Mass Spectrometry Peak Identification With Deep Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current Liquid Chromatography-Mass Spectrometry (LC-MS) data processing requires manual data review and correction due to high error rates, with existing methods failing to reliably determine the area under peaks.
Innovation Solution
A computer-implemented method using a deep learning regression architecture, specifically a convolutional neural network, to automatically identify and determine the start and end points of peaks in mass spectrometry response curves, enabling accurate peak area calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual data review and correction is used for LC-MS data processing, then reliability of peak identification can be maintained through human expertise, but productivity is reduced due to tedious visual analysis of hundreds of chromatography plots
Solution Approach 1:
The system performs automatic peak identification and area determination through self-learning deep neural networks, eliminating the need for manual review by trained operators. The model autonomously processes chromatography plots and identifies peaks with high accuracy, allowing the system to serve itself without human intervention while maintaining reliability.
Solution Approach 2:
The patent replaces the mechanical manual analysis process with an automated deep learning system. The deep neural network model substitutes human visual analysis and manual correction with automated computational processing, achieving both high productivity through rapid processing of hundreds of plots and maintained reliability through the model's learned patterns.
2Productivity
If automated peak identification methods are used to increase productivity, then processing speed improves, but measurement precision of peak area determination deteriorates due to high error rates
Solution Approach 1:
The system performs preliminary training of deep neural networks using labeled training data before actual peak identification. This preliminary learning phase enables the model to acquire accurate peak detection skills in advance, ensuring high measurement precision when processing actual chromatography data automatically without manual intervention.
Solution Approach 2:
The system uses feedback from training data with known ground truth peak locations to continuously improve the deep learning model's accuracy. The model learns from correct and incorrect predictions during training, adjusting its parameters to minimize errors, which ensures high measurement precision in automated peak area determination.
3Ease of operation
If simple peak presence detection methods are used, then ease of operation is improved, but reliability of quantitative analysis worsens because area under peak cannot be reliably determined
Solution Approach 1:
The patent merges peak detection and peak area determination into a single integrated deep learning model. Instead of separate simple detection followed by manual area calculation, the unified model simultaneously performs both functions, maintaining ease of operation through automation while ensuring reliable quantitative measurement of peak areas through learned patterns from training data.
Data Source
AI summary
A computer implemented method for identifying at least one peak in a mass spectrometry response curve is provided comprising: a) providing at least one mass spectrometry response curve by using at least one mass spectrometry device; b) evaluating the mass spectrometry response curve by using at least one trained model thereby identifying a start point and an end point of at least one peak of the mass spectrometry response curve, wherein the model was trained using a deep learning regression architecture.