WO2019160200A1 - 인-비보 병변 조직 검출을 위한 레이저 분광 기반의 독립 장치 및 방법 - Google Patents
인-비보 병변 조직 검출을 위한 레이저 분광 기반의 독립 장치 및 방법 Download PDFInfo
- Publication number
- WO2019160200A1 WO2019160200A1 PCT/KR2018/007297 KR2018007297W WO2019160200A1 WO 2019160200 A1 WO2019160200 A1 WO 2019160200A1 KR 2018007297 W KR2018007297 W KR 2018007297W WO 2019160200 A1 WO2019160200 A1 WO 2019160200A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- lesion tissue
- light
- laser
- generated
- generated light
- Prior art date
Links
Images
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/0059—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence
- A61B5/0075—Measuring for diagnostic purposes; Identification of persons using light, e.g. diagnosis by transillumination, diascopy, fluorescence by spectroscopy, i.e. measuring spectra, e.g. Raman spectroscopy, infrared absorption spectroscopy
Definitions
- the present invention relates to independent apparatus and methods based on laser spectroscopy for in vivo lesion tissue detection.
- the above-described '087' patent is a technique for detecting lesion tissue based on threshold values for specific components in a wavelength region, and it is not easy to ensure accuracy.
- an independent device based on laser spectroscopy for detecting in-vivo or ex-vivo lesion tissue.
- instruments for use in laser spectroscopy-based independent apparatus for detecting in-vivo or ex-vivo lesion tissue there are provided instruments for use in laser spectroscopy-based independent apparatus for detecting in-vivo or ex-vivo lesion tissue.
- a machine learning based lesion tissue detection method is provided.
- the spectrometer for measuring the spectrum of the generated light generated by the laser irradiation on the sample; And a disease analysis module configured to determine whether there is lesion tissue by applying a learning model for detecting lesion tissue to the non-discrete spectrum measurement result measured by the spectrometer, wherein the spectroscope comprises a moment at which the laser is irradiated onto the sample.
- a disease analysis module configured to determine whether there is lesion tissue by applying a learning model for detecting lesion tissue to the non-discrete spectrum measurement result measured by the spectrometer, wherein the spectroscope comprises a moment at which the laser is irradiated onto the sample.
- Machine learning-based lesion tissue detection method includes.
- Machine learning-based lesion tissue detection method includes.
- lesions can be detected more accurately through non-discrete spectrum measurement and machine learning based lesion tissue detection methods.
- present embodiments can support both in-vivo lesion detection as well as ex-vivo lesion detection.
- FIGS. 1 to 3 are diagrams for explaining a machine learning-based lesion tissue detection method according to an embodiment of the present invention.
- FIG. 4 is a view for explaining a laser spectroscopy-based independent device for detecting lesion tissue according to an embodiment of the present invention.
- FIG. 5 is a diagram for describing non-discrete spectral measurement according to an embodiment of the present invention.
- FIG. 6 shows an example of a laser spectroscopy-based independent device for detecting in-vivo lesion tissue described with reference to FIG. 4.
- 7 to 9 are diagrams for describing non-discrete spectrum measurement.
- ... unit refers to a unit for processing at least one function or operation, which may be implemented by hardware or software or a combination of hardware and software. .
- 'program' or 'algorithm' means 'a set of instructions suitable for computer processing,' and 'program' and 'algorithm' are used interchangeably.
- a program (or algorithm) performs (or executes) an operation (or step) means “an operation (or step) performed by an electronic device whose program (or algorithm) has a processor. To perform or to execute ”.
- the term 'laser' means a pulsed laser or continuous light laser.
- the frequency band of the 'laser' may have any frequency band, for example, may have a UV (ultra violet) band, a visible light band, or an IR-infrared band.
- 'generated light' is meant to include all light generated when the laser is irradiated to the sample (T).
- 'generated light' may mean, for example, plasma light, reflected light, scattered light, and / or fluorescent light.
- the term 'sample' means biological tissue, and may be, for example, human tissue or animal tissue.
- the term 'measurement data' means 'spectral data measured by a spectrometer for N0N-GATED generated light generated when a laser is irradiated to a sample', and 'non-discrete spectrum measurement' Data measured by '
- non-discrete spectrum measurement data or non-discrete spectrum data is a concept including filtered non-discrete spectrum data.
- the term 'non-discrete spectrum measurement' refers to all generated light measured over time, from the moment the laser is irradiated to the sample surface until the generated light is no longer generated, i.e. Measuring the spectrum of the NON-GATED generated light. That is, the result of 'non-discrete spectrum measurement' is not discrete in the wavelength band, that is, has a continuous value. Meanwhile, 'non-discrete spectrum measurement' is a concept including 'filtered non-discrete spectrum measurement'.
- the term 'Filtered Non-Discrete Spectrum Measurement refers to 'spectral measurement of some light in NON-GATED generated light, or to light belonging to a specific wavelength band in NON-GATED generated light. It means to measure the spectrum with respect to.
- 'Filtered Non-Discrete Spectrum' refers to data obtained as a result of 'Filtered Non-Discrete Spectrum Measurement'.
- the term 'parameter of the feature extractor' will be used as referring to the parameters constituting the feature extractor (eg, the weight of the feature function and the main component).
- Machine learning-based lesion tissue detection method includes a pre-processing step (hereinafter referred to as 'first step') and the determination step (hereinafter referred to as 'second step').
- the first step is to normalize, standardize, and normalize and normalize the results of non-discrete spectrum measurement (NSM) for NON-GATED generated light.
- normalization and standardization are steps of removing deviations and noise between measurement results.
