WO2018070760A1

WO2018070760A1 - Device and method for diagnosing breast cancer by using thermal imaging camera

Info

Publication number: WO2018070760A1
Application number: PCT/KR2017/011134
Authority: WO
Inventors: 남윤영; 공영선; 허지영
Original assignee: 순천향대학교 산학협력단
Priority date: 2016-10-10
Filing date: 2017-10-10
Publication date: 2018-04-19
Also published as: KR20180039466A; KR101887760B1

Abstract

The present invention relates to a device and a method for diagnosing breast cancer by using a thermal imaging camera. More particularly, the present invention relates to a device and a method for diagnosing breast cancer by using a thermal imaging camera, the method comprising: recognizing a breast from a thermographic image captured through a thermal imaging camera; acquiring breast cancer thermographic image analysis information from a breast thermographic image of the recognized breast; and comparing the acquired breast cancer thermographic image analysis information with breast cancer thermographic image analysis information of a normal person's breast so as to determine whether breast cancer is present.

Description

Breast cancer diagnosis apparatus and method using thermal imaging camera

The present invention relates to an apparatus and a method for diagnosing breast cancer using a thermal imaging camera. More particularly, the present invention relates to a breast cancer thermal image from a chest thermal image of a recognized chest. The present invention relates to an apparatus and method for diagnosing breast cancer using a thermal imaging camera that obtains analysis information and compares the obtained breast cancer thermal analysis information with breast cancer thermal analysis information about a breast of a normal person to determine whether breast cancer exists.

Breast cancer accounts for about 25% of female cancers worldwide and is one of the most proliferative cancers with high incidence and mortality. According to 2012 data for stage 1 to stage 3 breast cancer, the 1-year survival rate is greater than 97% and for stage 4, 71%. In addition, the 5-year survival rate of patients sharply drops from 97% at the first diagnosis to 15% at the fourth diagnosis.

Changes in survival and prognosis of breast cancer patients are highly dependent on early detection of breast cancer. Therefore, early diagnosis of breast cancer is very important, and various diagnostic methods have been developed for early diagnosis of breast cancer.

Methods for diagnosing breast cancer include self-examination, mammography, breast ultrasound and magnetic resonance imaging (MRI).

Self-checking method is to check the breasts themselves to check whether there is no lump or other abnormalities, there is no cost and there is no risk, but there is a disadvantage that the accuracy is low.

Mammography is to enlarge the specific area of the breast to perform the test to obtain the image of the breast in the compressed state to obtain the image necessary for diagnosis, causing pain in the test, has a problem that the test cost is high.

Breast ultrasound has advantages in that it is relatively inexpensive and does not expose patients to radiation, but it is difficult to find fine lime material.

Magnetic resonance imaging (MRI) can be used to detect cancerous tumors, but there is a problem in that the cost of installing and maintaining the device is significantly high, which causes a problem of increasing the number of medical treatments.

As such, it is required to develop a breast cancer diagnosis apparatus capable of performing more precisely for detecting breast cancer but requiring less installation and maintenance costs for the medical apparatus and lowering the number of medical treatments.

Due to these demands, advanced sensor technology, and chemical conversion processing technology have been developed, and the spread of mobile terminals such as smartphones equipped with thermal imaging cameras, considerable attention has been focused on diagnosing breast cancer using a small, low-cost thermal imaging camera.

Therefore, an object of the present invention is to recognize the breast from the thermal image taken by the thermal imaging camera, to obtain breast cancer thermographic analysis information from the chest thermal image of the recognized chest, the obtained breast cancer thermal analysis information and normal people The present invention provides a device and method for diagnosing breast cancer using a thermal imaging camera to determine breast cancer by comparing breast cancer thermal imaging analysis information on breasts.

Breast cancer diagnostic apparatus using a thermal imaging camera according to the present invention for achieving the above object: a thermal imaging camera including a thermal imaging camera, the body including a patient's chest portion by outputting a thermal image A thermal image obtaining unit; An image preprocessing unit which deletes a background from the thermal image and performs image preprocessing to output a left chest thermal image and a right chest thermal image that recognize a chest in a body from which the background is deleted; And breast cancer thermal image analysis information of a normal person, extracting feature vectors by analyzing the left and right chest thermal images processed before the image, classifying the extracted feature vectors by applying an artificial neural network, and mediating the feature vectors. And a breast cancer determination unit configured to generate breast cancer thermal analysis information using a variable, and compare the generated breast cancer thermal analysis information with the breast cancer thermal analysis information of the normal person to determine whether breast cancer exists in the left and right chests. It features.

