Data-driven head model individualization from digitized electrode positions or photogrammetry improves M/EEG source localization accuracy.

We propose a data-driven algorithm to approximate individual head anatomies to improve source localization accuracy over the widely used standard head models Colin27 and ICBM-152 when structural MRI/CT scans are not available. Based on a low-dimensional representation of a large head model database, we derive individual head shape parameters solely from additional knowledge of the subject's scalp, which is obtained, for example, from photogrammetry scans or precise electrode positions. We demonstrate in an experimental study of 16 subjects that our approach provides better-approximated head model anatomies than other existing approaches, even when using scalp proxies derived from a smartphone scan. Moreover, in an EEG simulation study involving 22 heads, we show that our head models outperform standard and other individualization approaches in terms of source localization accuracy. As our proposed head model individualization method does not require structural scans of each subject, it can help improve source localization with minimal effort in future M/EEG studies, particularly when MRI/CT scans are not available.