Extracellular recordings of electric potential remain a popular tool for investigations of brain activity on all scales in animals and humans, from single cells (spikes) to systems studied with depth electrodes (LFP, SEEG), subdural recordings (ECoG), and on the scalp (EEG). They are relatively easy to record but difficult to interpret: since electric field is long range one can observe neural activity several millimeters from its source. As a consequence, every recording reflects activity of many cells, populations and regions, depending on which level we focus. One way to overcome this problem is to reconstruct the distribution of current sources (CSD) underlying the measurement.

We recently proposed a kernel-based method of CSD estimation from multiple extracellular recordings from arbitrarily placed probes which we called kernel Current Source Density method (kCSD). In my talk I will present this method and show two recent developments, skCSD (single cell kCSD) and kESI (kernel Electrophysiological Source Imaging). skCSD assumes that we know which part of the recorded signal comes from a given cell and we have access to the morphology of the cell. This could be achieved by patching a cell, driving it externally while recording the potential on a multielectrode array, injecting a dye, and reconstructing the morphology. In this case we know that the sources must be located on the cell and this information can be successfully used in source estimation. In kESI we consider simultaneous recordings with subdural ECoG (strip and grid electrodes) and with depth electrodes (SEEG). Such recordings are taken on some epileptic patients prepared for surgical removal of epileptogenic zone. When MR scan of the patient head is taken and the positions of the electrodes are known as well as the brain’s shape, the ideas behind kCSD can be applied to constrain the possible distribution of sources facilitating localization of the epiletogenic foci.