- normalization and standardization may include area-normalization and interpolation operations after removing background noise from measurement results.
- Machine learning-based lesion tissue detection method may further comprise a non-discrete spectrum measurement (NSM) step for the NON-GATED generated light.
- NSM Non-Discrete Spectrum Measurement
- the Non-Discrete Spectrum Measurement (NSM) step consists of all the generated light measured over time, from the moment the laser is irradiated to the sample surface until the generated light is no longer generated, ie NON- In this step, the spectrum of the generated light is measured.
- the non-discrete spectrum measurement (NSM) step measures the spectrum for some light in the NON-GATED generated light, or for the light belonging to a specific wavelength band in the NON-GATED generated light.
- the method may be a step of measuring a filtered non-discrete spectrum measuring a spectrum.
- 'Non-discrete spectrum measurement result' means the data as 'Non-discrete spectrum measurement' as it is or with respect to such data. Means data after normalization and standardization.
- Machine learning-based lesion tissue detection method comprises the steps of irradiating the surface of the sample with a laser of a specific band; And non-discrete spectrum measurement (NSM) for NON-GATED generated light.
- NSM non-discrete spectrum measurement
- the wavelength of the laser irradiated onto the sample surface may be 1064 nm.
- the Non-Discrete Spectrum Measurement (NSM) step may measure the spectrum of some light in NON-GATED generated light, or may measure the spectrum of light belonging to a specific wavelength band in NON-GATED generated light.
- the method may be a step of measuring a filtered non-discrete spectrum.
- the wavelength of the generated light for which the spectrum is measured may be, for example, having a 200nm ⁇ 1000nm band.
- Principal Component Analysis is a step of extracting a feature for a principal component from non-discrete spectral measurement results.
- there are a plurality of principal components or also referred to as 'multi-dimensions'
- the principal component analysis step extracts features (or also referred to as 'feature values') for each of the plurality of principal components It includes an operation to do.
- the principal component analyzing step may extract each feature of the principal components in 16 dimensions.
- the 16th dimension is an exemplary value, the present invention is not limited to such a numerical value, and the main component analysis step may extract features for main components having more dimensions than the 16th dimension.
- 'extracting a feature for each of a plurality of principal components' will be referred to as 'multidimensional principal component analysis'.
- the first step described above is an electronic device including a memory (not shown), one or more processors (not shown), and one or more programs (not shown). It can be performed by).
- one or more programs (hereinafter, 'preprocessing programs') are stored in the memory and configured to be executed by the one or more processors.
- the programs for preprocessing include instructions for performing the above-described normalization and standardization and principal component analysis (PCA) steps.
- the preprocessing programs may include a normalization and standardization program and a multidimensional principal component analysis program. These programs each include instructions for performing the first operation.
- the memory (not shown), one or more processors (not shown), and preprocessing programs (not shown) for the first step may be described with reference to FIG. 4. It may be located in the analysis device 10 to be described later.
- some of the memory (not shown), one or more processors (not shown), and preprocessing programs (not shown) for the first step may be described with reference to FIG. 4.
- the rest Located in the analysis device 10 to be described later, the rest may be located in the handpiece 20.
- some of the memory (not shown), one or more processors (not shown), and preprocessing programs (not shown) for the first step may refer to FIG. 4. It may be located in the analysis device 10 to be described later, the remaining part may be located in an electronic device connected to the analysis device 10 and a communication network (network that can transmit or receive data, for example, the Internet).
- a communication network network that can transmit or receive data, for example, the Internet
- the second step is a step in which the machine learning algorithm applies a classifier, a learning model for lesion tissue detection, to the result of multidimensional principal component analysis in the first step to determine whether or not the sample has lesion tissue.
- the learning model for lesion tissue detection is a classifier generated by training (or learning) with labeled non-discrete spectral metrology data.
- the non-discrete spectrum measurement data may be Filtered Non-discrete spectrum data.
- the machine learning algorithm is a deep learning algorithm configured to include an input layer, at least one hidden layer, and an output layer.
- the hidden layer may reflect the functions and coefficients constituting the learning model for lesion tissue detection.
- the input layer receives the results of the preprocessing step, the hidden layer applies a learning model for lesion tissue detection on the data inputted by the input layer, and the output layer outputs the result in the hidden layer.
- the output value of the output layer may be, for example, a value that probably indicates the presence or absence of lesion tissue.
- SVM Space Vector Machine
- DNN Deep Neural Network
- CNN Convolution Neural Network
- the second step described above is an electronic device including a memory (not shown), one or more processors (not shown), and one or more programs (not shown). It can be performed by).
- one or more programs (hereinafter, 'machine learning programs') are stored in the memory and configured to be executed by the one or more processors.
- the machine learning program includes instructions for performing the step of determining the presence or absence of the lesion tissue described above.
- the machine learning program also includes instructions for performing a learning step for generating a learning model for lesion tissue detection.
- the memory (not shown), one or more processors (not shown), and the machine learning program (not shown) for the second step will be described later with reference to FIG. 4. It may be located in the analysis device (10).
- some of the memory (not shown), one or more processors (not shown), and machine learning programs (not shown) for the second step will be described later with reference to FIG. 4. May be located in the analytical instrument 10, the remainder being located in the handpiece 20.
- some of the memory (not shown), one or more processors (not shown), and machine learning programs (not shown) for the second step will be described later with reference to FIG. 4. It may be located in an analysis device 10, and the remaining part may be located in an electronic device connected to the analysis device and a communication network (a network capable of transmitting or receiving data, for example, the Internet).
- a communication network a network capable of transmitting or receiving data, for example, the Internet.
