EP3619618A1 - Procédé de configuration assistée par ordinateur d'un modèle fondé sur données en fonction de données d'apprentissage - Google Patents

Procédé de configuration assistée par ordinateur d'un modèle fondé sur données en fonction de données d'apprentissage

Info

Publication number
EP3619618A1
EP3619618A1 EP18734101.1A EP18734101A EP3619618A1 EP 3619618 A1 EP3619618 A1 EP 3619618A1 EP 18734101 A EP18734101 A EP 18734101A EP 3619618 A1 EP3619618 A1 EP 3619618A1
Authority
EP
European Patent Office
Prior art keywords
data
measurement series
measurement
measured values
target
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
EP18734101.1A
Other languages
German (de)
English (en)
Inventor
Thomas Engel
Alexander Michael Gigler
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Siemens AG
Original Assignee
Siemens AG
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Siemens AG filed Critical Siemens AG
Publication of EP3619618A1 publication Critical patent/EP3619618A1/fr
Withdrawn legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01J—MEASUREMENT OF INTENSITY, VELOCITY, SPECTRAL CONTENT, POLARISATION, PHASE OR PULSE CHARACTERISTICS OF INFRARED, VISIBLE OR ULTRAVIOLET LIGHT; COLORIMETRY; RADIATION PYROMETRY
    • G01J3/00—Spectrometry; Spectrophotometry; Monochromators; Measuring colours
    • G01J3/28—Investigating the spectrum
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology

Definitions

  • the invention relates to a method for computer-aided configuration of a data-driven model based on training data.
  • the invention can be used in a variety of technical applications.
  • the invention may be used in the medical arts to assist a physician in classifying tissue types.
  • spectroscopic methods are used in many areas of chemistry, food chemistry and biochemistry as well as biology and medicine for the determination or classification of substances.
  • the methods are based on the fact that recorded optical spectra are interpreted as characteristic fingerprints of the (bio) chemical composition of the examined sample and used for classification.
  • preliminary knowledge of the typical properties of the classes to be distinguished is often necessary. For example, peaks in known spectral regions are analyzed in order to determine the water and oxygen content and thus to isolate eye-catching test samples.
  • classes are to be distinguished whose biochemical and thus spectral properties are not or only partially known, these methods can not be applied.
  • the object of the invention is therefore to provide a method for computer-aided configuration of a data-driven model based on training data, which leads to a data-driven model with high prediction accuracy even with a small amount of training data.
  • the inventive method is used for computer-aided configuration of a data-driven model based on training data.
  • the training data includes a plurality of Records.
  • a respective data record contains a measurement series with a plurality of first input variables with assigned first measured values and a target vector belonging to the measurement series from one or more target variables with associated target values.
  • the first input quantities follow one another based on a predetermined order.
  • This predetermined order thus reflects an order of the input variables and the correlated measured values.
  • the predetermined order may be given by physical parameters, such as energies or wavelengths, where the order of the direction corresponds to larger or smaller values of the physical parameters.
  • the predetermined order may be represented by the time, which is equivalent to the fact that a respective measurement series represents a time series of successive measured values.
  • the measurement series of the training data are each subjected to preprocessing, whereby modified training data are obtained from modified data records.
  • a respective modified data record comprises a preprocessed measurement series and the same target vector as the measurement series without preprocessing.
  • a preprocessed measurement series includes a plurality of second input variables with associated second measured values, which are determined based on the first measured values.
  • a binning step is performed, in which first measured values of first and, in particular, adjacent first input variables of the respective measurement series are combined into measured value sections with assigned section values as a function of one or more measurement characteristics.
  • the measured value characteristic or characteristics were present in the respective measurements, from which the respective first measured values were obtained.
  • the binning step is carried out directly in the space of the first measured values. Nevertheless, there is also the possibility that the binning Step is applied to values that result from a conversion from the first measured values.
  • the number of first (preferably adjacent) input variables is determined, which are combined into measured value sections. This number can vary over the measurement series.
  • the offset of the individual measured value sections can also be suitably determined during the Binnings.
  • the above-defined plurality of second input variables corresponds to the measured value sections and the second measured values represent the section values. Nevertheless, after the binning step within the preprocessing, a further preprocessing step can be performed. In this case, the second input quantities and second measured values may be different sizes than the measured value sections and section values.
  • first input variables can also be combined into measured value sections.
  • isolated first input variables may not be correlated with further first input variables.
  • These isolated input variables are processed within the scope of the method according to the invention as measured value sections with the corresponding first measured value as the section value.
  • the data-driven model is learned based on the computer based on the modified training data, wherein the learned data-driven model allows the determination of target vectors based on preprocessed measurement series.
  • the method according to the invention is distinguished by the fact that information relating to the acquisition of the measured data is suitably taken into account in order to obtain information on a physical basis. to achieve the adaptation of the measurement series used as training data and thus to increase the information content of the measurement data. In this way, a good quality of the learned data-driven model can be ensured even in the case of a limited number of training data sets.
  • the measurement characteristic (s) comprise the noise in the respective measurements
  • the binning step combines first measurement values of first and in particular adjacent first input variables of the respective measurement series in such a way that the average signal-to-noise ratio over the measured value sections is maximized, it being preferably taken into account as a secondary condition of the maximization that a predetermined characteristic signal form, eg in the form of peak widths in waveforms.
  • a predetermined characteristic signal form eg in the form of peak widths in waveforms.
  • the pre-processing in step a) of the method according to the invention may also comprise one or more further preprocessing steps.
  • the pre-processing further comprises a rescaling and / or normalization of the first measured values or of values derived therefrom.
  • the section values of the above-defined measured value sections are determined by averaging over the first measured values of the respective measured-value sections.
  • the section values represent an average of these first measured values.
  • any desired mean value can be determined.
  • it may be a weighted or unweighted averaging.
  • the mean value determined is the arithmetic mean value.
  • the median can also be determined as an average value or a Gaussian profile can be averaged.
  • the first derivative and / or the second derivative is determined according to an order parameter characterizing the predetermined order as part of the preprocessing in step a), wherein the first derivative and / or the second derivative are used as second measured values of second input variables in the preprocessed measurement series.
  • the order parameter can be set differently.
  • the order parameter may be e.g. the wavelength or a quantity derived therefrom, e.g. the energy, his.
  • the order parameter can represent the time, if the measurement series is a time series.
  • the data-driven model used in the method according to the invention is a neural network comprising an input layer, one or more hidden layers and an output layer, wherein the input layer receives as input pre-processed measurement series and based on these inputs generates outputs in the form of corresponding target vectors ,
  • CNN Convolutional Neural Network
  • CNN Convolutional Neural Network
  • Such networks are known per se and contain, as a hidden layer, at least one convolution layer which has a linear transformation generates a so-called feature map.
  • at least one pooling layer is provided, which reduces the dimension of the features of the feature map.
  • the CNN network is a deep CNN network, also referred to as the Deep Convolutional Neural Network (DCNN), comprising a plurality of hidden layers
  • the invention is not limited to the use of neural networks as data driven models.
  • the data-driven model may also include a support vector machine and / or a cluster method (e.g., k-means clustering) and / or a decision tree and / or partial least squares (PLS) regression.
  • a support vector machine and / or a cluster method e.g., k-means clustering
  • PLS partial least squares
  • PLS regression a special variant of a PLSDA
  • PLSDA Partial Least Squares Discriminant Analysis
  • one or more target variables of the respective target vector in each case describe an assignment or non-assignment to a class.
  • the data-driven model is used to classify appropriate measurement series. Nonetheless, a target size may also represent a variable having a plurality of continuous or discrete values.
  • the measurement series of the training data are optical spectra for respective objects, wherein an optical spectrum for a respective object comprises first measurement values representing an absorption or a transmission of electromagnetic radiation for the respective object as a function of spectral values depend on the wavelength of the electromagnetic radiation.
  • the target vector specifies as target variables one or more features of the respective object. jekts.
  • the spectral values directly represent the wavelength of the electromagnetic radiation. Nonetheless, the spectral values can also represent, for example, the energy of the radiation.
  • a respective object is a biological tissue sample of the human or animal body.
  • the target vector preferably comprises a feature which specifies the biological tissue sample as pathological or non-pathological, e.g. as tumorous or non-tumorous. It is also possible that a respective object is an organic sample.
  • the target vector preferably comprises one or more features which specify the type of organic sample and / or its aging state.
  • the measurement characteristic (s) considered in the binning step represent the spectral resolution of the optical spectra as a function of the spectral values.
  • the Binning step monotonically with increasing spectral resolution. If the spectral values represent wavelengths, the spectral resolution toward higher wavelengths is generally greater, in which case the number of combined first measured values decreases towards higher wavelengths.
  • the measurement series represent optical spectra. Nevertheless, the invention is not limited thereto and the measurement series may also represent other ordered measurement data. As already mentioned above, the measurement series can also be time series, which comprise first measurement values which were obtained from measurements at different times.