The image preprocessor may include an image background remover configured to delete and output a background from the thermal image; And a region of interest setting unit configured to recognize a chest as a region of interest from the thermal image from which the background is deleted, and to output a left chest thermal image and a right chest thermal image.

The image background remover may include: a channel selector configured to select and output a thermal image of a red (R) channel among RGB channels of the thermal image output from the thermal image acquirer; A Gaussian filtering unit configured to output Gaussian filtering of the red channel thermal image; An outline detector detecting an outline of the body from the thermal image of the Gaussian filtered red channel; And a background deletion unit for deleting a background from a thermal image including all of the RGB channels based on the detected contour.

The image background removing unit may further include an outline reinforcing unit which performs a Hough transform on the detected outline to further reinforce the outline.

The ROI setting unit may include a channel selector configured to select and output only a green channel from a thermal image including all RGB channels from which the background is removed; A Gaussian filtering unit which outputs the thermal image of the green channel by Gaussian filtering; An outline detection unit for detecting and outputting an outline of a body including a chest part from a thermal image of the Gaussian filtered green channel; An outline reinforcing unit configured to perform a Hough circle transformation from the body outline detected by the outline detecting unit to detect a circle corresponding to the chest part; And a breast extracting a left chest thermal image and a right chest thermal image from a thermal image including all of the RGB channels based on a region of interest (ROI) detection based on a circle detected by the contour detector. Characterized in that it comprises a detection unit.

The contour reinforcement unit has a minimum radius value, which is a parameter of the Huff circle transformation, is set as an average radius value of a female chest, and a minimum distance value between a center of a left and a right circle defining a distance between chests for detecting a chest. It is set to 50. It is characterized by the above-mentioned.

The breast cancer determination unit may include: a generation matrix characteristic generator for extracting and outputting feature vectors based on the generation matrix for the vertical, horizontal, and two diagonals of the pre-processed left and right image images; And storing breast cancer thermography information of a normal person, generating artificial breast cancer thermography information by applying an artificial neural network to the extracted feature vectors, and comparing breast cancer thermography information of the normal person to compare the left and right breasts. It is characterized in that it comprises a breast cancer analysis unit for determining the presence of breast cancer.

The breast cancer determination unit may further include a histogram analyzer configured to analyze and output a distribution of histograms measured based on feature vectors for respective RGB channels of the left and right chest image images, and the breast cancer analyzer may include left and right images of normal persons. The histogram for the image is further stored, and the histogram of the normal person and the histogram measured based on the RGB channel characteristic information through the histogram analysis unit are primarily used to determine whether the breast cancer is present, and the breast cancer thermal image analysis information By judging whether or not secondary breast cancer caused by both the breast is characterized in that it is finally determined that the breast cancer is present in the breast.

The breast cancer determiner may include energy, entropy, contrast, correlation, homogeneity, and RGB in the vertical, horizontal, and two diagonal directions as the feature vector. Channel-specific means (Mean), variance (Variance), skewness (Skewness), kurtosis (Kurtosis) is characterized by including.

Breast cancer diagnosis method using a thermal imaging camera according to the present invention for achieving the above object: a thermographic image acquisition unit includes a thermal imaging camera, the body including the patient's chest by taking a thermal imaging camera A thermal image obtaining step of outputting a thermal image; An image preprocessing step of performing an image preprocessing unit to remove a background from the thermal image, and to output a left chest thermal image and a right chest thermal image which recognizes a chest in a body from which the background is deleted; And the breast cancer determination unit extracts feature vectors by analyzing the left and right chest thermal images processed before the image, classifies the extracted feature vectors by applying an artificial neural network, and uses the feature vectors as parameters. And a breast cancer determination step of determining whether breast cancer exists in the left and right chests by comparing the generated breast cancer thermal analysis information with the normal cancer breast thermal analysis information.

The image preprocessing step may include: an image background removing step of an image background removing unit deleting and outputting a background from the thermal image; And a region of interest setting step of outputting a left chest thermal image and a right chest thermal image by recognizing a chest, which is a region of interest, from the thermal image image from which the background is removed.

