Background

Normative mapping is a framework in which features of health-related variables are mapped at the population level. A common example is growth charts, used to evaluate whether a child’s growth falls within the normal range for their age. In neuroscience research, this framework is widely used to map normative ranges of brain activity, for example.

However, normative mapping is particularly challenging in modalities that do not support measurements in healthy control subjects, such as the highly invasive intracranial electroencephalogram (icEEG). An icEEG normative map of brain activity is valuable because it allows us to better understand populations of interest. Further, individuals can be compared to the normative map, helping to identify, characterise, and localise potentially pathological abnormalities. Such work is available in the literature [1–7,13–20], revealing, for example, that icEEG normative mapping can localise epileptogenic tissue [1], and that the abnormality maps that follow are temporally stable [5].

Despite these promising findings, there is currently no detailed guide available for transforming clinical data, such as neuroimaging and icEEG recordings, into a normative map of brain activity localised to standardised brain regions. This literature gap hinders the consistent application of the methodology across studies. To remedy this, our protocol consolidates existing methods and provides key decision points for constructing normative maps comprising regional signal properties, along with (optionally) regional abnormality maps for new subjects.

One of the protocol’s strengths lies in its modular design. Distinct sub-pipelines, such as the spatial processing pipeline, offer a standalone utility. For instance, our own lab has employed this pipeline to demonstrate that the incomplete resection of the icEEG seizure onset zone (SOZ) is not associated with post-surgical outcomes [10], amongst other results [8,9,11,12]. Furthermore, this protocol is written as a dataflow guide, allowing for flexibility. Researchers can choose their preferred programming language and data management software for implementation, decide whether to compute abnormalities, and apply the temporal processing pipeline to other modalities, such as scalp EEG (for instance). Our protocol allows for adaptation to best suit the reader.

While the protocol provides a standardised framework, certain aspects, such as manual resection mask delineation, could benefit from modernisation. Automated methods exist within our own lab [21] and externally [22–25], although their reliability varies [26]. The flexible, guide-like nature of our protocol is both a strength and a weakness. Importantly, we offer pseudo code and emphasise dataflow over low-level implementation, allowing researchers to tailor the process to their needs while fostering a deeper understanding of the method and related literature.

This protocol provides a cohesive data flow on icEEG normative and abnormality mapping, which is currently missing from the literature. We aim to equip neuroscience researchers with the tools needed to develop normative maps, allowing them to explore new hypotheses consistently and effectively.