  • the invention also relates to a method for computer-aided determination of a target vector based on a measurement series having a plurality of first input variables and associated first measured values, wherein the target vector comprises one or more target variables with target values to be determined and the first input quantities follow one another based on a predetermined order.
  • a learned data-driven model is provided, which is learned with the method according to the invention or one or more preferred embodiments of the method according to the invention.
  • the considered measurement series is subjected to the same preprocessing as a respective measurement series of the training data in step a) of the method with which the data-driven model is learned, whereby a preprocessed measurement series is obtained.
  • the invention further relates to a computer program product with a program code stored on a machine-readable carrier for carrying out the method according to the invention for the computer-aided configuration of a data-driven model or the method according to the invention for the computer-aided determination of a target vector or for carrying out one or more preferred variants this procedure.
  • the invention relates to a computer program with a program code for carrying out the method according to the invention for the computer-aided configuration of a data-driven model or the method according to the invention for computer-aided determination of a target vector or one or more preferred variants of these methods.
  • Fig. 1 is a schematic representation of a neural network which is used in a variant of the invention as a data-driven model
  • FIG. 2 shows a flowchart which illustrates the steps of an embodiment of the method according to the invention.
  • a variant of the invention will be described below with reference to the computer-aided configuration of a data-driven model in the form of a neural network.
  • a neural network NN is shown schematically in FIG.
  • this network is a DCNN network which has already been mentioned above.
  • the network NN comprises an input layer IL, several hidden layers HL1, HL2, HLn and an output layer OL.
  • the neural network is used to determine or forecast a target vector from one or more features based on a measurement series of measured values. This target vector is output via the output layer OL.
  • measurement series are considered which relate to optical spectra of tissue samples.
  • the input layer IL of the neural network NN is not directly supplied with the measurement series as input variables, but they are subjected to preprocessing in advance in order to appropriately take into account prior knowledge of the acquisition of the individual measured values.
  • This preprocessing is also applied when learning the neural network based on corresponding training data.
  • the training data represent data sets from a plurality of measurement series in the form of the abovementioned optical spectra, it being known whether the optical spectrum belongs to a tumorous tissue or not.
  • a data set of the training data also includes a target vector which, in the example considered here, represents a classification of the tissue as tumorous or non-tumorous.
  • the method according to the invention can also be used for other types of target vectors, which can also comprise several features, wherein a respective feature can also be represented by the corresponding value of a variable.
  • a respective feature can also be represented by the corresponding value of a variable.
  • ranges of values of a variable characterize different states of materials for which the optical spectra have been detected.
  • the aging state of materials can be categorized here.
  • Fig. 2 shows the learning of the neural network of Fig. 1.
  • training data TD comprising a plurality of data sets DS is used.
  • each data record contains an experimentally determined measurement series MR, which comprises a multiplicity of input variables EG, which are arranged one behind the other based on a predetermined order.
  • Each input variable is correlated with a measured value MW, whereby the input variables of different data sets DS are assigned different measured values MW.
  • measurement series in the form of optical spectra of tissue samples are considered, a corresponding spectrum being indicated in the diagram DI of FIG. 2.
  • the wavelength ⁇ of the light is indicated, which is irradiated to the corresponding tissue sample.
  • the ordinate represents the measured values MW, which in the example considered here correspond to absorption coefficients at the different wavelengths. This results in the measurement series MR in the form of the curve shown.
  • the order of the input quantities over the different wavelengths ⁇ is achieved.
  • the wavelength ⁇ represents an order parameter for describing the order of the input quantities.
  • Each spectrum of a data set DS of the training data TD is correlated with a target vector ZV, which in the embodiment described here comprises a single target size ZG occupied by a corresponding target value ZW.
  • This target value represents the information as to whether the tissue sample of the corresponding data set DS is tumorous or not.
  • the learning of the neural network NN is performed.
  • the training data are not used directly, but they are subjected to preprocessing, which is indicated by step Sl of FIG.
  • the pre-processing can include various sub-steps.
  • BS binning step
  • measured values which have been determined for adjacent wavelengths are combined taking into account measurement characteristics MC relating to the measurement of the respective measured values.
  • the summarized measured values represent measured value sections, to each of which a section value is assigned, which in the example considered here is the mean value of the combined measured values.
  • the noise in the respective measurement is taken into consideration.
  • the number of combined measured values is selected in such a way that the average signal-to-noise ratio is maximized over the combined measured value sections and at the same time the signal shape remains easily recognizable via the resulting averaging.
  • Corresponding solution methods for such an optimization problem are well known to the person skilled in the art and are therefore not explained in detail.
  • the number of measured values which are combined does not remain constant but can vary as a function of the wavelength ⁇ .
  • the measured value characteristic of the spectral resolution of the measured values within the respective optical spectra can also be taken into account in the method of FIG. In doing so, one makes use of the knowledge that the resolution increases towards longer wavelengths, since there are lower energies and can be obtained, for example, in the case of larger wavelengths. In the middle infrared distinguish the characteristic eigenmodes of the chemical groups of the material in fingerprint manner. This results in sharper peaks in the spectrum at longer wavelengths.
  • This measured value characteristic can now be included in preprocessing in such a way that the number of measured values which are combined into measured value sections decreases due to the higher spectral resolution towards higher wavelengths ⁇ .
  • the preprocessing according to the step Sl of FIG. 2 may also comprise further steps in addition to the binning step BS.
  • the measured values can be rescaled or normalized.
  • the measured values or the resulting measured value sections can not be plotted along a wavelength scale, but along an energy scale, which represents a non-linear rescaling.
  • additional measured values can be added, which result from the original measured values with mathematical calculations.
  • the slope (first derivative) or the curvature (second derivative) of the curve of the respective optical spectra can be considered as a further variable.
  • modified training data TD 'consisting of modified data records DS' is contained.
  • the target vector ZV with the target size ZG and the target value ZW remains unchanged for the modified data set DS ', i. the target vector is the same target vector as in the case of the unmodified data set DS.
  • the neural network NN of FIG. 1 is learned.
  • This network receives as input data the preprocessed measurement series MR 'with the modified input quantities EG'.
  • the learning of the neural network is done with methods known per se and is therefore not described in detail.
  • the learned neural network NN is obtained, which is subsequently used in the prediction step PR shown in FIG. 2 in order to correctly determine the corresponding target vector ZVN for a new measurement series MRN with an unknown target vector.
  • the same preprocessing is used, which was applied to the training data during learning of the neural network.
  • a calculated target vector ZVN is obtained, which in the present case indicates whether the optical spectrum belongs to a tissue sample which is tumorous.
  • a suitable multi-dimensional target vector can also be used, which further specifies which type of tissue, such as muscle, blood vessel, tendon, nerve, fat, etc., belongs to the tissue sample.
  • a method for learning a data-driven model in which the observed measurement data are suitably pre-processed taking into account measurement characteristics, whereby a learning adapted to the physical conditions of the measurement can be achieved.
  • the data quality can be improved before the actual learning process, thereby increasing the predictive quality of the learned data-driven models.
  • the quality of learning can be improved even with a small number of training data sets. This is very helpful and advantageous, especially for medical questions, since the access to large amounts of data in training and validation phases is very complex to almost impossible.
  • the inventive method can be used in a variety of applications. As described above, e.g. optical spectra are analyzed. Nevertheless, there is also the possibility that measurement series are processed in the form of time series with successive measured values.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Computational Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Databases & Information Systems (AREA)
  • Algebra (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Probability & Statistics with Applications (AREA)
  • Operations Research (AREA)
  • Investigating Or Analysing Materials By Optical Means (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