The image background removing may include a channel selecting step of selecting and outputting a red (R) channel among the RGB channels of the thermal image output from the thermal image obtaining unit; A Gaussian filtering step of performing a Gaussian filtering on the red image of the red channel by a Gaussian filtering unit; An outline detection step of detecting an outline of a body from a thermal image of the Gaussian filtered red channel by an outline detector; And a background deleting step of deleting a background from a thermal image including all of the RGB channels based on the detected contour.

The image background removing step may further include an outline enhancement step of performing a Hough transform on the detected outline to further reinforce the outline.

The ROI setting step may include: selecting and outputting only a green channel from a thermal image including all of the RGB channels from which the background is removed; Gaussian filtering to output a thermal image of the green channel by Gaussian filtering; Contour detection step of detecting and outputting the contour of the body including the chest portion from the thermal image of the Gaussian filtered green channel; Contour reinforcement step of performing a Hough circle transformation from the body contour detected by the contour detection unit to perform a circle detection corresponding to the breast portion; And a breast detection step of extracting a left chest thermal image and a right chest thermal image from a thermal image including all of the RGB channels through ROI detection based on a circle detected by a contour detector. It is characterized by.

In the step of strengthening the contour, the contour enhancement unit sets a minimum radius value, which is a parameter of the Huff circle transformation, as an average radius value of the female breast, and between the centers of the left and right circles defining a distance between the breasts for detecting the breasts. Characterized in that the minimum distance value of 50 is set.

The breast cancer determining step may include a histogram analysis step of analyzing and outputting histograms of each of the left and right chest thermal images; A feature information generation step of extracting and outputting feature vectors by analyzing the left and right chest thermal images processed before the image; And storing breast cancer thermography information of a normal person, classifying the extracted feature vectors by applying an artificial neural network, generating breast cancer thermography information using the feature vectors as a parameter, and analyzing breast cancer thermal images of the normal person. And comparing the information with the breast cancer to determine whether breast cancer exists in the left and right breasts.

In the breast cancer analysis step, the breast cancer analyzer further stores histograms of left and right chest thermal images of a normal person, and primarily compares the histogram of the normal person and the histogram measured by the histogram analyzer to determine whether breast cancer is primary. The method further includes a histogram analysis step, and after determining the breast cancer by the histogram analysis and breast cancer by the breast cancer thermal analysis information after determining the breast cancer, the breast cancer is finally included in the breast. It is characterized by judging that it exists.

The present invention can lower the production cost of the breast cancer diagnostic apparatus by diagnosing breast cancer using a thermal imaging camera, and has the effect of lowering the number of medical treatments.

In addition, the present invention uses the thermal imaging camera has the effect that the patient is not exposed to harmful elements such as radiation.

That is, the breast cancer diagnosis apparatus using the thermal imaging camera of the present invention is faster, more economical, and safer than other breast cancer diagnosis methods, and thus can be used in sensitive patients such as pregnant women.

In addition, the present invention is a simple and inexpensive primary breast cancer diagnosis means for the general public because it can be directly diagnosed by the individual even through a mobile terminal such as a smart phone equipped with a thermal imaging camera that is held in one person per person Since it can be provided has the effect of early diagnosis of breast cancer.

1 is a view showing the configuration of a breast cancer diagnosis apparatus using a thermal imaging camera according to the present invention.

2 is a view showing the detailed configuration of the image background removal unit of the breast cancer diagnosis apparatus using a thermal imaging camera according to the present invention.

3 is a diagram illustrating a detailed configuration of an ROI setting unit of a breast cancer diagnosis apparatus using a thermal imaging camera according to the present invention.

4 is a diagram showing the detailed configuration of the breast cancer determination unit of the breast cancer diagnosis apparatus using a thermal imaging camera according to the present invention.

5 is a view showing a thermal image of a normal person and breast cancer patients applied according to an embodiment of the present invention.

6 is a diagram illustrating an original image image and an image image from which a background is removed according to the present invention.

7 is a view illustrating a thermal image for each RGB channel according to an embodiment of the present invention.

FIG. 8 is a view illustrating contour line images for explaining a result of comparing edge detection according to whether Gaussian filtering is performed according to the present invention.

9 is a diagram illustrating a contour image detected when performing a Hough transform according to the present invention.

10 is a view for explaining a method of extracting left and right chest regions separated by breast recognition and ROI detection using Hough transform according to the present invention.