- the step of determining the parameter of the feature extractor from the labeled non-discrete spectrum measurement data and the classifier (classifier) which is a learning model for lesion tissue detection may further include defining by learning.
- the non-discrete spectrum measurement data may be, for example, filtered non-discrete spectrum data.
- Determining the parameter of the feature extractor described above is a step in which the feature extractor receives labeled non-discrete spectral measurement data and determines a parameter of the feature extractor.
- the step of defining the classifier described above by learning is the step in which the machine learning program is learned by labeled non-discrete spectral measurement data.
- a classifier generated by learning from labeled non-discrete spectral measurement data may perform a step of determining whether lesion tissue exists from unknown non-discrete spectral measurement data.
- FIGS. 1 to 3 are diagrams for explaining a machine learning-based lesion tissue detection method according to an embodiment of the present invention.
- Machine learning-based lesion tissue detection method to be described with reference to these drawings is an embodiment using a deep learning algorithm, the figures or functions mentioned below are illustrative and the scope of the present invention is limited to those figures or functions only. Those skilled in the art should know.
- the feature extractor 200 is a program for extracting features of principal components from non-discrete spectral measurement data ('multidimensional principal component analysis program'), and the learning model 400 for lesion tissue detection is Classifier to determine whether a lesion is present.
- the non-discrete spectrum measurement data may be, for example, filtered non-discrete spectrum data.
- the feature extractor 200 extracts each feature for a plurality of main components.
- the feature extractor 200 receives non-discrete spectrum measurement data as shown in FIG. 1A and extracts a feature for each of the five main components from the measurement data.
- the feature extractor 200 calculates the intensities y1, y2, y3, y4, y5, ... of the respective signals of the components lambda 1, lambda 2, lambda 3, lambda 4, lambda 5, ... in the wavelength band.
- the values of the functions f1, f2, f3, f4, and f5 are calculated.
- the functions f1, f2, f3, f4, f5 calculated in the feature extractor 200 may have different coefficients (weights) for the inputs.
- f1 (y1, y2, y3, y4, y5,%) (a1 * y1) + (b1 * y2) + (c1 * y3) + (d1 * y4) + (e1 * y5) +.
- F2 (y1, y2, y3, y4, y5,...) (a2 * y1) + (b2 * y2) + (c2 * y3) + (d2 * y4) + (e2 * y5) +.
- A1 and a2 may be different, b1 and b2 may be different, c1 and c2 may be different, d1 and d2 may be different, and / or e1 and e2 may be different.
- the feature extractor 200 outputs a feature calculated by a feature extraction function.
- the feature extractor 200 has five (i.e., five-dimensional) feature extraction functions
- the five feature extraction functions F1, F2, F3, F4, and F5 are each a function. The value is calculated and output, and the outputs are input to the deep learning algorithm 300.
- the deep learning algorithm 300 determines whether there is a lesion tissue by applying the learning model 400 for lesion tissue detection to the received features.
- 2 is a diagram for describing the deep learning algorithm 300.
- the deep learning algorithm 300 receives features from the feature extractor 200 and applies the learning model 400 for lesion tissue detection to the features to determine whether there is a lesion tissue.
- the deep learning algorithm 300 is configured to include an input layer 302, a hidden layer 304 (first hidden layer 303, a second hidden layer 305), and an output layer 306.
- the deep learning algorithm 300 will be described by way of example.
- the input layer 302, the hidden layer 304, and the output layer 306 each consist of one or more nodes, each of which receives a plurality of inputs, the same number of coefficients as the number of inputs' Weights'). That is, the node calculates a predetermined coefficient for each of the inputs received by the node.
- the weight computed for inputs at a node is defined differently for each node.
- the input layer 302 is composed of five nodes, and these five nodes (hereinafter referred to as 'input nodes') receive five features extracted from the feature extractor 200, respectively, and are hidden. Output to layer 304.
- IN1 may be the input feature F1 as it is, or may be a value of any function used as the F1 input.
- the first hidden layer 303 is composed of a plurality of nodes (hereinafter, 'first hidden nodes'), and each of the first hidden nodes calculates and outputs a function using the output values of the input nodes as inputs.
- the plurality of first hidden nodes is m (where m is a positive integer) and h11, h12, h13, h14,. , h1m.
- the function applied at the h11 node of the first hidden node is defined as h11, and the result (or 'output') of the function h11 is defined as H11 .
- the remaining nodes h12, h13, h14,..., H1m of the first hidden node may also be displayed in a similar manner and defined by a formula. That is, h12 of the first hidden node calculates a function h12 that takes IN1, IN2, IN3, IN4, and IN5 as inputs , and outputs the calculated result H12 .
- the remaining first hidden nodes are respectively H13, H14,... And outputs the H1M.
- each of the coefficients included in the functions h11, h12, h13, h14,..., H1m calculated at the first hidden nodes may be at least one function h11, h12, h13, h14. ,..., H1m).
- f11 and f12 may be different, g11 and g12 may be different, j11 and j12 may be different, k11 and k12 may be different, and / or p11 and p12 may be different.
- the function of the h11 node of the first hidden layer 303 is defined to receive IN1, IN2, IN3, IN4, and IN5 , and to calculate a weight corresponding to each of those inputs , and to calculate the weight of the h12 node of the first hidden node.
- the function also receives IN1, IN2, IN3, IN4, and IN5 , and is defined such that weights are computed for each of those inputs, wherein the weights used in the function of the h11 node and the weights used in the function of the h12 node are at least one.
- the weights used in the function of the h11 node and the weights used in the function of the h12 node are at least one.
- at least one or more weights used for the functions included in the first hidden node are defined differently from each other.
