Breast cancer analysis and prediction system and method based on artificial intelligence

Abstract

The invention discloses a breast cancer analysis and prediction system and method based on artificial intelligence, relates to the technical field of breast cancer, solves the problems of slow manual check and large numerical deviation, and locks associated feature vectors based on different change features of different focus areas in different images. According to the method and the system provided by the invention, the medical personnel can quickly identify the deterioration condition of the corresponding kitchen area and timely display the signal, so that the medical personnel can conveniently and timely know the related condition and timely take related medical treatment measures, and the medical personnel can conveniently and rapidly know the related condition and timely take related medical treatment measures according to the scientific and quantitative proportion standard. According to the method, highly suspicious focus areas are accurately screened out, fuzziness and uncertainty of manual judgment of burr features are overcome, breast cancer focuses with high concealment and complex forms are free from hiding, and a precedent is made for timely intervention and prevention of disease deterioration.

Description

Technical Field

[0001] The present invention relates to the technical field of breast cancer prediction, and in particular to a breast cancer analysis and prediction system and method based on artificial intelligence. Background Art

[0002] Breast cancer, also known as breast cancer, refers to a malignant tumor formed by the uncontrolled proliferation of breast epithelial (ductal or lobular epithelial) tissue under the action of multiple carcinogenic factors; 99% of breast cancers occur in women and only 1% in men.

[0003] The application with publication number CN118333960A discloses a breast cancer analysis and prediction system and method based on artificial intelligence. The method includes: obtaining a breast X-ray image of a patient subject to be detected; extracting X-ray image features of the breast X-ray image to obtain a breast X-ray image feature map; performing feature constraints and feature significant processing on the breast X-ray image feature map to obtain a significant breast X-ray image feature map; and determining a prediction result based on the significant breast X-ray image feature map. In this way, the risk level of breast cancer of the patient subject to be detected can be intelligently identified, assisting doctors in making more accurate diagnoses.

[0004] In the process of analyzing and predicting breast cancer, medical staff generally identify the deterioration of lesions based on corresponding pathological images. However, in the actual treatment process, medical staff cannot perform detailed verification well, but instead rely on personal experience. The verification values ​​are greatly deviated from those of artificial intelligence, and the image verification time is long, which cannot achieve the rapid verification effect of artificial intelligence. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a breast cancer analysis and prediction system and method based on artificial intelligence, which solves the problems of slow manual verification and large numerical deviation.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a breast cancer analysis and prediction method based on artificial intelligence, comprising the following steps:

[0007] Step 1: Confirm the breast X-ray image associated with the patient, and based on the grayscale values ​​associated with different points in the breast X-ray image, locate the abnormal area from the breast X-ray image:

[0008] S11, based on the mammary X-ray image confirmed this time, determine the grayscale values ​​associated with different points in the image, and calibrate the different grayscale values ​​associated with different points as HD i , where i represents the different points in this image;

[0009] S12, the different gray values ​​HD associated with different points i , check with the preset interval, and set the gray value HD i Satisfaction: HD i ∈The relevant points in the preset interval are marked as abnormal points, otherwise, no calibration is performed;

[0010] S13, marking the area covered by the groups of abnormal points as an abnormal area, and marking it in the breast X-ray image;

[0011] Step 2: Based on the abnormal area marked in the mammary X-ray image, the gradient features of several points in the abnormal area are confirmed in combination with the mammary X-ray image, and the gradient points are locked from the gradient features determined in sequence, thereby locking the overall area contour of the abnormal area:

[0012] S21. Based on the determined abnormal area and the several points existing in the abnormal area, the several points existing in the corresponding single group of abnormal areas are marked as pending points, and the pending point is taken as the center point, and the adjacent points around the center point are taken as subsidiary points. There are eight groups of subsidiary points, and the center point is located at the center position. The pending points and the determined subsidiary points are taken as the point sequence of the pending points, and a group of gray value sequences is determined based on the determined point sequence and the gray values ​​corresponding to different points: Where D k is the gray value associated with the determined point to be determined, and H1-H8 are the gray values ​​corresponding to the surrounding points of the point to be determined;

[0013] S22, based on the gray value sequence associated with the undetermined point, determine the horizontal vertical gradient Gx associated with the undetermined point k And the vertical gradient Gy k :

[0014] Where Gx k =(-1)×H1+0×H2+1×H3+(-2)×H4+0×D k +2×H5+(-1)×H6+0×H7+1×H8;

[0015] Gy k =(-1)×H1+(-2)×H2+(-1)×H3+0×H4+0×D k +0×H5+1×H6+2×H7+1×H8;

[0016] use Determine the gradient feature G associated with this undetermined point k ;

[0017] S23, based on the different gradient features G associated with different points to be determined in this abnormal area k , G k >Y1 is the undetermined point to be calibrated as the contour point, where Y1 is the preset value and G k No calibration is performed for the undetermined points ≤Y1;

[0018] S24, confirming a plurality of contour points in the abnormal area, and connecting adjacent contour points based on the determined plurality of contour points to confirm the overall area contour belonging to the abnormal area;

[0019] Step 3: Based on the overall regional contour determined by the abnormal region, select the burr segment from the overall regional contour of the abnormal region, and evaluate whether the abnormal region belongs to the focal region based on the proportion of the burr segment in the overall regional contour:

[0020] S31, marking the overall area contour determined by the abnormal area as the pending contour, determining the contour points in the pending contour in sequence, and identifying the point angle JJ associated with the corresponding contour point from the determined contour points q , where q represents different contour points. A group of contour points is randomly selected, two groups of contour points adjacent to this contour point are connected, two groups of angles between the two groups of connecting lines are determined, and the angle with the smallest value is selected from the two groups of angles as the point angle of this contour point;