11 is a diagram showing a histogram of a normal person and a breast cancer patient for explaining a breast cancer detection method according to the present invention.

12 is a view for explaining a method for generating a concurrent matrix for explaining a breast cancer detection method according to the present invention.

13 is a diagram illustrating the intensity levels of the co-occurrence matrix generated according to the present invention in different colors.

Hereinafter, a configuration and operation of a breast cancer diagnosis apparatus using a thermal imaging camera according to the present invention will be described with reference to the accompanying drawings, and a breast cancer diagnosis method in the apparatus will be described.

1 is a view showing the configuration of a breast cancer diagnostic apparatus using a thermal imaging camera according to the present invention, Figure 5 is a view showing a thermal image of a normal person and breast cancer patients applied according to an embodiment of the present invention, Figure 6 2 is a diagram illustrating an original image image and an image image from which a background is removed. 5A is an RGB thermal image of a normal person, and (B) is an RGB thermal image of a cancer patient. Hereinafter, a description will be given with reference to FIGS. 1, 5, and 6.

The apparatus for diagnosing breast cancer using a thermal imaging camera according to the present invention includes a thermal image acquisition unit 10, an image preprocessor 100, and a breast cancer determination unit 400.

The thermal image acquisition unit 10 may include a thermal imaging camera (not shown) to photograph a body part including a chest part of a patient's body and output an RGB thermal image of the body including the chest part to an image preprocessor ( 100). The RGB thermal image may refer to a thermal image including all of red (R), green (G), and blue (B) as shown in FIG. 5. A thermal image including only red is hereinafter referred to as a red channel thermal image, and a thermal image including only green is hereinafter referred to as a green channel thermal image. In addition, the RGB thermal image may further include a gray channel in addition to the red channel, the green channel, and the blue channel.

As shown in FIG. 5, the thermal image obtaining unit 10 outputs an RGB thermal image as shown in FIG. 5A when a patient who has undergone thermal imaging is normal, and when the patient has breast cancer. It will output an RGB thermal image as shown in b). However, it may not be possible to determine whether breast cancer is only shown in FIG. 4. However, in general, the RGB thermal image of a normal person has a uniform heat distribution, but the RGB thermal image of a breast cancer patient shows that the heat distribution of the right chest with a tumor is drastically different from other areas.

The image preprocessing unit 100 receives an RGB thermal image as shown in FIG. 5A from the thermal image acquisition unit 10, and in FIG. 6 (B) as shown in FIG. Acquire an RGB thermal image with the background removed, and recognize the chest from the RGB thermal image from which the background is removed, and determine the left chest RGB channel thermal image and the right chest RGB channel thermal image by the breast cancer determination unit 400. )

The breast cancer determination unit 400 stores histogram and breast cancer thermal analysis information of a normal person, and determines the primary breast cancer based on the histogram and the histogram feature vector, and uses the feature vector of the breast cancer thermal image analysis information. It may be configured to determine whether the subject patient breast cancer by one or more of the breast cancer determination. However, even if the primary breast cancer is judged to be breast cancer, it cannot be confirmed that the patient has breast cancer. Therefore, the breast cancer determination unit 400 may perform only the determination of the secondary breast cancer or apply the primary breast cancer determination as an auxiliary determining means to determine the secondary breast cancer.

The breast cancer determination unit 400 extracts feature vectors by analyzing the left and right RGB channel thermal images processed before the image, and generates breast cancer thermal image analysis information by applying an artificial neural network to the extracted feature vectors. The breast cancer thermal image analysis information of the normal person is compared to determine whether breast cancer exists in the left and right chests (second breast cancer determination).

2 is a view showing a detailed configuration of the image background removal unit of the breast cancer diagnosis apparatus using a thermal imaging camera according to the present invention, Figure 7 is a view showing a thermal image for each RGB channel according to an embodiment of the present invention, FIG. 8 is a view illustrating contour line images for explaining contour detection comparison results according to whether Gaussian filtering is performed according to the present invention, and FIG. 9 is a diagram illustrating contour images detected when Huff transform is performed according to the present invention. Hereinafter, a description will be given with reference to FIGS. 2 and 7 to 9.

The image background remover 200 includes a channel selector 210, a (first) Gaussian filter 220, an outline detector 230, and a background deleter 250, and optionally an outline enhancer 240. It may include more.