- the second hidden layer 305 is composed of a plurality of nodes (hereinafter, 'second hidden nodes'), and each of the second hidden nodes calculates and outputs a function whose output values of the first hidden nodes are variables. do.
- the plurality of second hidden nodes is n (where n is a positive integer), and h21, h22, h23, h24,. , h2n.
- the function applied at the h21 node of the second hidden node is defined as h21, and the result of the function h21 is defined as H21 .
- H21 h21 ( H11, H12, H13, H14, ..., H1m ). That is, the h21 node of the second hidden node is H11, H12, H13, H14,... Calculating a function h21 for the, H1m as input variable, and outputs the calculation result H21.
- the remaining nodes h22, h23, h24,..., H2m of the second hidden node may be displayed in a similar manner and defined by a formula. That is, h22 of the second hidden node is H11, H12, H13, H14,... , Calculates a function h22 that takes H1m as an input variable, and outputs the calculated result H22 .
- the remaining second hidden nodes are respectively H23, H24,... , And it outputs a H2N.
- each of the coefficients included in the functions h21, h22, h23, h24,..., H2n calculated at the second hidden nodes are at least one function h21, h22, h23, h24. ,..., H2n).
- it is defined as + (p22 * HIM )
- f21 and f22 are different, g21 and g22 are different, j21 and j22 are different, k21 and k22 are different,.
- And / or p21 and p22 may be different.
- the function of the h21 node of the second hidden node is H11, H12, H13, H14,... , H1N are defined as variables, weighted to each of those variables, and the function of the h22 node of the second hidden node is also H11, H12, H13, H14,... , H1N is input as variables, and weights are respectively calculated for those variables, except that weights used in the function of the node h21 and weights used in the function of the node h22 are defined differently from each other. .
- the weights used for the functions included in the second hidden node are defined at least one differently.
- the output layer 306 is composed of one node (hereinafter, 'output node'), and the output node calculates and outputs a predefined function that takes as input the output values of the second hidden node.
- the functions and operations calculated at the output nodes are shown by way of example.
- the function of the output node as out, and the result of the function out as OUT .
- OUT may be a value that can determine whether the lesion tissue. For example, it may be defined as 0 ⁇ OUT ⁇ 1 ( OUT is a real number) or may be defined as 0 ⁇ OUT ⁇ 100 ( OUT is a percentage value).
- the deep learning algorithm 300 generates a learning model 400 for lesion tissue detection by performing learning on labeled measurement data.
- Each of the hidden nodes has one function (hereinafter referred to as a 'hidden node function'), and the hidden nodes each output an input value of the hidden node function corresponding to the input.
- the function of the hidden node mathematically operates (e.g., multiplies) the inputs and coefficients. Training of the learning model 400 for lesion tissue detection will be described later in detail with reference to FIG. 3.
- the learning model 400 for lesion tissue detection is optimized by learning (or training).
- Learning the training model 400 for detecting lesion tissue refers to a process of optimizing respective coefficients included in the function of each node of the hidden layers. The learning will be described later in detail with reference to FIG. 3.
- the deep learning algorithm 300 directly receives measurement data without applying the feature extractor 200 and applies the learning model 400 for lesion tissue detection to determine whether there is lesion tissue. That is, without using the feature extractor 200, it is determined whether the lesion tissue is present by applying the learning model 400 for lesion tissue detection to all values of the wavelength band.
- the feature extractor 200 may optimally determine the parameters of the feature extractor 200 using the plurality of labeled measurement data, and the deep learning algorithm 300 may determine the plurality of labeled measurement data.
- the plurality of labeled measurement data may be, for example, filtered non-discrete spectrum data.
- a learning model for detecting lesion tissue after being trained will be specifically called a classifier.
- measurement data labeled "cancer” refers to non-discrete spectral measurement data collected from a patient's tissue determined to be cancer by a doctor's diagnosis.
- the feature extractor 200 receives all of the measurement data labeled (for example, a label meaning cancer) (all collected measurement data), so that the measurement data can be effectively classified. 200 determines the parameters of feature extractor 200.
- the feature extractor 200 may determine weights included in a function for extracting a feature as an optimized value.
- the feature extractor 200 may determine not only the weights included in the function for extracting the feature, but also the main component (number of principal components and / or positions of the principal components) as an optimized value.
- the feature extractor 200 determines weights included in a function for extracting a feature as an optimized value, and a main component may be defined by a person (for example, the person implementing the present invention). have. According to this embodiment, the feature extractor 200 may determine the weights included in the function of extracting the feature as an optimized value based on the number of principal components and / or the location of the principal components defined by the present inventor. .
- the deep learning algorithm 300 sequentially receives the labeled data (all collected data) labeled with a label (e.g., a label meaning cancer), and the deep learning algorithm includes a learning model for detecting lesion tissue. Update the coefficients constituting 400).
- a label e.g., a label meaning cancer
- FIG. 4 is a view for explaining an independent apparatus based on laser spectroscopy (hereinafter, referred to as an 'independent apparatus') for detecting lesion tissue according to an embodiment of the present invention.
- an independent apparatus irradiates a laser to a sample T, collects generated light generated from the sample T, and describes the spectrum of the generated generated light.
- the stand apparatus in-vivo in- vivo
- the lesion diagnostic X-VIVO ex-vivo
- the independent device can diagnose a disease such as cancer.
- the present standalone device can diagnose diseases such as skin cancer, and can diagnose not only skin cancer but also other types of cancer.
- the skin cancer may be, for example, squamous cell carcinoma, basal cell carcinoma, or melanoma.
