Deep Learning Waveform Analysis for Epilepsy IED Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual analysis of interictal epileptiform discharges (IEDs) in epilepsy diagnosis using magneto-encephalography or electro-encephalography is time-consuming and prone to false or missed detections due to the complexity of identifying characteristic waveform information across multiple sensors.
Innovation Solution
A waveform generation identifying method that uses a deep learning model to calculate probability information for IED appearance, extracting feature maps, generating attention maps to highlight important regions, and identifying the probability of IED occurrence by multiplying feature maps with attention maps, thereby improving the accuracy of time and sensor detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual analysis is used to identify IEDs, then detection accuracy can be maintained through expert judgment, but the analysis time becomes excessively long and productivity is low
Solution Approach 1:
The patent replaces the manual mechanical analysis process with an automated deep learning system. The neural network model automatically processes waveform data from multiple sensors to detect IEDs, substituting human expert manual review with computational algorithms that achieve both high accuracy and fast processing speed.
Solution Approach 2:
The patent introduces an attention map as an intermediary component between the feature extraction layer and the detection output. This attention map highlights important regions in the waveform data, allowing the system to focus computational resources on critical areas while maintaining detection accuracy and improving overall processing efficiency.
2Productivity
If deep learning is used to automatically extract IED information, then analysis speed and productivity improve, but detection accuracy may decrease due to false positives and missed detections
Solution Approach 1:
The patent applies local quality by generating attention maps that assign different importance weights to different regions of the waveform data. Instead of uniform processing, the system identifies and emphasizes local critical regions where IEDs are likely to occur, improving detection accuracy in specific areas while maintaining overall processing speed.
Solution Approach 2:
The attention map serves as an intermediary that refines the raw deep learning output. It acts as a filtering mechanism that enhances true positives while suppressing false detections, thereby improving detection accuracy without sacrificing the automated processing speed advantage.
3Device complexity
If conventional deep learning methods are used without attention mechanisms, then the system complexity remains low, but the ability to identify important regions and improve detection accuracy is limited
Solution Approach 1:
The patent segments the deep learning model into distinct functional components: feature extraction, attention map generation, and detection output. This segmentation allows the system to maintain manageable complexity while incorporating the attention mechanism that significantly improves detection accuracy by focusing on important regions.
Solution Approach 2:
By introducing the attention map mechanism, the system enhances local quality of processing without overwhelming complexity. The attention mechanism selectively enhances important features while suppressing irrelevant information, achieving high detection accuracy with a relatively simple additive component to the deep learning architecture.
Data Source
AI summary
A waveform generation identifying method including: acquiring waveform data of biological signals measured by a plurality of sensors; calculating probability information of appearance of IEDs (interictal epileptiform discharges) from a deep learning model trained using the waveform data with labels indicating whether the characteristic waveform information appears or not; and first extracting a time and a sensor at which the characteristic waveform information appears in the waveform data, based on the probability information. The calculating includes: second extracting a feature map indicating waveform data characteristics from the waveform data; generating an attention map indicating an important region in IED recognition, from the feature map; and identifying the probability information by inputting information obtained by multiplying the feature map by the attention map.