Cancer color detection device and method using mid-infrared

Abstract

The present invention relates to a cancer color detection device and method using mid-infrared, and more specifically, to a cancer color detection device and method using mid-infrared, which generates a grouped image based on the main components of the pixel units of the mid-infrared image acquired for any tissue, and compares multiple grouped images with tissue types such as cancer tissue (for example, adenocarcinoma) and normal tissue for the above-mentioned main components and pre-learned tissue learning data, and classifies the organized images according to tissue types such as cancer tissue and general tissue, and reflects the pre-mapped colors on the classified grouped images according to the classified tissue types, generates color index mapping images, and then combines them to generate cancer color shooting images and display them.

Description

Technical Field

[0001] The present invention relates to a cancer color detection device and method using mid-infrared, and more specifically, to a cancer color detection device and method using mid-infrared, which generates a grouped image based on the main components of the pixel units of the mid-infrared image acquired for any tissue, and compares multiple grouped images and tissue types such as cancer tissue (for example, adenocarcinoma) and normal tissue for the above-mentioned main components with pre-learned tissue learning data, thereby classifying the organized images according to tissue types such as cancer tissue and general tissue, and reflecting the pre-mapped colors on the classified grouped images according to the classified tissue types, generating color index mapping images and then combining them to generate cancer color shooting images and display them. Background Art

[0002] The most well-known disease is probably cancer. Cancer is a disease that forms a mass or tumor composed of undifferentiated cells that ignore order and proliferate indefinitely in tissues. Cancer is a general term for a group of diseases that infiltrate and destroy surrounding normal tissues or organs, and can metastasize from the primary lesion to any organ of an individual, forming a new growth site and eventually taking the individual's life.

[0003] Cancer is considered a terrible disease because of its uncontrollable cell proliferation, which destroys the structure and function of tissues and internal organs. Representative types of cancer include stomach cancer, colorectal cancer, lung cancer, breast cancer, liver cancer, prostate cancer, pancreatic cancer, gallbladder cancer, and other bile duct cancers.

[0004] The incidence of this cancer is so high that it ranks first in mortality. Early detection will increase the probability of recovery, so early detection is very important.

[0005] Usually, cancer is confirmed and diagnosed by doctors through infrared images of the corresponding body tissues taken with their eyes.

[0006] Therefore, when it comes to detecting cancer cells, there will be differences depending on the doctor's level of expertise.

[0007] In other words, depending on the doctor's expertise, there is a possibility that cancer may not be detected even if the patient has it, which will delay the discovery of cancer, thereby reducing the patient's chances of recovery and ultimately taking the patient's life.

[0008] Therefore, it is necessary to develop a cancer detection device that can objectively and accurately determine the presence of cancer through captured infrared images.

[0009] Prior art literature

[0010] Patent Literature

[0011] Patent Document 1: Korean Patent No. 10-1384211 (published on April 14, 2014) Summary of the invention

[0012] Problems to be solved by the invention

[0013] Therefore, the purpose of the present invention is to provide a cancer color detection device and method using mid-infrared, which generates a grouped image based on the main components of the pixel units of the mid-infrared image acquired for any tissue, and compares multiple grouped images with tissue types such as cancer tissue (for example, adenocarcinoma) and normal tissue for the above-mentioned main components and pre-learned tissue learning data, classifies the grouped images into tissue types such as cancer tissue and general tissue, and reflects the pre-mapped colors on the classified grouped images according to the classified tissue types, generates a color index mapping image, and then combines them to generate a cancer color capture image and display it.