L'invention concerne un procédé de configuration assistée par ordinateur d'un modèle fondé sur données (NN) en fonction de données d'apprentissage (TD). Le procédé selon l'invention est caractérisé en ce que des séries de mesures (MR) sont soumises à une étape de compartimentage (BS) comprenant un prétraitement approprié, des caractéristiques de mesure (MC) étant prises en compte, lesquelles, lors de la mesure, présentent des valeurs mesurées (MW) respectives. Un modèle fondé sur données approprié, par exemple, un réseau de neurones artificiels, est ensuite appris en fonction de la série de mesures (MR') pré-traitée. Ledit modèle fondé sur données (NN) appris permet une prédiction précise de vecteurs cibles en fonction de séries de mesures (MR) correspondantes. Le procédé peut être utilisé, par exemple, pour l'analyse des spectres optiques. En particulier, le modèle appris permet de prédire si l'échantillon de tissu, pour lequel un spectre optique a été détecté, représente un tissu pathologique.
EP18734101.1A 2017-07-05 2018-06-07 Procédé de configuration assistée par ordinateur d'un modèle fondé sur données en fonction de données d'apprentissage Withdrawn EP3619618A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
EP17179817.6A EP3425519A1 (fr) 2017-07-05 2017-07-05 Procédé de configuration assistée par ordinateur d'un modèle commandé par des données en fonction de données d'apprentissage
PCT/EP2018/065029 WO2019007626A1 (fr) 2017-07-05 2018-06-07 Procédé de configuration assistée par ordinateur d'un modèle fondé sur données en fonction de données d'apprentissage