- the independent device guides the laser beam generated by the analysis device 10, the laser generator 11, and the laser generator 11 to be irradiated onto the sample T.
- the analysis device 10 irradiates a laser on a sample T, performs non-discrete spectrum measurement on NON-GATED generated light collected therefrom, and the measurement result. It is an electronic device that determines whether lesion tissue exists by performing the first and second steps.
- the analysis instrument 10 determines from the labeled non-discrete spectral measurement data, determining parameters of a feature extractor that extracts a feature, and from the labeled non-discrete spectral measurement data, Defining a learning model for detecting lesion tissue may be additionally performed.
- the analysis device 10 includes a plurality of electronic devices.
- the analysis device 10 includes a spectrometer 21, a disease analysis module 23, a power supply 25, and a display unit 27.
- the spectrometer 21 irradiates a laser on the sample T and performs non-discrete spectrum measurement on the NON-GATED generated light collected therefrom.
- FIG. 5 is a diagram illustrating a non-discrete spectrum according to an embodiment of the present invention.
- the present invention it is very important to use non-discrete spectrum measurement data.
- the non-discrete spectrum will be described in detail with reference to FIG. 5.
- a pulse laser having a pulse length of several nanoseconds or less (fs to ns) is focused and irradiated onto the surface of the sample, and when the energy is generally 1 GW / cm 2 or more on the surface of the sample, a very small amount of the surface of the sample is irradiated. It is ablation and becomes plasma.
- a "plasma light” (including an electron emission spectrum, a molecular emission spectrum, a continuous emission, etc.) of the atom is generated on the sample surface, which is illustrated in FIG. 5.
- a certain time for example, 1 us
- the initial continuous emission spectrum Is excluded, and the electron emission spectrum of the atom can be obtained mainly.
- the spectrum of NON-GATED generated light irradiated from a laser and collected from the sample is non-discrete spectrum data including all of the electron emission spectrum, the molecular emission spectrum, and the continuous emission spectrum of the atom.
- the spectrometer 21 may measure the spectrum of the NON-GATED generated light.
- the spectrometer 21 is configured to measure the filtered non-discrete spectrum in the NON-GATED generated light.
- the spectrometer 21 when measuring a filtered non-discrete spectrum, measures the spectrum of the remaining light except for reflected light, scattered light, and fluorescent light in the NON-GATED generated light irradiated with a sample irradiated from the laser. That is, the spectrometer 21 is configured to perform a filtering operation for excluding the reflected light, scattered light, and fluorescent light from the NON-GATED generated light and measuring the spectrum of the remaining light, whereby the reflected light, scattered light, and fluorescent light are generated from the generated light. The light is excluded and a non-discrete spectrum is obtained for the remaining generated light.
- the spectrometer 21 is configured to perform a filtering operation for measuring the spectrum for plasma light.
- a non-discrete spectrum for plasma light is obtained.
- the spectrometer 21 is only for the generated light of a specific wavelength band (200 nm to 1000 nm). It is configured to measure the spectrum. By this configuration, a non-discrete spectrum for the generated light in a specific wavelength band (200 nm to 1000 nm) is obtained.
- the disease analysis module 23 sequentially performs a pretreatment step (first step) and a step of determining whether there is lesion tissue (second step) with respect to a non-discrete spectrum measurement result. It is an electronic device which judges whether the lesion tissue exists in the sample T by doing so.
- the disease analysis module 23 determines from the labeled non-discrete spectral metrology data, determining parameters of the feature extractor, and detects lesion tissue from the labeled non-discrete spectral metrology data.
- the step of defining a learning model can be additionally performed.
- disease analysis module 23 includes a memory (not shown), one or more processors (not shown), and one or more programs.
- the one or more programs include preprocessing programs and / or deep learning programs.
- the pre-processing programs are for performing the pre-treatment step, and include a multidimensional principal component analysis program, and the deep learning program performs an operation of determining whether the lesion tissue is present. Please refer to the above section for more details on these programs.
- the power source 25 supplies power to the laser generator 11 and the disease analysis module 23.
- the display unit 27 outputs the analysis result by the disease analysis module 23 in a form that the user can recognize by visual and / or auditory hearing.
- the laser generator 11 may be used generally, for example, for skin treatment.
- the pulsed laser beam by the laser generator 11 When the pulsed laser beam by the laser generator 11 is focused and irradiated on the sample surface, not only the plasma light but also the reflected light, scattered light, or fluorescence emission is generated.
- plasma light is excluded except for reflected light, scattered light, and fluorescence emission. It is desirable to obtain a non-discrete spectrum only for
- the laser generator 10 irradiates a surface of a sample with a laser having a wavelength of a specific band, and the spectrum is measured by only the generated light belonging to the wavelength of a specific band among the generated light generated therefrom. do.
- a laser generator called Q-switched Nd: YAG laser produces 1064 nm and 532 nm wavelength lasers.
- the laser of this wavelength is irradiated on the surface of the sample, 200 nm ⁇
- Light with a wavelength band of 1000 nm was generated when the tissue on the surface of the sample broke down. That is, the spectra measured for the generated light having a wavelength band of 200 nm to 1000 nm are both electron emission spectra and molecular emission spectra of atoms.
- a laser called a Q-switched Nd: YAG laser is irradiated onto the sample surface, and the spectrometer 21 measures the spectrum only for light in the wavelength band of 200 nm to 1000 nm (1 um) among the generated light generated at that time. Non-discrete spectrum can be obtained.
