Scalogram Selection for EEG High-Frequency Oscillation Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated tools for detecting high-frequency oscillations in electroencephalographic signals, such as Fast Ripples, are inefficient due to high false positive rates and require extensive manual analysis, making it difficult to diagnose epileptic zones effectively.
Innovation Solution
A method using a convolutional neural network to preprocess electroencephalographic data into scalograms, followed by characteristic calculation and selection, significantly reduces false positives by applying deep-learning and signal processing techniques in a two-step segregation process, allowing for efficient detection of high-frequency oscillations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated detection methods are used to identify high-frequency oscillations, then detection speed increases, but false positive rate increases
Solution Approach 1:
The patent divides the detection process into multiple sequential stages: initial automated detection to identify candidate events, followed by manual verification of each candidate. This segmentation allows the system to process large volumes of data quickly while maintaining high accuracy through human review of potential false positives.
Solution Approach 2:
The patent introduces an intermediary manual verification step between automated detection and final confirmation. This intermediary process acts as a filter to eliminate false positives while preserving true detections, resolving the contradiction between speed and accuracy.
2Reliability
If manual analysis is used to detect high-frequency oscillations, then detection accuracy is maintained, but time consumption increases
Solution Approach 1:
The patent applies partial automation by using automated detection only for the initial identification of candidate events, while reserving manual analysis for verification. This partial action approach maintains high accuracy while reducing the overall time investment compared to complete manual analysis.
Solution Approach 2:
The patent performs preliminary automated detection to pre-sort and identify candidate events before manual verification. This preliminary action filters out clearly non-relevant segments, reducing the time required for manual review while maintaining detection accuracy.
3Adaptability or versatility
If the scale of analysis is increased to capture more neural populations, then representativeness improves, but high-frequency oscillations become undetectable
Solution Approach 1:
The patent applies local quality analysis by examining small, localized segments of neural activity for high-frequency oscillations rather than analyzing large-scale aggregated data. This allows the detection of subtle local phenomena while maintaining the ability to map results across broader neural networks.
Solution Approach 2:
The patent transitions from analyzing raw amplitude data to analyzing the temporal derivatives and frequency characteristics of neural signals. This dimensional change allows detection of high-frequency oscillations that are invisible in standard amplitude representations, enabling precise local detection while maintaining broader contextual understanding.
Data Source
AI summary
A method for selecting data derived from an electroencephalogram, in the form of a set of starting scalograms, each scalogram being calculated from a portion of an electroencephalographic signal. The method includes: extracting, via an artificial neural network, a set of candidate scalograms; and for some candidate scalograms of the set of candidate scalograms: calculating characteristics of the electroencephalographic signal portion corresponding to the candidate scalogram; and when the plurality of characteristics are within prerequisite value ranges, selecting the electroencephalographic signal portion of the candidate scalogram within an electroencephalographic signal selection data structure.