The channel selector 210 selects and outputs only the red (R) channel thermal image from the RGB thermal image input from the thermal image acquirer 10.

This is because the mean and standard deviation of the thermal image of the red channel are smaller than the mean and standard deviation of the other channels as shown in FIG. 7 and Table 1 below, making it easier to detect the contour of the body including the patient's chest. to be.

The Gaussian filtering unit 220 performs Gaussian filtering on the red channel thermal image input from the channel selecting unit 210 and outputs the Gaussian filtering.

The contour detection unit 230 detects the contour from the Gaussian filtered red channel thermal image by the Gaussian filtering unit 220 and outputs the contour, and a Canny Edge process is applied.

In the image background remover 200 of the present invention, the Gaussian filter 220 may be selectively configured. However, the contour 712 extracted by performing the Canny edge process on the Gaussian filtered red channel thermal image 711 in FIG. 8 (b) is used to display the red channel thermal image 701 as shown in FIG. It can be seen that the canny edge process is performed more sharply than the extracted contour 702.

Therefore, after Gaussian filtering is performed through the Gaussian filtering unit 220, it may be desirable to detect the contour.

The contour reinforcement unit 240 may also be selectively configured to perform the Hough Transform process to more clearly process the contour detected by the contour detection unit 230 as shown in FIG. 9.

The background deleting unit 250 receives the RGB channel thermal image from the thermal image obtaining unit 10, receives the contour from the contour reinforcing unit 240, and performs Gaussian filtering based on the contour. As described above, the background is deleted from the RGB channel thermal image and then output to the ROI setting unit 300.

3 is a diagram illustrating a detailed configuration of a region of interest setting unit of a breast cancer diagnosis apparatus using a thermal imaging camera according to the present invention, and FIG. 10 is a left and right chest region separated by breast recognition and ROI detection using a Hough transform according to the present invention. Is a diagram for explaining a method of extracting. A description with reference to FIGS. 3 and 10 is as follows.

The ROI setting unit 300 includes a channel selector 310, a Gaussian filtering unit 320, an outline detector 330, an outline enhancer 340, and a breast detector 350.

The channel selector 310 selects and outputs only the green (G) channel thermal image image from the RGB thermal image from which the background input from the image background remover 200 is removed. Choosing the green channel will be more advantageous because of the detection of the chest part of the patient taken

The Gaussian filtering unit 320 performs Gaussian filtering on the green channel thermal image from which the background is removed and then outputs it to the contour detection unit 330. By performing Gaussian filtering, the contour reinforcement unit 340 to be described later may increase the degree of agreement of the circular contour corresponding to the actual chest.

The contour detection unit 330 detects and outputs a contour of a body including a chest contour from the green channel thermal image by applying a Canny edge process to the Gaussian filtered green channel thermal image, as shown in FIG. 10. do.

The contour reinforcement unit 340 detects a circular contour corresponding to the humidified part by applying a Huff circle conversion process to a body contour including a breast contour detected by the contour detector 330. . FIG. 10B illustrates a case where the detected circle contour is applied to the green channel thermal image from which the background is removed.

The breast detection unit 350 applies the detected circle contour to the RGB channel thermal image from which the background is removed, and thus, the left chest thermal image and the right chest thermal image of the RGB channel of the circular shape as shown in FIG. The detection is output to the breast cancer determination unit 400. In order to increase the ROI extraction rate, the breast detector 350 sets the minimum radius of the circle as the average radius of the female breast, and the minimum distance value between the centers of the circles detected by the Hough circle transformation is 50. Was set.

4 is a view showing a detailed configuration of the breast cancer determination unit of the breast cancer diagnosis apparatus using a thermal imaging camera according to the present invention, Figure 11 is a diagram showing a histogram of a normal person and a breast cancer patient for explaining a breast cancer detection method according to the invention 12 is a view for explaining a method of generating a co-generation matrix for explaining a breast cancer detection method according to the present invention, Figure 13 is to distinguish the intensity level for the co-generation matrix generated in accordance with the present invention in different colors The figure shown. Hereinafter, a description will be given with reference to FIGS. 4 and 11 to 13.

The breast cancer determination unit 400 includes a histogram analyzer 410, a simultaneous matrix characteristic generator 420, and a breast cancer analyzer 430.

