Probabilistic electrical stimulation mapping of human medial frontal cortex

The medial frontal cortex remains functionally ill-understood; this is reflected by the heterogeneity of behavioural outcomes following damage to the region. We aim to use the rich information provided by extraoperative direct electrical cortical stimulation to enhance our understanding of its functional anatomy. Examining a cohort of 38 epilepsy patients undergoing direct electrical cortical stimulation in the context of presurgical evaluation, we reviewed stimulation findings and classified them in a behavioural framework (positive motor, negative motor, somatosensory, speech disturbances, and “other”). The spatially discrete cortical stimulation-derived data points were then transformed into continuous probabilistic maps, thereby enabling the voxel-wise spatial inference widely used in the analysis of functional and structural imaging data. A functional map of stimulation findings of the medial wall emerged. Positive motor responses occurred in 141 stimulations (31.2%), anatomically located on the paracentral lobule (threshold at p<.05), extending no further than the vertical anterior commissure (VCA) line. Thirty negative motor responses were observed (6.6%), localised to the VCA line (at p < .001 uncorrected). In 43 stimulations (9.5%) a somatosensory response localised to the caudal cingulate zone (at p < .001 uncorrected), with a second region posterior to central sulcus. Speech disturbances were elicited in 38 stimulations (8.4%), more commonly but not exclusively from the language fMRI dominant side, just anterior to VCA (p < .001 uncorrected). In only 2 stimulations, the patient experienced a subjective “urge” to move in the absence of overt movement. Classifying motor behaviour along the dimensions of effector, and movement vs arrest, we derive a wholly data-driven stimulation map of the medial wall, powered by the largest number of stimulations of the region reported (n = 452) in patients imaged with MRI. This model and the underlying data provide a robust framework for understanding the architecture of the region through the joint analysis of disruptive and correlative anatomical maps.