The Project CLEPSYDRA (Towards a Closed-Loop Epileptogenic Prediction SYstem based on sub-Dural Recording Arrays
) aims to provide enabling technologies for the integration and miniaturization of electronic systems, which can be used for building neurocortical implants suitable for scientific (to allow new advances in neuroscience) and clinical (to provide neuroprosthesis for the treatment of neurological diseases) issues. Within this general scenario, our objective will be to provide the basis towards a reliable and efficient closed-loop epileptogenic prediction system with higher specificity and sensitivity than current therapeutic devices. This will be accomplished by the heterogeneous integration and 3D stacking of three different System-on-Chips (SoC); one for the acquisition and conditioning of neural activity, another to process the recorded information and extract only those parameters relevant to effective seizure prediction, and one final SoC for powering and communication purposes. Additionally, novel versatile setups for measuring the system under in vivo
and in vitro
conditions will be provided.
The seamless integration of all these functional blocks involves many exciting design challenges and, thereby, represents a fertile scenario for new ideas in the field of ultra-low power applications (needed to avoid tissue damages due to excessive heating) in a wide range of frequencies, from nearly DC up to the tens of MHz band. At low frequencies, in the mHz to kHz range, advances on the design of low-noise, low-power amplifiers for biopotential measurements, as well as on the design of band-pass filtering and converting stages will be pursued. In the MHz range, short-distance wireless data transfer/powering techniques based on inductive links will be proposed. This is done to avoid large size battery or percutaneous physical links that are prone to damage due to mismatch in material properties or scarring at the tissue interface. At intermediate frequencies, a sophisticated bio-signal processor involving tasks such as artifact removal, real-time spike sorting and classification, or multivariate seizure prediction will be targeted.
This work has been supported by the Spanish Ministry of Science & Innovation under grant TEC2012-33634 and the FEDER Program.
, Instituto de Microeletrónica de Sevilla-CNM (CSIC-Univ. Sevilla)