The histogram analyzing unit 410 stores histograms of the left and right chest thermal images of the normal person in advance, and each of the histogram analysis unit 410 includes a left chest thermal image and a right chest thermal image input from the ROI setting unit 300. Create and print a histogram. The histogram is a histogram of the color (thermal) intensity level versus the number of pixels of the thermal image as shown in FIG.

The histogram represents the existence probability distribution function of the color intensity in a given left and right chest thermal image, and can be expressed by Equation 1 below.

Where p is the pixel value of the thermal image, and m and n represent the width and height of the image.

FIG. 11 is a histogram generated by Equation 1, wherein (a) is a histogram of a normal person, and (b) shows a histogram of a patient who is likely to have breast cancer.

Feature vectors that may be applied to the histogram may be Mean, Standard Deviation, Skewness, Kurtosis, and the like.

The mean is the average pixel value of the RGB channel of the RGB channel thermal image, the standard deviation is the square root of the variance of the image, the asymmetry is the symmetry of the color distribution, and the kurtosis is the normal distribution. It is a value measured for the distribution.

The mean, standard deviation, asymmetry rate and kurtosis may be obtained by Equation 2 below.

As shown in FIG. 12, the simultaneous matrix characteristic generator 420 simultaneously generates a gray level (channel) 1201 from a left chest thermal image and a right chest thermal image input from the ROI setting unit 300. A generation matrix 1202 is generated, the frequency of occurrence is specified as a specific value using an image N * N matrix mask, and pairs of pixels are obtained in two diagonal relationship, horizontal, vertical and different. Simultaneous matrices are used to define secondary statistical features.

Feature vectors applied to the co-occurrence matrix may include energy, contrast, homogeneity, and correlation.

The energy refers to the uniformity of the sum of the squared elements of the co-occurrence matrix, the entropy is a measure of statistical randomness, or uncertainty, the contrast can measure local variability in the image, and the homogeneity is two different diagonals. The element distribution of the co-occurrence matrix is measured with respect to the direction, and the correlation may be calculated by Equation 3 by indicating the relationship between pixels.

Since the feature vectors of Equation 3 are well known to those skilled in the art, detailed descriptions thereof will be omitted.

The co-generation matrix characteristic generator 420 may be configured to convert the obtained pixel pairs into color intensity levels as shown in FIG. 13.

The breast cancer analysis unit 430 is a feature vector such as energy, entropy, contrast, homogeneity, correlation, etc., for each RGB matrix based on histogram, average, variance, asymmetry, kurtosis, vertical horizontal, and co-occurrence in two different diagonal directions. The values are classified by applying an artificial neural network, the histogram of the normal person and the calculated histogram are compared as shown in FIG. 11, and the color intensity level information and the chest of the normal person which are output from the co-generation characteristic generator 420 as shown in FIG. 13. Comparing the color intensity level information for the thermal image may be configured to determine whether or not breast cancer of the patient taking the thermal image, and compares the feature vector of the normal person with the feature vector of the patient for each feature vector You can also judge.

In addition, the breast cancer analyzer 430 calculates the relative entropy of the right and left chest thermal image of the patient by the following Equation 4, and compares the calculated entropy of the patient with the relative entropy of the normal person to determine whether the breast cancer has occurred. It may be configured to judge. Table 2 below shows the difference of relative entropy in the green channel and gray channel of the breast cancer patient and the normal person and the result of breast cancer determination.

In addition, the breast cancer analyzer 430 may be configured to determine whether the patient invented breast cancer by combining two or more of the above-described methods.

On the other hand, the present invention is not limited to the above-described typical preferred embodiment, but can be carried out in various ways without departing from the gist of the present invention, various modifications, alterations, substitutions or additions in the art Anyone who has this can easily understand it. If the implementation by such improvement, change, replacement or addition falls within the scope of the appended claims, the technical idea should also be regarded as belonging to the present invention.

