Polymicrobial Detection via Spectral Decomposition

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for identifying microbial mixtures using spectrometry or spectroscopy are limited to single-type microorganisms, requiring time-consuming sample preparation and culture processes, and fail to accurately detect multiple microorganisms, leading to errors and inefficiencies.

Innovation Solution

A method that analyzes biological samples using a single measurement to detect and identify multiple microorganisms by constructing candidate models from reference intensity vectors, selecting the best model based on reconstruction error and complexity, and determining the presence of microorganisms through multidimensional digital signal processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If classification-based identification methods are used, then single microorganism identification is achieved, but polymicrobial mixture detection fails

Engineering Contradiction:
Improveidentification accuracyVSAvoidpolymicrobial detection capability
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The spectrum of a polymicrobial mixture is segmented into multiple spectral contributions, each corresponding to a different microorganism present in the sample. The algorithm decomposes the composite spectrum into individual microbial spectra, enabling identification of each component organism rather than treating the mixture as a single entity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The method transitions from traditional single-dimension classification to a multi-dimensional spectral decomposition approach. By analyzing the spectrum as a combination of multiple reference spectra with varying weights, the system adds a dimensional layer of complexity that enables simultaneous identification of multiple microorganisms.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Reliability

If biological treatment and isolation steps are implemented, then sample preparation is thorough, but analysis time increases significantly

Engineering Contradiction:
Improvesample preparation completenessVSAvoidincubation and culture time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The method extracts the identification capability directly from the raw spectral data without requiring physical extraction or isolation of individual microorganisms through culture methods. The spectral decomposition algorithm extracts information about each microorganism present in the mixture directly from the combined spectrum.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent replaces the mechanical/biological isolation system (culture media, incubation, colony separation) with a computational spectral decomposition system. Instead of physically separating microorganisms through biological processes, the algorithm separates their spectral signatures through mathematical decomposition.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Quantity of substance

If culture media and incubation processes are used, then microorganism growth is achieved, but cost and complexity increase

Engineering Contradiction:
Improvemicroorganism growthVSAvoidculture medium and incubation requirements
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The method enables direct analysis of the original sample without requiring the sample to undergo additional processing or growth steps. The spectral decomposition algorithm allows the sample to 'serve itself' by providing sufficient spectral information for identification of multiple microorganisms in their native mixture state.

Inventive Principle:
Principle #25Self-service

4Measurement precision

If preliminary isolation steps are performed, then identification accuracy is improved, but productivity decreases

Engineering Contradiction:
Improveidentification accuracyVSAvoidanalysis throughput
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The spectral decomposition process enables continuous analysis of polymicrobial samples without interruption for isolation or culture steps. The algorithm processes the spectrum directly and continuously identifies multiple microorganisms, maintaining a continuous workflow rather than requiring discrete isolation and sequential analysis steps.

Inventive Principle:
Principle #20Continuity of useful action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Enables the detection and identification of multiple microorganisms in a biological sample with high accuracy and efficiency, reducing the need for extensive sample preparation and minimizing errors, by using a single measurement to reconstruct intensity vectors and determine taxon presence.

Implementation Method 1

using measurement techniques producing a multidimensional digital signal representative of the sample

Methodology Applied
Scientific EffectSpectroscopy: Absorption Spectroscopy

Implementation Method 2

use spectrometry or spectroscopy to identify microorganisms

Methodology Applied
Scientific EffectSpectrometry: Absorption Spectroscopy

Data Source

PatentEP3028202B1Method and device for analysing a biological sample
Publication Date: 2017.09.13 BIOMERIEUX SA
  • EP3028202B1 patent drawingFigure 1
  • EP3028202B1 patent drawingFigure 2A
  • EP3028202B1 patent drawingFigure 2B

AI summary

A method for detecting, in a biological sample, at least two microorganisms belonging to two different taxa, represented by intensity vectors Pj obtained by a multidimensional measuring technique, comprises: ■ acquiring a digital signal from the biological sample by means of measurement technology; ■ determining an intensity vector x on the basis of the acquired digital signal; ■ constructing a set {Ŷl} of candidate models Ŷl =(Ŷj, Ŷo)l modelling intensity vector x according to the equation: expression in which: expression in which: o ∀j ∈ [[1, K]], Pj (a)=:Σi K=1aijPi; and o ∀(i,j) ∈ [[1, K]]2, aij is a predefined coefficient; selecting a candidate model Ŷsel from set {Ŷl} according to the equation: Ŷsel= argmin(Cv (Ŷl) + Cc(Ŷl) expression in which: o Cv(Ŷl) is a criterion quantifying a reconstruction error between the intensity vector of biological sample x and the reconstruction of intensity vector xl by a candidate model (Ŷl); and o Cv(Ŷl) is a criterion quantifying the complexity of a candidate model Ŷl; and determining the presence in the biological sample of at least two taxa from when at least two components of vector Ŷj; of Ŷsel are greater than a positive threshold.