Chromatographic Waveform Peak Detection for Overlapping Signals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing peak detection methods in chromatography struggle to accurately identify peaks due to variations in analyzer configuration and measurement conditions, particularly when dealing with overlapping peaks and fluctuating baselines, leading to incorrect peak separation.
Innovation Solution
A waveform-analyzing method and device that constructs a trained model using machine learning with diverse reference waveform data, each with known peak positions and baseline shapes, to output indices for peak portions and suitable separation techniques, including tailing processing, complete separation, and vertical partitioning, regardless of device configuration and conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional peak detection methods are used, then the analysis can be performed with simple processing, but the peak detection accuracy deteriorates when dealing with overlapping peaks and fluctuating baselines
Solution Approach 1:
The patent applies parameter changes by transforming the chromatogram data through multiple preprocessing steps including normalization, derivative calculation, and wavelet transformation. These parameter transformations convert the original signal into different domains where overlapping peaks and baseline fluctuations can be more effectively distinguished, thereby improving peak detection accuracy without requiring complex manual intervention
Solution Approach 2:
The patent introduces an intermediary classification model that acts as a mediator between the raw chromatogram data and the final peak detection results. This model receives preprocessed data and outputs classification results that guide the peak detection process, effectively decoupling the complexity of handling overlapping peaks and baseline fluctuations from the final detection step
2Adaptability or versatility
If a trained model is constructed with diverse reference waveform data covering different baseline shapes and detector types, then the adaptability to different measurement conditions improves, but the amount of teaching data and model construction complexity increases
Solution Approach 1:
The patent implements universality by constructing a classification model that can handle multiple types of baseline shapes (increasing, decreasing, fluctuating) and different detector types (mass spectrometry, UV-Vis, PDA) within a single unified framework. The model is designed to process diverse reference waveform data and apply the same classification logic across different measurement conditions, eliminating the need for separate models for each detector type or baseline condition
Solution Approach 2:
The patent applies segmentation by dividing the teaching data into multiple categories based on baseline shapes and detector types. Rather than treating all data uniformly, the system segments reference waveforms into distinct groups (e.g., increasing baseline, decreasing baseline, fluctuating baseline) and trains the model to recognize patterns within each segment. This segmented approach allows the model to learn specific characteristics of each category while maintaining overall versatility
Data Source
AI summary
A waveform-analyzing device includes a trained-model storage section (44) for a trained model which detects a peak from a waveform. The model is constructed by machine learning using reference waveform data as teaching data. Each reference waveform has a different baseline shape and a known position of a peak portion including an overlap peak, with tailing processing, complete separation or vertical partitioning related to this peak. For an input of measurement data, the model outputs an index which represents a single-peak, overlap-peak or non-peak portion and to which the tailing processing, complete separation or vertical partitioning is related as a peak separation technique. A n index outputter (55-57) inputs analysis-target data into the model to obtain an output of the index which represents a single-peak, overlap-peak or non-peak portion and to which the tailing processing, complete separation or vertical partitioning is related as the technique for separating the overlap-peak portion.