[0014] Solutions for solving problems

[0015] In order to achieve the above-mentioned object, the adenocarcinoma color detection device using mid-infrared according to the present invention includes: a display unit; an input unit; a mid-infrared shooting unit, which shoots a region of interest of a tissue sample to generate and output a mid-infrared image for the region of interest; a storage unit, which includes a cancer pixel learning DB, and the cancer pixel learning DB stores tissue type learning data corresponding to the main components of the spectrum of the pixel unit; and a control unit, which displays the H&E staining image acquired when the cancer diagnosis activity occurs on the above-mentioned display unit, receives the region of interest through the input unit, controls the above-mentioned mid-infrared shooting unit to acquire the mid-infrared image for the above-mentioned region of interest, and extracts a plurality of main components from the spectrum of the pixel unit of the acquired mid-infrared image with reference to the above-mentioned tissue type learning data for the main components, groups the pixels according to the extracted main components to generate grouped images of each tissue type, and then combines them to generate a combined grouped image, and in the above-mentioned combined group image, assigns and applies any color to the grouped image of the tissue type of cancer tissue to generate a color index mapping image, and displays it on the above-mentioned display unit.

[0016] The apparatus further includes an HE imaging unit that images the tissue sample to generate an H&E staining image and outputs the image to the control unit. When the cancer diagnosis activity occurs, the control unit controls the HE imaging unit to acquire the H&E staining image.

[0017] The control unit includes: a HE staining image acquisition unit, which acquires an H&E staining image through the HE shooting unit and displays it on the display unit when the cancer diagnosis activity occurs; a region of interest setting unit, which receives and sets the region of interest of the H&E staining image displayed on the display unit input by the user through the input unit; a mid-infrared image acquisition unit, which controls the mid-infrared shooting unit to shoot a corresponding region of the tissue sample corresponding to the set region of interest to acquire a mid-infrared image; a feature extraction unit, which extracts a plurality of principal components as features of the mid-infrared image from the spectrum of the pixel unit of the acquired mid-infrared image with reference to the tissue type learning data for the principal component, groups the pixels according to the extracted principal components to generate grouped images of each tissue type and corresponding tissue information corresponding to each grouped image, generates a combined grouped image combining the grouped images, and outputs it together with the tissue information; and a color mapping unit, which receives the tissue information and the combined grouped image from the feature extraction unit, applies a preset color to the grouped image having tissue information corresponding to the cancer tissue in the combined grouped image to generate a color index mapping image, and displays it on the display unit.

[0018] The feature extraction unit includes: a spectrum enumeration unit, which extracts the spectrum of the pixel unit of the mid-infrared image and performs baseline correction on the extracted spectrum; a principal component extraction unit, which extracts the principal component from the spectrum of the pixel unit on which the baseline correction has been performed based on the learning data stored in the cancer pixel learning DB; and a pixel grouping unit, which groups the pixels according to the principal component to generate a grouped image, and combines the grouped images of the respective principal components to generate a combined group image and outputs it.

[0019] The control unit further includes a learning unit configured to update previous learning data by learning the obtained spectra of each tissue type, ie, each main component in pixel units.

[0020] The above-mentioned main component extraction unit extracts lipids, amide I, nucleic acids, and carbohydrates as main components from the spectrum of the above-mentioned pixel unit. The above-mentioned pixel grouping unit regards the pixels of the spectrum with large absorbance of amide I in the spectrum of the pixel unit as pixels of cancer tissue based on the learning data of each tissue type of the above-mentioned cancer pixel learning DB, and determines the grouped image for the corresponding pixels as a group image of cancer.

[0021] The color mapping unit generates the color index mapping image by applying red to the grouped image of cancer in the combined group image.

