Cardiovascular Signal Segmentation Using Bayesian Verification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for dividing cardiovascular signals into segments often fail due to non-periodic signals, high noise levels, and incorrect segmentation, which can lead to inaccurate diagnostic parameters.
Innovation Solution
A method that identifies segments based on prior knowledge of the signal characteristics and verifies them using statistical parameters, such as duration, frequency, and energy, to ensure accurate segmentation and discard noisy segments, utilizing a Bayesian network for probability calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If periodic segmentation is used to divide the signal into segments, then the signal can be divided into equal time periods, but the segmentation becomes incorrect when the signal is not 100% periodic
Solution Approach 1:
The patent applies preliminary action by pre-defining segment characteristics (duration, frequency content, energy levels) before segmentation. The method uses prior knowledge of cardiovascular signal structure to establish expected parameters for systolic and diastolic segments, then verifies actual segments against these predefined criteria, ensuring accurate segmentation even when signals are not perfectly periodic
Solution Approach 2:
The patent implements feedback by verifying each segmented portion against statistical parameters and prior knowledge. The system continuously checks whether segmented signals match expected cardiovascular patterns (e.g., duration ranges, frequency content), and can adjust or reject segments that do not conform, thereby maintaining high segmentation accuracy despite signal variability
2Reliability
If characteristic-based segmentation is used to find special characteristics in the signal, then segmentation can adapt to signal variations, but the segmentation fails when the signal to noise ratio is very bad
Solution Approach 1:
The patent applies parameter changes by analyzing multiple signal parameters simultaneously (duration, frequency content, energy levels) rather than relying on a single characteristic. By examining the signal from multiple parameter perspectives and comparing against statistical ranges for each parameter, the method can reliably identify cardiovascular segments even in high-noise conditions where individual parameters might be obscured
3Reliability
If individual segments contain a lot of noise, then the segments become unusable for further data handling, but filtering noise reduces the useful signal
Solution Approach 1:
The patent applies taking out by extracting and removing only the identified noisy segments from further analysis. The verification step identifies which segments meet quality criteria and which do not, allowing the system to exclude only the unusable portions while preserving all valid segment data for diagnostic processing, thus avoiding both noise contamination and information loss
Data Source
AI summary
This invention describes a method for dividing a substantial cycli cardiovascular signal into segments by determining the characteristics of said cyclic signal, wherein each cycle in said signal comprises at least two characteristic segments and wherein the method comprises the steps o identifying segments in a cycle based on prior knowledge of said segment characteristics and the step of verifying said identified segments based on a number of statistical parameters obtained from prior knowledge relating to said cyclic signal. Furthermore, the invention describes a system adapted to perform the above-described method.