Real-time Neural Spike Detection on Low-power Implantable SoC
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing neural spike detection methods suffer from high computational resource consumption and false positives, making them unsuitable for low-power, real-time applications in implantable devices.
Innovation Solution
A system-on-chip (SoC) approach that filters and models neural signals to identify characteristic values, compares them to thresholds, and classifies spikes using fixed-point arithmetic, reducing power consumption and processing requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If machine learning algorithms are used for neural spike detection, then detection accuracy is improved, but power consumption increases
Solution Approach 1:
The patent segments the neural signal processing into distinct stages: filtering to extract relevant frequency components, feature extraction to identify characteristic spike patterns, and classification to categorize detected spikes. This segmentation allows each stage to be optimized independently, achieving accurate spike detection while minimizing computational complexity and power consumption compared to applying machine learning algorithms to the entire signal processing pipeline.
Solution Approach 2:
The patent extracts specific characteristic features from the neural signal that are indicative of spikes, such as amplitude thresholds, frequency components, and temporal patterns. By extracting only these relevant features rather than processing the entire signal with complex machine learning models, the system achieves accurate spike detection with significantly reduced computational requirements and lower power consumption.
2Measurement precision
If complex algorithms are used for spike detection, then detection accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by using different processing strategies for different aspects of the signal. Simple threshold-based detection is applied for amplitude-based spike identification, while more sophisticated filtering and feature extraction are applied only where needed to distinguish true spikes from artifacts. This localized application of complexity achieves high detection accuracy without requiring the entire system to be complex.
Solution Approach 2:
The patent changes parameters such as filtering coefficients, threshold values, and feature extraction parameters to optimize detection accuracy for different neural signal characteristics. By adjusting these parameters rather than changing the fundamental algorithmic approach, the system maintains low computational complexity while adapting to different detection requirements and signal types.
3Productivity
If simple threshold-based detection is used, then computational resources are reduced, but false positives increase
Solution Approach 1:
The patent performs preliminary filtering and feature extraction before applying threshold-based detection. By pre-processing the signal to remove artifacts and highlight characteristic spike features, the system reduces the false positive rate of subsequent threshold-based detection while maintaining high processing efficiency. This preliminary action prepares the signal in advance, making simple threshold comparison more reliable.
Solution Approach 2:
The patent incorporates feedback mechanisms where detection results are used to adjust processing parameters in real-time. When false positives are detected, the system adjusts thresholds or filtering parameters to reduce future false positives. This feedback loop maintains high processing efficiency while continuously improving detection reliability based on observed signal characteristics.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
A method is described for real-time detecting and classifying of a characteristic signal, such as a neural spike, and forwarding information for further processing if it meets certain criteria. A system (e.g., an on-chip system implanted in a subject's cranium with limited processing power) receives an electrical biological signal. The system filters the signal to generate a filtered signal and fits the filtered signal to a model. The system identifies a set of fit values based on the model, the set of fit values comprising a plurality of sample amplitude values and a respective plurality of time values. Based on the fit values, the system computes a set of characteristic values. The system compares the characteristic values to a corresponding set of threshold values. Based on the comparison, the system determines whether the received biological signal corresponds to a neural spike and, if a spike is detected, forwards on information.