[0022] In order to achieve the above-mentioned object, the method for color detection of adenocarcinoma using mid-infrared according to the present invention comprises: a HE staining image acquisition step, in which when a cancer diagnosis event occurs, the control unit acquires an H&E staining image through an HE shooting unit and displays it on the display unit; a region of interest setting step, in which the control unit receives and sets a region of interest in the H&E staining image displayed on the display unit and input by a user through an input unit; a mid-infrared image acquisition step, in which the control unit controls the mid-infrared shooting unit to shoot a corresponding region of a tissue sample corresponding to the set region of interest to acquire a mid-infrared image; a feature extraction step, in which the control unit refers to a composition of a cancer pixel learning DB for a principal component tissue type learning data, extracting a plurality of principal components as features of the mid-infrared image from the spectrum of the pixel units of the acquired mid-infrared image, grouping the pixels according to the extracted principal components to generate grouped images of each tissue type and corresponding tissue information corresponding to each grouped image, generating a combined grouped image combining the grouped images, and outputting it together with the tissue information; and a color mapping step, the control unit receiving the tissue information and the combined grouped image from the feature extraction unit, applying a preset color to the grouped image having tissue information corresponding to the cancer tissue in the combined grouped image to generate a color index mapping image, and displaying it on the display unit.

[0023] The feature extraction process includes: a spectrum enumeration step, in which the control unit extracts the spectrum of the pixel unit of the mid-infrared image and performs baseline correction on the extracted spectrum; a principal component extraction step, in which the control unit extracts the principal component from the spectrum of the pixel unit on which the baseline correction has been performed based on the learning data stored in the cancer pixel learning DB; and a pixel grouping step, in which the control unit groups the pixels according to the principal component and generates a grouped image, and combines the grouped images of each principal component to generate a combined group image and outputs it.

[0024] The method further includes: a learning step, wherein the control unit causes the learning unit to learn the obtained spectra of each tissue type, ie, the pixel unit of each main component, to update the previous learning data.

[0025] In the above-mentioned principal component extraction step, the principal component extraction unit extracts lipids, amide I, nucleic acids, and carbohydrates as principal components from the above-mentioned pixel unit spectrum. In the above-mentioned pixel grouping step, the pixel grouping unit regards the pixels of the spectrum with large absorbance of protein amide I in the spectrum of the pixel unit as pixels of cancer tissue based on the learning data of each tissue type of the above-mentioned cancer pixel learning DB, and determines the grouped image for the corresponding pixels as a group image of cancer.

[0026] In the color mapping step, the color mapping unit applies red to the grouped image for cancer in the combined group image to generate the color index mapped image.

[0027] Effects of the Invention

[0028] The present invention utilizes mid-infrared to have the effect of being able to export the protein amide I component that can determine whether or not there is cancer into the mid-infrared image.

[0029] The present invention assigns different colors to the tissue types constituting the tissue sample to be examined and displays them, thereby having the effect of making it easy to confirm with the eyes when cancer cells (cancer tissue) exist in the tissue sample to be examined.

[0030] That is, the present invention has the effect that even ordinary people without professional knowledge can easily determine whether or not cancer tissue exists in a tissue simply by viewing the analyzed cancer color photographic image.

[0031] Furthermore, the present invention uses mid-infrared light for substances (lipids, proteins, nucleic acids, carbohydrates, etc.) that constitute tissues having absorption bands in the mid-infrared region, thereby achieving an effect of obtaining images that are more accurate and easier to distinguish between tissues. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 1 is a diagram showing the structure of a cancer color detection device using mid-infrared according to the present invention.

[0033] Figure 2 2 is a diagram showing an H&E staining image and a mid-infrared image according to an embodiment of the present invention.

[0034] Figure 3 FIG. 1 is a diagram showing a spectral waveform of a mid-infrared image in pixel units and a main component of each wavelength according to an embodiment of the present invention.

[0035] Figure 4 2 is a diagram showing principal component images classified by the number of principal components according to the present invention.

[0036] Figure 5 is a diagram showing grouped images classified by K-clustering according to the present invention.

[0037] Figure 6 is a diagram showing a color index mapping image according to an embodiment of the present invention.

[0038] Figure 7 is a flow chart showing a cancer color detection method using mid-infrared according to the present invention.

