LC/MS Peak Picking With AI Chromatogram Preprocessing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing AI algorithms struggle to accurately identify and quantify peaks in complex biological samples due to contaminations, impurities, and matrix effects in LC/MS data, leading to reduced performance in peak picking.

Innovation Solution

A method involving pre-processing steps to generate an artificial chromatogram by subtracting median values from zero samples, applying retention time windows, scaling, smoothing, and using AI algorithms like U-net to enhance peak detection accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If AI algorithms are used for peak picking in complex biological samples, then automation and processing speed are improved, but measurement precision deteriorates due to contaminations, impurities, and matrix effects

Engineering Contradiction:
Improveautomation of peak pickingVSAvoidpeak detection accuracy
Core Design Contradiction:
Extent of automationVSMeasurement precision

Solution Approach 1:

The patent applies preliminary action by performing pre-processing steps before AI-based peak picking. Specifically, median subtraction of zero samples is performed to remove background contaminations and matrix effects before the AI algorithm processes the chromatogram. This preliminary cleaning action improves the precision of subsequent automated peak detection.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If pre-processing steps are added to clean up the chromatogram, then measurement precision is improved, but device complexity increases

Engineering Contradiction:
Improvepeak detection accuracyVSAvoidcomplexity of data processing system
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies parameter changes by modifying the chromatogram data through mathematical operations (median subtraction) rather than adding physical components. The pre-processing step changes the parameter values of the signal by subtracting the median background from zero samples, thereby improving peak detection accuracy without significantly increasing system complexity.

Inventive Principle:
Principle #35Parameter changes

3Device complexity

If traditional peak picking methods are used without pre-processing, then device complexity is reduced, but measurement precision deteriorates in complex samples

Engineering Contradiction:
Improvesimplicity of processing methodVSAvoidpeak picking accuracy
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent introduces a preliminary action step (median subtraction of zero samples) that can be integrated into existing AI-based peak picking workflows. This addition maintains the automation and simplicity of the overall system while significantly improving measurement precision in complex biological samples by removing background contaminations.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4597098A1Method for peak picking in LC/ms
Publication Date: 2025.08.06 BIOCRATES LIFE SCIENCES AG
  • EP4597098A1 patent drawingFigure 1~4
  • EP4597098A1 patent drawingFigure 5~8
  • EP4597098A1 patent drawingFigure 9

AI summary

The present invention relates to a new method for peak picking by using at least one AI algorithm wherein the presence of at least one compound corresponds to the regions in a Liquid Chromatography/Mass Spectrometry (LC/MS) chromatogram by detecting and identifying peaks with pre-processing steps for cleaning up the chromatogram in order to improve the results of the AI algorithm on complex samples.