- the wavelengths of the reflected light, the scattered light, and the fluorescent light remain at 1064 nm with little change in the wavelength of the source laser, and inelastic scattering corresponding to only a fraction of the actual scattered light
- the change in wavelength occurs at, but the signal strength is very weak compared to elastic scattering and negligible. Therefore, the wavelengths of reflected light, scattered light, and fluorescent light (1064 nm) of the generated light can be completely separated from the wavelength band (200 to 1000 nm) at which the electron emission spectrum and the molecular emission spectrum of the atom to be measured are emitted.
- the spectrometer 21 when the laser generating device 11 generates a laser having a wavelength of 1064 nm and irradiates the sample surface, and the spectrometer 21 no longer generates plasma light from the moment when the laser is irradiated on the sample surface. It is designed to measure light over time as non-gated, but to measure non-discrete spectrum only for the wavelength band between 200 nm and 1000 nm. In this case, the spectrometer 21 can remove the reflected light, the scattered light, and the fluorescent light, and can effectively acquire the electron emission spectrum and the molecular emission spectrum of a significant atom.
- the first optical elements 13 may comprise a medium and / or optical elements (eg a lens) for guiding the light and for adjusting the focus of the light such that the laser is irradiated onto the sample T.
- a medium and / or optical elements eg a lens
- the second optical elements 15 may be media and / or optical elements (eg, for collecting and guiding the generated light and adjusting the focus of the light, for collecting the generated light and guiding the collected generated light to the cable 31).
- it may include a lens).
- the cable 31 may comprise a light transmission medium that provides a path through which the generated light collected by the handpiece can be moved to the analysis instrument.
- the optical transmission medium may be composed of, for example, optical fibers.
- FIG. 6 shows an example of a laser spectroscopy based independent device for detecting in-vivo lesion tissue described with reference to FIG. 4.
- an independent apparatus based on laser spectroscopy for detecting in-vivo lesion tissue includes an analyzer 10, a handpiece 20, and a cable 30.
- the analysis device 10 has a substantially cylindrical case, and inside the analysis device 10, the spectrometer 21, the disease analysis module 23, the power supply 25, and the display unit as described with reference to FIG. 4. 27).
- the handpiece 20 has a shape for holding by hand, and the laser generated by the laser generator 11 and the laser generator 11 is irradiated to the body tissue T inside the handpiece 20.
- the cable 30 may include a wire for providing power and an optical transmission medium.
- the spectrometer 20 performs non-discrete spectrum measurement on the generated light collected by the second optical elements 15, and the disease analysis module 23 is
- the lesion tissue may be detected by the machine learning based lesion tissue detection method as described above.
- the spectrometer 20 measures the filtered non-discrete spectrum on the generated light collected by the second optical elements 15, and the disease analysis module 23 is based on the machine learning as described above.
- the lesion tissue can be detected by the lesion tissue detection method.
- 7 to 9 are diagrams for describing non-discrete spectrum measurement.
- FIG. 7 illustrates a discrete spectrum measurement result according to the prior art
- FIG. 8 illustrates a non-discrete spectrum measurement result according to an embodiment of the present invention
- FIG. 9 compares the results of FIGS. 7 and 8.
- a discrete spectrum measurement result according to the prior art is only non-discrete according to an embodiment of the present invention, compared with being able to measure only the threshold value of the wavelength value corresponding to a specific component It can be seen that the result of non-discrete spectrum measurement is to measure values for all wavelength values.
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Pathology (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Biophysics (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Signal Processing (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
- Investigating Or Analysing Materials By Optical Means (AREA)
Abstract
Description
Claims (20)
- 레이저가 시료에 조사되어 생성되는 발생 광의 스펙트럼을 계측하는 분광기; 및상기 분광기에 의해 계측된 넌-디스크리트 스펙트럼 계측 결과에 병변 조직 검출용 학습 모델을 적용하여 병변 조직이 있는지 여부를 판단하는 질병 분석 모듈;을 포함하며,상기 분광기는 상기 레이저가 시료에 조사되는 순간부터 생성되는 모든 발생 광에 대한 스펙트럼을 계측하는 것인, 레이저 분광 기반의 독립 장치.
- 제1항에 있어서,상기 분광기는 상기 모든 발생 광에서 일부 광에 대하여만 스펙트럼을 계측하는 것인, 레이저 분광 기반의 독립 장치.
- 제2항에 있어서,상기 분광기는 상기 모든 발생 광에서, 반사광, 산란광, 및 형광 광을 제외하고 플라즈마 광에 대하여만 스펙트럼을 계측하는 것인, 레이저 분광 기반의 독립 장치.
- 제1항에 있어서,상기 분광기는 상기 모든 발생 광에서 특정 대역의 발생 광에 대하여 스펙트럼을 계측하는 것인, 레이저 분광 기반의 독립 장치.
- 제4항에 있어서,상기 분광기는 상기 모든 발생 광에서 200 nm~1000nm 대역의 발생 광에 대하여 스펙트럼을 계측하는 것인, 레이저 분광 기반의 독립 장치.
- 제1항 또는 제2항에 있어서,상기 병변 조직 검출용 학습 모델은, 병변 조직이 있는지 여부를 나타내는 레이블이 붙은 넌-디스크리트 스펙트럼 계측 데이터들로부터 학습에 의해 정의된 분류자(classifier)것인, 레이저 분광 기반의 독립 장치.
- 제1항 또는 제2항에 있어서,상기 질병 분석 모듈은상기 넌-디스크리트 스펙트럼 계측 결과에서 주 성분들의 특징을 추출하는 주 성분 분석 동작을 수행하고, 상기 주 성분 분석 결과에서 추출한 상기 주 성분들의 특징에 병변 조직 검출용 학습 모델을 적용하여 병변 조직이 있는지 여부를 판단하는 것인, 레이저 분광 기반의 독립 장치.
