神经尖峰检测方法及装置
By using a sample-by-sample update mechanism for the state estimated by the center and the state estimated by the scale, combined with high-pass filtering and exponential moving average, an adaptive threshold is generated. This solves the problems of large memory overhead and unstable detection performance in the existing technology, and realizes adaptive and robust spike detection with extremely low memory, which is suitable for brain-computer interface neural signal processing chips.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- 宁波时识科技有限公司
- Filing Date
- 2026-04-21
- Publication Date
- 2026-07-17
AI Technical Summary
Existing on-chip online spike detection technology struggles to achieve robust adaptive threshold detection under extremely low memory constraints, resulting in high storage overhead, unstable detection performance, poor cross-scenario adaptability, and software/hardware inconsistency issues.
A sample-by-sample update mechanism using center-estimated state and scale-estimated state is adopted, combined with high-pass filtering and exponential moving average to generate an adaptive threshold. Period-by-period peak detection is achieved through fixed-point arithmetic and integrated into a brain-computer interface neural signal processing chip.
It achieves extremely low memory usage, adaptive and robust spike detection, has real-time response capability, adapts to different signal scenarios, and has high consistency in software and hardware implementation, and is compatible with multi-channel neural signal processing.
Smart Images
Figure CN122196550B_ABST