Publications (1)

Publication Number Publication Date
EP3619618A1 true EP3619618A1 (fr) 2020-03-11

Family

ID=59298289

Family Applications (2)

Application Number Title Priority Date Filing Date
EP17179817.6A Withdrawn EP3425519A1 (fr) 2017-07-05 2017-07-05 Procédé de configuration assistée par ordinateur d'un modèle commandé par des données en fonction de données d'apprentissage
EP18734101.1A Withdrawn EP3619618A1 (fr) 2017-07-05 2018-06-07 Procédé de configuration assistée par ordinateur d'un modèle fondé sur données en fonction de données d'apprentissage

Family Applications Before (1)

Application Number Title Priority Date Filing Date
EP17179817.6A Withdrawn EP3425519A1 (fr) 2017-07-05 2017-07-05 Procédé de configuration assistée par ordinateur d'un modèle commandé par des données en fonction de données d'apprentissage

Country Status (3)

Country Link
US (1) US11610112B2 (fr)
EP (2) EP3425519A1 (fr)
WO (1) WO2019007626A1 (fr)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP3435295A1 (fr) 2017-07-26 2019-01-30 Siemens Aktiengesellschaft Pré-traitement pour un algorithme de classification
CN109859849B (zh) * 2019-03-01 2022-05-20 南昌大学 一种基于分段人工神经网络的软组织穿刺力建模方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7084973B1 (en) * 2002-06-11 2006-08-01 Dalsa Inc. Variable binning CCD for spectroscopy
US9449283B1 (en) * 2012-08-20 2016-09-20 Context Relevant, Inc. Selecting a training strategy for training a machine learning model
US10963810B2 (en) * 2014-06-30 2021-03-30 Amazon Technologies, Inc. Efficient duplicate detection for machine learning data sets
EP4425506A2 (fr) * 2019-05-22 2024-09-04 Grail, Inc. Systèmes et procédés pour déterminer si un sujet a une pathologie cancéreuse à l'aide d'un apprentissage par transfert

Also Published As

Publication number Publication date
US20200218972A1 (en) 2020-07-09
WO2019007626A1 (fr) 2019-01-10
EP3425519A1 (fr) 2019-01-09
US11610112B2 (en) 2023-03-21