- 제1항에 있어서,시료에 조사할 레이저를 생성하는 레이저 생성장치와, 상기 레이저 생성장치가 생성한 레이저를 상기 시료까지 안내하는 제1 광학적 소자들과, 상기 발생 광을 수집하는 제2 광학적 소자들을 구비한 핸드피스;를 더 포함하며,상기 제2 광학적 소자들이 수집한 발생 광은 광 전달 매체에 의해 상기 질병 분석 모듈에게 제공되는 것인, 레이저 분광 기반의 독립 장치.
- 제8항에 있어서,상기 분광기는 상기 모든 발생 광에서 일부 광에 대하여만 스펙트럼을 계측하는 것인, 레이저 분광 기반의 독립 장치.
- 제9항에 있어서,상기 분광기는 상기 모든 발생 광에서, 반사광, 산란광, 및 형광 광을 제외하고 플라즈마 광에 대하여만 스펙트럼을 계측하는 것인, 레이저 분광 기반의 독립 장치.
- 제8항에 있어서,상기 분광기는 상기 모든 발생 광에서 특정 대역의 발생 광에 대하여 스펙트럼을 계측하는 것인, 레이저 분광 기반의 독립 장치.
- 제11항에 있어서,상기 분광기는 상기 모든 발생 광에서 200 nm~1000nm 대역의 발생 광에 대하여 스펙트럼을 계측하는 것인, 레이저 분광 기반의 독립 장치.
- 제12항에 있어서,상기 레이저 발생 장치가 생성하는 레이저의 파장이 1064nm 인 것인, 레이저 분광 기반의 독립 장치.
- 제8항에 있어서,상기 질병 분석 모듈은 16개 이상의 주성분들 각각에 대하여 특징을 추출하고, 이들 16개 이상의 특징들에 대하여 상기 병변 조직 검출용 학습 모델을 적용하여 병변 조직이 있는지 여부를 판단하는 것인, 레이저 분광 기반의 독립 장치.
- 제1항 또는 제2항에 있어서,상기 병변 조직은 피부암을 포함하며, 상기 피부암은 편평세포암(Squamous cell carcinoma), 기저세포암 (Basal Cell Carcinoma), 또는 흑색종 (melanoma)인 것인, 레이저 분광 기반의 독립 장치.
- 레이저가 시료에 조사되어 발생 광이 생성되는 순간부터 발생 광이 더 이상 생성되지 않을 때까지 생성된 모든 발생 광에 대한 스펙트럼을 계측하는 넌-디스크리트 스펙트럼 계측 단계; 및상기 넌-디스크리트 스펙트럼 계측 결과에 병변 조직 검출용 학습 모델을 적용하여 병변 조직이 있는지 여부를 판단하는 단계; 를 포함하는 것인, 머신 러닝 기반 병변 조직 검출 방법.
- 제16항에 있어서,상기 병변 조직 검출용 학습 모델은, 병변 조직이 있는지 여부를 나타내는 레이블이 붙은 넌-디스크리트 스펙트럼 계측 데이터들로부터 학습에 의해 정의된 분류자(classifier)것인, 머신 러닝 기반 병변 조직 검출 방법.
- 제16항 또는 제17항에 있어서,상기 판단하는 단계는,상기 넌-디스크리트 스펙트럼 계측 결과에서 주 성분들의 특징을 추출하는 주 성분 분석 동작을 수행하고, 상기 주 성분 분석 결과에서 추출한 상기 주 성분들의 특징에 병변 조직 검출용 학습 모델을 적용하여 병변 조직이 있는지 여부를 판단하는 것인, 머신 러닝 기반 병변 조직 검출 방법.
- 제18항에 있어서,상기 판단하는 단계는 16개 이상의 주성분들 각각에 대하여 특징을 추출하고, 이들 16개 이상의 특징들에 대하여 상기 병변 조직 검출용 학습 모델을 적용하여 병변 조직이 있는지 여부를 판단하는 것인, 머신 러닝 기반 병변 조직 검출 방법.
- 제16항에 있어서,상기 병변 조직은 피부암을 포함하며, 상기 피부암은 편평세포암(Squamous cell carcinoma), 기저세포암 (Basal Cell Carcinoma), 또는 흑색종 (melanoma)인 것인, 머신 러닝 기반 병변 조직 검출 방법.