[0039] Description of reference numerals:

[0040] 10: Storage unit 11: Cancer pixel learning DB

[0041] 20: Display unit 30: Input unit

[0042] 40: HE imaging unit 50: Mid-infrared imaging unit

[0043] 100: Control unit 110: HE staining image acquisition unit

[0044] 120: Region of interest setting unit 130: Mid-infrared image acquisition unit

[0045] 140: Feature extraction unit 141: Spectrum listing unit

[0046] 142: Main component extraction unit 143: Pixel grouping unit

[0047] 150: Color mapping unit 160: Learning unit. DETAILED DESCRIPTION

[0048] Hereinafter, the structure and operation of the adenocarcinoma color detection device using mid-infrared according to the present invention will be described in detail with reference to the accompanying drawings, and the adenocarcinoma color detection method of the above device will be described.

[0049] Figure 1 is a diagram showing the structure of a cancer color detection device using mid-infrared according to the present invention, Figure 2 is a diagram showing an H&E staining image and a mid-infrared image according to an embodiment of the present invention, Figure 3 is a diagram showing a spectral waveform of a mid-infrared image in pixel units and a main component of each wavelength according to an embodiment of the present invention, Figure 4 is a diagram showing a principal component image classified by the number of principal components according to the present invention, Figure 5 is a diagram showing grouped images classified by K-clustering according to the present invention, Figure 6 is a diagram showing a color index mapping image according to an embodiment of the present invention. Figures 1 to 6 Provide explanation.

[0050] The cancer color detection device using mid-infrared according to the present invention includes a storage unit 10 , a display unit 20 , an input unit 30 , an HE imaging unit 40 , a mid-infrared imaging unit 50 , and a control unit 100 .

[0051] The storage unit 10 includes: a program area for storing a control program for controlling the operation of the cancer color detection device using mid-infrared according to the present invention; a temporary area for temporarily storing data generated when the control program is executed; and a data area for semi-permanently storing data generated when the control program is executed and data required for executing the control program. According to the present invention, a cancer pixel learning DB 11 can be formed in the data area.

[0052] The cancer pixel learning DB 11 stores learning data for light waves in pixel units so as to identify tissue types such as cancer tissue and general tissue based on principal components. The learning data may be data generated by learning methods such as deep learning, support vector machine (SVM), and K-nearest neighbors (KNN).

[0053] like Figure 3 As shown, in the above learning data, the input data may be the wavelength and absorbance of the light wave, and the output data may be the main component corresponding to the wavelength and absorbance of the light wave.

[0054] According to the measurement results of the mid-infrared microscope, the learning data is a grouped image of the classification results calculated by K-means clustering based on the spectral data between 950nm and 1800nm ​​wavelengths of each file based on 480*480 pixels. The output data is the data of each pixel output as a cluster (1, 2, 3, ..., k) by training two learning data. Based on this clustering information, the pixel parts corresponding to each cluster are clustered to reconstruct the image.

[0055] The display unit 20 is controlled by the control unit 100 to display the input / output information as any one or more of text, graphics, static video, and video. According to the present invention, the output information may be an H&E stained image, a mid-infrared image, a principal component image, a grouped image, a color index mapping image, analysis data, etc.

[0056] The input unit 30 includes any one or more of a key input device such as a keyboard capable of inputting multiple functions and texts, a mouse, and a touch pad that is integrated with the screen of the above-mentioned display unit 20 and is used to output position information of the position touched on the screen, so as to provide the user with information input and function selection means.

[0057] The HE imaging unit 40 images the tissue sample and transmits the image to Figure 2The hematoxylin and eosin (H&E or HE) stained image shown in 201 is output to the control unit 100.

[0058] The mid-infrared imaging unit 50 is controlled by the control unit 100 to capture a specific area of ​​the tissue sample, for example, Figure 2 The region of interest (ROI) of the H&E staining image shown in 201 is photographed using mid-infrared, and the Figure 2 The mid-infrared image of the specific region (ROI region) shown in 202 is output to the control unit 100. The mid-infrared image may be an enlarged image corresponding to the specific region of the H&E staining image.

