During the two-year fellowship, the project investigated how individual neurons behave in the human brain during epilepsy, combining unique in vivo and in vitro approaches.
Neuronal activity was analysed from 26 patients undergoing pre-surgical evaluation for drug-resistant epilepsy, using hybrid clinical–research electrodes capable of detecting both large-scale brain signals and the tiny voltage changes produced by single neurons. In addition, recordings were obtained from 27 postoperative brain tissue samples, where neurons could be studied under controlled laboratory conditions using high-density microelectrode arrays.
The in vivo part of the project exceeded expectations. From these recordings, the electrical activity of over one thousand individual neurons was successfully isolated and classified into two main categories—principal cells and interneurons—based on the shape and duration of their action potentials. This large-scale dataset provided one of the most detailed views to date of human neuronal behaviour in epileptic brain tissue.
A key outcome of the project was the development of a new computational algorithm to identify and analyse burst firing—short, rapid sequences of electrical impulses that indicate intense neuronal activation. Using this method, the study provided one of the first systematic descriptions of bursting behaviour in human single-unit recordings. The results revealed that not only excitatory principal cells but also inhibitory interneurons are capable of generating bursts, challenging earlier assumptions and suggesting that bursting is a more universal feature of human neuronal communication than previously thought.
The in vitro recordings complemented these results by offering a controlled environment to explore intrinsic neuronal excitability. Comparative analyses between the two datasets showed that, when isolated from their natural network, interneurons fired less frequently, while principal cells retained similar levels of spontaneous activity. This indicates that interneurons are more dependent on network input, whereas principal cells maintain self-driven firing even in isolation.
Importantly, the same waveform-based classification criteria applied to both datasets produced consistent results, confirming the robustness and reproducibility of the analytical framework.
Finally, an additional measure of firing variability, the short-timescale irregularity coefficient (CV2), revealed that neurons in vitro fired more regularly than those in vivo. This suggests that the natural synaptic environment of the human brain introduces additional variability—an expected signature of ongoing network activity.