Applications Claiming Priority (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
KR10-2018-0018857 | 2018-02-15 | ||
KR20180018857 | 2018-02-15 | ||
US201862647032P | 2018-03-23 | 2018-03-23 | |
US62/647,032 | 2018-03-23 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2019160200A1 true WO2019160200A1 (ko) | 2019-08-22 |
Family
ID=67619511
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/KR2018/007297 WO2019160200A1 (ko) | 2018-02-15 | 2018-06-27 | 인-비보 병변 조직 검출을 위한 레이저 분광 기반의 독립 장치 및 방법 |
Country Status (2)
Country | Link |
---|---|
KR (3) | KR102167946B1 (ko) |
WO (1) | WO2019160200A1 (ko) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009300131A (ja) * | 2008-06-11 | 2009-12-24 | Tokyo Institute Of Technology | 生体組織識別装置および方法 |
KR20150036345A (ko) * | 2012-07-02 | 2015-04-07 | 내셔널 유니버시티 오브 싱가포르 | 광섬유 라만 분광법을 이용하는 내시경검사로 실시간 암 진단과 관련된 방법 |
US20170175169A1 (en) * | 2015-12-18 | 2017-06-22 | Min Lee | Clinical decision support system utilizing deep neural networks for diagnosis of chronic diseases |
KR20170114973A (ko) * | 2016-04-04 | 2017-10-16 | 스페클립스 주식회사 | 레이저 기반의 질병 진단 시스템 |
KR20180011646A (ko) * | 2016-07-25 | 2018-02-02 | 삼성전자주식회사 | 생체 내 물질 추정 장치 및 방법, 단위 스펙트럼 획득 장치 및 웨어러블 기기 |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6135965A (en) * | 1996-12-02 | 2000-10-24 | Board Of Regents, The University Of Texas System | Spectroscopic detection of cervical pre-cancer using radial basis function networks |
US10182757B2 (en) * | 2013-07-22 | 2019-01-22 | The Rockefeller University | System and method for optical detection of skin disease |
-
2018
- 2018-06-26 KR KR1020180073304A patent/KR102167946B1/ko active IP Right Grant
- 2018-06-27 KR KR1020180073687A patent/KR20190098900A/ko not_active Application Discontinuation
- 2018-06-27 WO PCT/KR2018/007297 patent/WO2019160200A1/ko active Application Filing
- 2018-06-27 KR KR1020180073690A patent/KR20190098901A/ko unknown
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009300131A (ja) * | 2008-06-11 | 2009-12-24 | Tokyo Institute Of Technology | 生体組織識別装置および方法 |
KR20150036345A (ko) * | 2012-07-02 | 2015-04-07 | 내셔널 유니버시티 오브 싱가포르 | 광섬유 라만 분광법을 이용하는 내시경검사로 실시간 암 진단과 관련된 방법 |
US20170175169A1 (en) * | 2015-12-18 | 2017-06-22 | Min Lee | Clinical decision support system utilizing deep neural networks for diagnosis of chronic diseases |
KR20170114973A (ko) * | 2016-04-04 | 2017-10-16 | 스페클립스 주식회사 | 레이저 기반의 질병 진단 시스템 |
KR20180011646A (ko) * | 2016-07-25 | 2018-02-02 | 삼성전자주식회사 | 생체 내 물질 추정 장치 및 방법, 단위 스펙트럼 획득 장치 및 웨어러블 기기 |
Also Published As
Publication number | Publication date |
---|---|
KR20190098898A (ko) | 2019-08-23 |
KR20190098901A (ko) | 2019-08-23 |
KR102167946B1 (ko) | 2020-10-20 |
KR20190098900A (ko) | 2019-08-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2020197005A1 (ko) | 레이저 유도 붕괴 스펙트럼 분석을 이용하는 진단 방법 및 이를 수행하는 진단 장치 | |
US11672425B2 (en) | Stand-alone apparatus and methods for in vivo detection of tissue malignancy using laser spectroscopy | |
WO2017175913A1 (ko) | 레이저 조사 장치를 이용한 질병 진단 장치 및 질병 진단용 탈부착 핸드피스 | |
US20160007840A1 (en) | Handheld Device for Identification of Microbiological Constituents in the Middle Ear | |
EP3281578B1 (en) | System for diagnosing diseases on basis of laser | |
WO2018101569A1 (ko) | 방수형 pdms 스캐너를 이용한 광음향 수술중 조직검사 장치 및 그를 이용한 광음향 영상 획득 방법 | |
JPH03139346A (ja) | 外科用レーザ装置 | |
US20200237421A1 (en) | Device for Tissue Coagulation | |
WO2016153132A1 (ko) | 피부 측정기기 및 그 제어방법 | |
WO2021107310A1 (ko) | 뇌파 분석 기반 환자진단 지원 시스템 및 방법 | |
US20190246971A1 (en) | Non-discrete spectral analysis algorithms and methods for in vivo detection of tissue malignancy based on laser spectroscopy | |
WO2013066107A1 (ko) | 테라헤르츠파를 이용한 중이염 진단용 프로브, 중이염 진단 시스템 및 방법 | |
WO2017078226A1 (ko) | 조직절제기 및 조직절제시스템 | |
WO2019160200A1 (ko) | 인-비보 병변 조직 검출을 위한 레이저 분광 기반의 독립 장치 및 방법 | |
WO2019160201A1 (ko) | 레이저 분광 기반의 인-비보 병변 조직 검출을 위한 넌-디스크리트 스펙트럼 분석 알고리즘 및 이를 위한 방법 | |
WO2019164375A2 (ko) | 블루라이트에 의한 피부 변화 측정 장치, 방법 및 블루라이트 조사 장치 | |
WO2011037299A1 (en) | Capsule endoscope having biopsy function and method for controlling the same | |
CA2227363A1 (en) | Determination of data concerning a person's auditory capacity | |
WO2017176039A1 (ko) | 색소 병변의 진단 시스템 및 방법 | |
JP2023174715A (ja) | 過熱保護を有するレーザ手術システムのための光学スプリッタ | |
WO2020105954A1 (ko) | 가변파장의 피부 영상 정보를 이용한 휴대용 피부질환 진단장치 | |
WO2019022433A1 (ko) | 타겟에 조사되는 레이저 에너지와 초점 크기를 조절할 수 있는 진단용 레이저 핸드피스용 팁 | |
WO2024101531A1 (ko) | 프로브에 카메라가 탑재된 레이저 진단 기기 | |
WO2017176037A2 (ko) | 레이저 기반의 질병 진단 시스템 | |
WO2018139855A1 (ko) | 색소 병변의 진단 장치 및 방법 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 18906110 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 18906110 Country of ref document: EP Kind code of ref document: A1 |
|
32PN | Ep: public notification in the ep bulletin as address of the adressee cannot be established |
Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 12.02.2021) |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 18906110 Country of ref document: EP Kind code of ref document: A1 |