.. Quality control checks and outputs from pipeline B for an example subject.

(A) Correctly aligned post-implantation CT scan and orig.mgz file following co-registration (quality control, step B2). Overlaid in blue is the resection mask, which delineates the tissue that was subsequently resected during surgery (quality control, step B5). The mask was generated using the RAMPS pipeline [21]. (B) Rendering of the subject’s implanted electrodes following channel localisation (quality control, step B4). The channels on each electrode are indicated by coloured dots; blue channels were defined as recording from subsequently resected tissue, whilst red channels were spared (step B9). Looking at both panels A and B, it can be determined that the subject had surgery on the temporal lobe. (C) Visualisation of the Desikan–Killiany parcellation [36] to which channels will be localised.

Figure 5.. Visualisation of the recommended database structure as described in detail throughout pipeline C.

All black lines indicate one-to-many relationships. The yellow box indicates the highest level of the database structure, Subject at level one. Subjects can have anywhere between 0 and 3 of the subsequent level-two classes, indicated by their blue colour (Epilepsy Status, Exam, Treatment). When a subject has an icEEG exam, there is also the option to input any (or none) of the three level-three classes, indicated in red. These are Channel, icEEG Segment, and Seizure. Within each class, required fields are indicated in bold font, whilst optional fields are in standard font. Constrained inputs are indicated by italics in parentheses.