Similar Documents

Publication Publication Date Title
DE60114036T2 (de) Verfahren zur charakterisierung von spektrometern
EP0581023B1 (fr) Procédé pour l'analyse d'un constituant d'un échantillon médical
DE102012205686B3 (de) Verfahren zur Substanzidentifizierung und zugehöriges Computerprogrammprodukt
EP3435295A1 (fr) Pré-traitement pour un algorithme de classification
EP3159681B1 (fr) Procédé et dispositif de détermination pouvant être automatisée de la limite de détection et de l'erreur relative lors de la quantification de la concentration d'une substance à analyser dans un échantillon
DE102018221695A1 (de) Verfahren zur quantitativen Magnetresonanzbildgebung, Magnetresonanzeinrichtung, Computerprogramm und elektronisch lesbarer Datenträger
WO2019007626A1 (fr) Procédé de configuration assistée par ordinateur d'un modèle fondé sur données en fonction de données d'apprentissage
WO2002033400A1 (fr) Procede de caracterisation, d'identification, et de marquage de melanges microbiens par comparaison de donnees d'analyse ir, ir-tf, de raman, de raman-tf, et cg avec des donnees d'une base de donnees de reference
DE69811088T2 (de) Klassifizieranlage und klassifizierverfahren unter verwendung einer kombination von warscheinlichkeitsverfahren und neuronalen netzwerken
DE102013200058B3 (de) Automatisierte Auswertung der Rohdaten eines MR-Spektrums
EP3513355A1 (fr) Procédé de classification de spectres d'objets à contenu d'information complexe
DE3855783T2 (de) Gerät zur Analyse von partiellen molekularen Strukturen
DE112022001991T5 (de) Tumorzelen-isolinien
DE102014224916B4 (de) Verfahren zur rechnergestützten Analyse eines oder mehrerer Gewebeschnitte des menschlichen oder tierischen Körpers
DE102017220103A1 (de) Identifizierung eines oder mehrerer spektraler Merkmale in einem Spektrum einer Probe für eine Inhaltsstoffanalyse
WO2018059662A1 (fr) Sélection de longueur d'onde dans la classification de tissus par imagerie hyperspectrale
DE19549300C1 (de) Verfahren zur rechnergestützten Ermittlung einer Bewertungsvariablen eines Bayesianischen Netzwerkgraphen
DE102014218354B4 (de) Verfahren zur Gewinnung von in einem Ergebnis einer NMR-Messung kodierter Information
DE19608734C1 (de) Verfahren zur Klassifikation einer meßbaren Zeitreihe, die eine vorgebbare Anzahl von Abtastwerten aufweist, insbesondere eines elektrischen Signals, durch einen Rechner und Verwendung des Verfahrens
EP1687756B1 (fr) Procede pour classifier des valeurs de mesures dans des analyses medicales et biochimiques
WO2022242812A1 (fr) Procédé et système mis en œuvre par ordinateur pour déterminer des paramètres de système optimisés d'un système technique à l'aide d'une fonction de coût
DE10317717B4 (de) Verfahren zur Diagnose von Erkrankungen unter Verwendung von Indikatorstoffen
DE10032029A1 (de) IntelllIdent:Auswerteverfahren für die intelligente Identifikation spektralanalytischer Daten
DE102023107378A1 (de) Vorverarbeitung von Spektren in Spektrometrie
WO2018059668A1 (fr) Dispositif d'analyse et procédé pour l'analyse d'un tissu biologique

Legal Events

Date Code Title Description
STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: UNKNOWN

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE

PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE

17P Request for examination filed

Effective date: 20191206

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

AX Request for extension of the european patent

Extension state: BA ME

DAV Request for validation of the european patent (deleted)
DAX Request for extension of the european patent (deleted)
GRAP Despatch of communication of intention to grant a patent

Free format text: ORIGINAL CODE: EPIDOSNIGR1

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: GRANT OF PATENT IS INTENDED

INTG Intention to grant announced

Effective date: 20220110

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN

18D Application deemed to be withdrawn

Effective date: 20220521