[Description of the code]

10: thermal image acquisition unit 100: image preprocessor

200: image background remover 210: channel selector (red)

220: Gaussian filtering unit 230: Contour detection unit

240: contour enhancement unit 250: background deleting unit

300: region of interest setting unit 310: channel selection unit (green)

320: Gaussian filtering unit 330: contour detection unit

340: contour enhancement unit 350: breast detection unit

400: breast cancer determination unit 410: histogram analysis unit

420: generation characteristics of the concurrent matrix 430: breast cancer analysis unit

Claims

A thermal image acquisition unit including a thermal imaging camera and outputting a thermal image by capturing a body including a chest of a patient with the thermal camera;

An image preprocessing unit which deletes a background from the thermal image and performs image preprocessing to output a left chest thermal image and a right chest thermal image that recognize a chest in a body from which the background is deleted; And

It stores the breast cancer thermal image analysis information of a normal person, extracts feature vectors by analyzing the left and right chest thermal images processed before the image, classifies the extracted feature vectors by applying an artificial neural network and parameterize the feature vectors And a breast cancer determination unit configured to generate breast cancer thermal image analysis information, and compare the generated breast cancer thermal image analysis information with the breast cancer thermal image analysis information of the normal person to determine whether breast cancer exists in the left and right breasts. Breast cancer diagnostic apparatus using a thermal imaging camera.
The method of claim 1,

The image preprocessing unit,

An image background remover configured to delete and output a background from the thermal image; And

And a region of interest setting unit for recognizing a chest of interest from the thermal image of which the background is deleted and outputting a left chest thermal image and a right chest thermal image.
The method of claim 2,

The image background remover,

A channel selector which selects and outputs a thermal image of a red (R) channel among the RGB channels of the thermal image output from the thermal image acquisition unit;

A Gaussian filtering unit configured to output Gaussian filtering of the red channel thermal image;

An outline detector detecting an outline of the body from the thermal image of the Gaussian filtered red channel; And

And a background deletion unit for deleting a background from a thermal image including all of the RGB channels based on the detected contour line.
The method of claim 3,

The image background remover,

The apparatus for diagnosing breast cancer using a thermal imaging camera according to claim 1, further comprising an outline reinforcement unit for performing a Hough transform on the detected contour to further reinforce the outline.
The method of claim 2,

The ROI setting unit,

A channel selector which selects and outputs only a green channel from a thermal image including all of the RGB channels from which the background is removed;

A Gaussian filtering unit which outputs the thermal image of the green channel by Gaussian filtering;

An outline detection unit for detecting and outputting an outline of a body including a chest part from a thermal image of the Gaussian filtered green channel;

An outline reinforcing unit configured to perform a Hough circle transformation from the body outline detected by the outline detecting unit to detect a circle corresponding to the chest part; And

A breast detection unit extracting a left chest thermal image and a right chest thermal image from a thermal image including all of the RGB channels based on a region of interest (ROI) detection based on a circle detected by the contour detector. Breast cancer diagnostic apparatus using a thermal imaging camera comprising a.
The method of claim 5,

The contour reinforcement unit,

The minimum radius value, which is a parameter of the Hough circle transformation, is set to the average radius value of the female breast, and the minimum distance value between the centers of the left and right circles defining the distance between the chests for detecting the chest is set to 50 Breast cancer diagnostic apparatus using a thermal imaging camera, characterized in that.
The method of claim 1,

Breast cancer judgment part,

A simultaneous matrix characteristic generator for extracting and outputting feature vectors based on the concurrent matrix for vertical, horizontal, and two diagonals of the pre-processed left and right image images; And

It stores breast cancer thermography information of a normal person, generates artificial breast cancer thermography information by applying artificial neural network to the extracted feature vectors, compares breast cancer thermography information of the normal person, and compares breast cancer of the left and right breasts. Breast cancer diagnosis apparatus using a thermal imaging camera, characterized in that it comprises a breast cancer analysis unit for determining the presence.
The method of claim 7, wherein

The breast cancer determination unit,

And a histogram analyzer configured to analyze and output a distribution of histograms measured based on feature vectors of respective RGB channels of the left and right chest image images.

