fig5

Figure 5. Neuromorphic computing for EEG-based seizure detection based on two-terminal devices. (A) Schematic of a-In₂Se₃-based two-terminal memristor; (B) Current-voltage (I-V) characteristics of the device; (C) Process of computation involved in seizure detection within the system; (D) Current output from dynamic memristor from interictal and ictal data input. The ictal data clip showed a higher response current than the interictal data clip; (E) Crossbar array connected to multichannel sensors. Detailed structure of the 1T-1R memristor device (right); (F) Analog conductance modulation behaviors in RESET process for different V_SL; (G) The average conductance change of 128 memristors in response to pulse train applications across different input signal waveforms; (H) Relationship between average ΔG and signal energy across eight consecutive segments depending on the interictal or preictal states. (A-D) Reproduced with permission Copyright 2023, APL Machine Learning^[82]. (E-H) Reproduced with permission Copyright 2020, Science Advances^[86]. EEG: Electroencephalogram.