[0059] The control unit 100 includes an HE-stained image acquisition unit 110 , an ROI setting unit 120 , a mid-infrared image acquisition unit 130 , a feature extraction unit 140 , a color mapping unit 150 , and a learning unit 160 , and controls the overall operation of the cancer color detection device according to the present invention.

[0060] Specifically, when a cancer diagnosis activity occurs, the HE staining image acquisition unit 110 controls the HE imaging unit 40 to photograph the tissue sample located at the focus position of the camera of the HE imaging unit 40 , and acquires the H&E staining image of the tissue sample from the HE imaging unit 40 , and displays it on the display unit 20 .

[0061] like Figure 2 As shown in 201 , the region of interest setting unit 120 obtains and sets the region of interest specified by the user through the input unit 30 for the H&E staining image displayed on the display unit 20 .

[0062] The mid-infrared image acquisition unit 130 controls the mid-infrared photographing unit 50 to photograph the region of interest set by the region of interest setting unit 120 , and receives and acquires the photographed mid-infrared image from the mid-infrared photographing unit 50 .

[0063] The feature extraction unit 140 includes a spectrum enumeration unit 141 , a principal component extraction unit 142 and a pixel grouping unit 143 , and extracts group images of each tissue classification from the mid-infrared image, and outputs a combined group image by combining the extracted group images to the color mapping unit 150 .

[0064] Specifically, the spectrum listing unit 141 calculates the pixel unit of the mid-infrared image input from the mid-infrared image acquisition unit 130 as follows: Figure 3 The spectrum shown is subjected to baseline correction (Baseline Correction) and output.

[0065] The main component extraction unit 142 extracts the main component from the spectrum of each pixel that has been baseline corrected. Figure 2 As shown, the above-mentioned main components can be lipids, proteins, nucleic acids, carbohydrates, etc.

[0066] The pixel grouping unit 143 compares the principal components extracted from the principal component extraction unit 142 with the learning data of the tissue types of the principal components pre-stored in the cancer pixel learning DB 11, determines the tissue type of the pixel, and groups the pixels according to the same principal component, that is, groups the pixels classified as the same tissue type, generates a grouped image, provides the tissue type information of the grouped image to the color mapping unit 150, and outputs the combined grouped image combined with the above-mentioned grouped image to the color mapping unit 150.

[0067] The number of the above principal components is at most 426, from which one can selectively apply. Figure 4 As shown in 301 , the number of principal components can be set to 5 to generate a grouped image, or as shown in 302 , all 426 principal components can be used to generate a grouped image.

[0068] Furthermore, the pixel grouping unit 143 reduces noise by using an arbitrary number of principal components and generates a grouped image by applying K-means clustering.

[0069] The pixel grouping unit 143 of the present invention can be Figure 5 As shown in 401, K is set to 10 (10Group), as shown in 402, K is set to 15 (15Group), etc., and K can also be set to 20 (20Group).

[0070] The color mapping unit 150 receives the tissue type information of the grouped images from the pixel grouping unit 143, applies the colors assigned to the tissue types to the grouped images constituting the combined group image, and generates a Figure 6 The color index mapping image shown is displayed on the display unit 20. For example, Figure 6 The black color (Black) in the figure represents general tissue, the blue color (Blue) represents lipid tissue, the green color (Green) represents nucleic acid tissue, and the red color (Red) represents cancer tissue whose protein contains Amide I as the main component.

[0071] The learning unit 160 updates the previous learning data by learning the spectrum of each tissue type, that is, each main component, in units of pixels.

[0072] Figure 7 is a flow chart showing a cancer color detection method using mid-infrared according to the present invention.

[0073] Reference Figure 7 First, the control unit 100 checks whether a cancer diagnosis activity occurs (S111).

[0074] When the cancer diagnosis activity occurs, the control unit 100 acquires an H&E staining image by controlling the HE imaging unit 40 ( S113 ).