The breast cancer analysis unit,

The histogram of the left and right image of the normal person is further stored, and the histogram of the normal person and the histogram measured based on the RGB channel characteristic information through the histogram analyzer are primarily used to determine whether breast cancer is present. A breast cancer diagnosis apparatus using a thermal imaging camera, characterized in that it is judged whether breast cancer exists in a corresponding breast when judging whether or not secondary breast cancer is based on breast cancer thermal image analysis information.
The method of claim 8,

The breast cancer determination unit,

Energy, Entropy, Contrast, Correlation, Homogeneity, and Mean for each RGB channel in the vertical, horizontal, and two diagonal directions with the feature vectors ), Variance (Variance), asymmetry (Skewness), kurtosis (Kurtosis) comprising a diagnostic apparatus for breast cancer using a thermal imaging camera.
A thermal image acquisition step of obtaining a thermal image by capturing a body including a chest portion of a patient with the thermal imager and outputting a thermal image;

An image preprocessing step of performing an image preprocessing unit to remove a background from the thermal image, and to output a left chest thermal image and a right chest thermal image which recognizes a chest in a body from which the background is deleted; And

The breast cancer determination unit analyzes the left and right chest thermal images pre-processed to extract feature vectors, classifies the extracted feature vectors by applying an artificial neural network, and analyzes the breast cancer thermal image analysis information using the feature vectors as parameters. And a breast cancer determination step of determining whether breast cancer exists in the left and right chests by comparing the generated breast cancer thermal imaging information with the normal breast cancer thermal imaging information. How to diagnose breast cancer.
The method of claim 10,

The image preprocessing step,

An image background removing step of removing and outputting a background from the thermal image by an image background removing unit; And

The ROI setting unit may include a ROI setting step of outputting a left chest thermal image and a right chest thermal image by recognizing a chest as the ROI from the thermal image from which the background is removed. How to diagnose breast cancer.
The method of claim 11,

The image background removing step,

A channel selecting step of selecting, by a channel selecting unit, a red (R) channel among the RGB channels of the thermal image output from the thermal image obtaining unit;

A Gaussian filtering step of performing a Gaussian filtering on the red image of the red channel by a Gaussian filtering unit;

An outline detection step of detecting an outline of a body from a thermal image of the Gaussian filtered red channel by an outline detector; And

And a background deleting step of deleting a background from a thermal image including all of the RGB channels based on the detected contour.
The method of claim 12,

The image background removing step,

And a contour enhancement step of performing an Huff transform on the detected contour to further enhance the contour.
The method of claim 11,

The area of interest setting step is

A channel selection step of selecting and outputting only a green channel from a thermal image including all of the RGB channels from which the background is removed;

Gaussian filtering to output a thermal image of the green channel by Gaussian filtering;

Contour detection step of detecting and outputting the contour of the body including the chest portion from the thermal image of the Gaussian filtered green channel;

Contour reinforcement step of performing a Hough circle transformation from the body contour detected by the contour detection unit to perform a circle detection corresponding to the breast portion; And

And a breast detection step of extracting a left chest thermal image and a right chest thermal image from a thermal image including all of the RGB channels through ROI detection based on a circle detected by a contour detector. Breast cancer diagnosis method using a thermal imaging camera characterized in that.
The method of claim 14,

The contour strengthening step,

The contour enhancer sets the minimum radius value, which is the parameter of the Huff circle transformation, to the average radius value of the female breast, and sets the minimum distance value between the centers of the left and right circles to define the distance between the chests for detecting the chest. Breast cancer diagnostic method using a thermal imaging camera, characterized in that set to.
The method of claim 10,

Breast cancer determination step,

A histogram analysis step of analyzing and outputting histograms of each of the left and right chest thermal images;

A feature information generation step of extracting and outputting feature vectors by analyzing the left and right chest thermal images processed before the image; And

It stores breast cancer thermography information of normal people, classifies the extracted feature vectors by applying an artificial neural network, generates breast cancer thermography information using the feature vectors as a parameter, and analyzes breast cancer thermography information of the normal people. Comparing the breast cancer diagnosis method comprising a breast cancer analysis step of determining the presence of breast cancer in the left and right chest.
The method of claim 16,

The breast cancer analysis step,

The breast cancer analyzer further stores histograms of left and right chest thermal images of a normal person, and compares the histogram of the normal person and the histograms measured by the histogram analyzer to primarily determine a histogram analysis step of determining breast cancer. Including,

After the breast cancer analysis step, if it is determined that both the breast cancer by the histogram analysis and breast cancer based on the breast cancer thermal analysis information is determined to be breast, it is finally determined that the breast cancer is present in the breast Breast cancer diagnosis method using a thermal imaging camera.
The method of claim 17,

The feature vector,

Energy, Entropy, Contrast, Correlation, Homogeneity, and Mean, Variance by RGB Channel Method for diagnosing breast cancer using a thermal imaging camera, characterized in that it includes variance, skewness, and kurtosis.