Recordings in the Human Temporal Lobe: Data Analysis and Spike Sorting

In the present thesis we study the recording of neurons in the human medial temporal lobe (MTL). We take a dual approach to the topic concordant to the scientific nature of the problem. On the one hand, the MTL is one of the most studied structures on the brain, strongly correlated to the formation and retrieval of conscious memories. On the other hand, the direct recording of neurons is a challenging operation requiring advanced methods of signal processing. We used recordings from electrodes implanted in epileptic patients to study the behaviour of MTL neurons with strong responses to visual stimuli. We studied how the repeated stimulus presentation modulated the firing of these neurons. The results showed decreased activity with each presentation and differences between areas in the line of the suggested roles in previous works. We analysed the performance provided by the algorithms used for the extraction of single unit activity out of the signal recorded by electrodes in the brain: a process called spike sorting. The results quantified an inherent limitation to these algorithms, with a maximum number of detected units and significantly reduced performance for those neurons with low firing rate. In summary, the work presented here contributes to the understanding of the ongoing processes in MTL neurons and the problematic of recording the activity of neurons, in especial the ones with low levels of activity.