[0075] When the H&E staining image is acquired, the control unit 100 displays the H&E staining image on the display unit 20 and requests setting of a region of interest (ROI) ( S115 ).

[0076] When receiving a request to set a region of interest, the control unit 100 monitors whether the region of interest has been set (S117). When the user sets the region of interest, the mid-infrared shooting unit 50 is controlled to shoot the area corresponding to the region of interest of the above sample to obtain a mid-infrared image of the above region of interest (S119).

[0077] When acquiring a mid-infrared image, the control unit 100 extracts the following information from the acquired mid-infrared image: Figure 3 The spectrum of each pixel unit is shown (S121).

[0078] When extracting the spectrum of each pixel unit, the control unit 100 performs baseline correction on the pixel unit spectrum (S123), and performs principal component analysis (PCA) based on the tissue type learning data of each principal component learned and stored in the cancer pixel learning DB 11 to extract the principal component (S125).

[0079] When the principal component is extracted, the control unit 100 performs tissue type learning on the pixel-based principal component of the currently acquired mid-infrared image, and updates the learning data stored in the cancer pixel learning DB 11 ( S127 ).

[0080] When the principal component is extracted, the control unit 100 applies K-means clustering to group pixels according to the principal component distribution to generate a grouped image, and after combining the generated grouped images to generate a combined group image (S129), the color preset for the grouped images is applied to the combined group image to generate the combined group image. Figure 6 The color index mapping image shown is then displayed on the display unit 20 (S131).

[0081] Furthermore, the present invention is not limited to the above typical preferred embodiments. A person skilled in the art can easily understand that various improvements, changes, substitutions or additions can be made to implement the present invention without exceeding the scope of the present invention. As long as the implementation methods implemented by such improvements, changes, substitutions or additions fall within the scope of the attached claims, their technical ideas should also be regarded as belonging to the present invention.

Claims

1. A cancer color detection device using mid-infrared, characterized in that: include: Display unit; Input section; A mid-infrared photographing unit, which photographs a region of interest of a tissue sample to generate and output a mid-infrared image of the region of interest; A storage unit including a cancer pixel learning DB storing tissue type learning data corresponding to a main component of a spectrum of a pixel unit; a control unit that displays an H&E staining image acquired when a cancer diagnosis activity occurs on the display unit, receives a region of interest through an input unit, controls the mid-infrared imaging unit to acquire a mid-infrared image for the region of interest, extracts a plurality of principal components from a spectrum of a pixel unit of the acquired mid-infrared image with reference to the tissue type learning data for the principal component, groups pixels according to the extracted principal components to generate grouped images of respective tissue types, combines and generates a combined grouped image, and assigns and applies an arbitrary color to the grouped image of the tissue type of cancer tissue in the combined group image to generate a color index mapping image, and displays the image on the display unit; and an HE imaging unit that captures the tissue sample to generate an H&E stained image and outputs the image to the control unit; in, The control unit includes: A HE staining image acquisition unit, which, when the cancer diagnosis activity occurs, acquires an H&E staining image through the HE imaging unit and displays it on the display unit; A region of interest setting unit, which receives a region of interest of the H&E staining image displayed on the display unit input by a user through the input unit and sets the region of interest; a mid-infrared image acquisition unit, which controls the mid-infrared shooting unit to shoot a corresponding area of ​​the tissue sample corresponding to the set area of ​​interest to acquire a mid-infrared image; a feature extraction unit, which refers to the tissue type learning data for the principal components, extracts a plurality of principal components as features of the mid-infrared image from the spectrum of the acquired pixel unit of the mid-infrared image, groups pixels according to the extracted principal components to generate grouped images of each tissue type and corresponding tissue information corresponding to each grouped image, generates a combined grouped image combining the grouped images, and outputs the combined grouped image together with the tissue information; and a color mapping unit, which receives the tissue information and the combined grouped images from the feature extraction unit, applies a preset color to the grouped images having tissue information corresponding to the cancerous tissue in the combined grouped images to generate a color index mapping image, and displays the image on the display unit; The feature extraction unit comprises: a spectrum enumeration unit, which extracts a spectrum of a pixel unit of the mid-infrared image and performs a baseline correction on the extracted spectrum; a principal component extraction unit that extracts principal components from the spectrum of pixel units on which baseline correction has been performed based on the learning data stored in the cancer pixel learning DB; and a pixel grouping unit that groups pixels according to the principal component to generate a grouped image, and combines the grouped images of the principal components to generate a combined group image and outputs the combined group image, The control unit also includes: a learning unit, which updates the previous learning data by learning the spectrum of each tissue type, that is, the spectrum of each main component in pixel units, The main component extraction unit extracts lipids, amide I, nucleic acids, and carbohydrates as main components from the spectrum of the pixel unit. The pixel grouping unit regards pixels of a spectrum with a large absorbance of protein amide I in a pixel-by-pixel spectrum as pixels of cancer tissue based on the learning data of each tissue type in the cancer pixel learning DB, and determines a grouped image for the corresponding pixels as a cancer group image.

