Receiver design for a bionic nervous system: Modeling the dendritic processing power

Intrabody nanonetworks for nervous system monitoring are envisioned as a key application of the Internet of Nano-Things (IoNT) paradigm, with the aim of developing radically new medical diagnosis and treatment techniques. Indeed, very recently, bionic devices have been implanted inside a living human brain as innovative treatment for drug-resistant epilepsy. In this context, this paper proposes a systems-theoretic communication model to capture the actual behavior of biological neurons. Specifically, biological neurons exhibit physical extension due to their projections called dendrites, which propagate the electrochemical stimulation received via synapses to the soma. Experimental evidences show that the dendrites exhibit two main features: 1) the compartmentalization at the level of the dendritic branches of the neuronal processes and 2) the location-dependent preference for different frequencies. Stemming from these experimental evidences, we propose to model the dendritic tree as a spatiotemporal filter bank, where each filter models the behavior in both space and time of a dendritic branch. Each filter is fully characterized along with the overall neuronal response. Furthermore, sufficient conditions on the incoming stimulus for inducing a null-neuronal response are derived. The conducted theoretical analysis shows that: 1) the neuronal information is encoded in the stimulus temporal pattern, i.e., it is possible to select the neuron to affect by changing the stimulus frequency content; in this sense, the communication among neurons is frequency-selective and 2) the spatial distribution of the dendrites affects the neuronal response; in this sense, the communication among neurons is spatial-selective. The theoretical analysis is validated through a real neuron morphology.