2. The cancer color detection device using mid-infrared according to claim 1, characterized in that: The color mapping unit generates the color index mapping image by applying red to the grouped image of cancer in the combined group image.

3. A cancer color detection method using mid-infrared, characterized in that: include: HE staining image acquisition step, when a cancer diagnosis event occurs, the control unit acquires the H&E staining image through the HE imaging unit and displays it on the display unit; a region of interest setting step, wherein the control unit receives, through the input unit, a region of interest input by a user in the H&E staining image displayed on the display unit and sets the region of interest; a mid-infrared image acquisition step, wherein the control unit controls the mid-infrared photographing unit to photograph a corresponding area of ​​the tissue sample corresponding to the set area of ​​interest to acquire a mid-infrared image; a feature extraction step, wherein the control unit extracts a plurality of principal components as features of the mid-infrared image from the spectrum of the acquired mid-infrared image in pixel units with reference to the tissue type learning data of the cancer pixel learning DB for the principal components, groups the pixels according to the extracted principal components to generate grouped images of each tissue type and corresponding tissue information corresponding to each grouped image, generates a combined grouped image combining the grouped images, and outputs the combined grouped image together with the tissue information; and In a color mapping step, the control unit receives the tissue information and the combined grouped images from the feature extraction unit, applies a preset color to the grouped images having tissue information corresponding to the cancerous tissue in the combined grouped images to generate a color index mapping image, and displays the image on the display unit. in, The feature extraction process includes: a spectrum enumeration step, wherein the control unit extracts a spectrum of a pixel unit of the mid-infrared image and performs a baseline correction on the extracted spectrum; a principal component extraction step in which the control unit extracts a principal component from the spectrum of the pixel unit on which baseline correction has been performed based on the learning data stored in the cancer pixel learning DB; and a pixel grouping step, wherein the control unit groups pixels according to the principal component to generate a grouped image, and combines the grouped images of the principal components to generate a combined group image and outputs the combined group image, in, The cancer color detection method using mid-infrared further includes: a learning step, wherein the control unit causes the learning unit to learn the obtained spectra of each tissue type, that is, the pixel unit of each main component, to update the previous learning data, in, In the main component extraction step, the main component extraction unit extracts lipids, amide I, nucleic acids, and carbohydrates as main components from the spectrum of the pixel unit. In the pixel grouping step, the pixel grouping unit regards the pixels with the largest absorbance spectrum of protein amide I in the spectrum of pixel units as pixels of cancer tissue based on the learning data of each tissue type in the cancer pixel learning DB, and determines the grouped image for the corresponding pixels as a group image of cancer.

4. The method for color detection of cancer using mid-infrared according to claim 3, characterized in that: In the color mapping step, the color mapping unit applies red to the grouped image of cancer in the combined group image to generate the color index mapped image.

Images