[0021] S32, confirm the different point angles associated with different contour points in turn, and satisfy: JJ q Contour points ≤Y2 are marked as burr points, where Y2 is the preset value and does not meet JJ q No calibration is performed on the contour points ≤Y2, and the two sets of lines associated with the burr points are taken as burr segments;

[0022] S33, determining the total length ZL of the burr segment in the undetermined contour, and then determining the total length ZD of the undetermined contour, using ZL÷ZD=ZB to determine the proportion ZB of the burr segment, and marking the abnormal area with ZB≥40% as the focal area; no marking is performed for the abnormal area with ZB<40%;

[0023] Step 4: Based on the focal area determined in the breast X-ray image of the associated patient, confirm the breast X-ray image of the corresponding associated patient taken last time, and use the same method to simultaneously determine the focal area from the breast X-ray image taken last time, and compare and analyze the two confirmed groups of focal areas to assess whether there are any abnormal changes in this associated patient, and display them in real time:

[0024] S41, based on the patient associated with the current breast X-ray image, confirm the most recent breast X-ray image of the patient from the historical data and mark it as the main image, and process the confirmed main image in the same manner as steps 1 to 3, confirm the focal area associated with the main image and record it as the main focal area;

[0025] S42, based on the overall edge contour of the main focal area, confirm the center point of the main focal area, place the overall edge contour in a two-dimensional coordinate system, and based on different two-dimensional coordinates associated with different contour points, average the two-dimensional coordinates of several groups of contour points to determine the average coordinates, and the determined average coordinates are the location of the center point of the corresponding main focal area;

[0026] Then, based on the confirmed main image, a set of horizontal reference lines is generated, and the horizontal reference lines are gradually moved upward from the bottom of the main image. The intersection segments of the horizontal reference lines and the main image during the movement are recorded, and the longest intersection segment is marked as a feature segment.

[0027] Record the intersection point between the feature segment and the main image on the left side. The left and right sides of the main image have been calibrated in advance. Take the left intersection point as the starting point and the center point of the main focal area as the end point to confirm a set of position vectors, which are recorded as the main position vectors.

[0028] S43, using the current breast X-ray image as a secondary image, using the focal area determined in the secondary image as a secondary focal area, determining a horizontal reference line in the secondary image in the same manner, and synchronously determining a left intersection point in the secondary image, combining the initial point of the main position vector with the currently confirmed left intersection point based on the confirmed main position vector, locking the position of the end point of the main position vector and recording it as a characteristic position: identifying whether the characteristic position has a secondary focal area, if so, using the secondary focal area to which the characteristic position belongs as a verification area for the main focal area, and directly generating an error signal if not.

[0029] S44. Compare the confirmed primary focus area and secondary focus area: preferentially calibrate the area of ​​the primary focus area as M1, and the area of ​​the secondary focus area as M2; calibrate the total length of the burr segment of the primary focus area as ZL1, and calibrate the total length of the burr segment of the secondary focus area as ZL2; use: HD = (M2-M1) × C1 + (ZL2-ZL1) × C2 to confirm the verification value HD, wherein C1 and C2 are both preset fixed coefficient factors; compare the verification value HD with the preset value Y3; if HD>Y3, generate an abnormal patient change signal; otherwise, continue monitoring.

[0030] Preferably, a breast cancer analysis and prediction system based on artificial intelligence comprises:

[0031] The abnormal area confirmation terminal confirms the breast X-ray image associated with the associated patient, and based on the grayscale values ​​associated with different points in the breast X-ray image, locks the abnormal area from the breast X-ray image;

[0032] The region contour confirmation end confirms the gradient features of several points in the abnormal region based on the abnormal region marked in the mammary X-ray image and in combination with the mammary X-ray image, and locks the gradient points from the gradient features determined in sequence, thereby locking the overall region contour of the abnormal region;

[0033] The focal region locking end selects a burr segment from the overall regional contour of the abnormal region based on the overall regional contour determined by the abnormal region, and evaluates whether the abnormal region belongs to the focal region based on the proportion of the burr segment in the overall regional contour;

[0034] The patient abnormality verification end confirms the last breast X-ray image of the corresponding associated patient based on the focal area determined in the breast X-ray image of the associated patient, and simultaneously determines the focal area from the last breast X-ray image in the same way, and verifies and analyzes the two confirmed focus areas to assess whether there are any abnormal changes in the associated patient, and displays them in real time.

[0035] The present invention provides a breast cancer analysis and prediction system and method based on artificial intelligence. Compared with the prior art, it has the following beneficial effects:

[0036] With the scientific and quantitative proportion standard (ZL÷ZD=ZB, ZB≥40% is marked as focal area), highly suspicious focal areas are accurately screened out, overcoming the ambiguity and uncertainty of manual judgment of burr characteristics, making breast cancer foci with strong concealment and complex morphology nowhere to hide, and gaining the opportunity for timely intervention and blocking the deterioration of the disease;

[0037] Based on the different changing characteristics of different focal areas in different images, the associated feature vectors are locked, and endpoint confirmation and comparison are performed to make the verification of the confirmed focal areas more accurate. At the same time, the deterioration of the corresponding focal areas can be quickly identified, and the signals can be displayed in time, so that medical staff can be informed of the relevant situation in time and take relevant medical treatment measures in time. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of the process of the present invention;

[0039] Figure 2 It is a schematic diagram of the principle framework of the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0041] First embodiment

[0042] See also Figure 1 , the present application provides a breast cancer analysis and prediction method based on artificial intelligence, comprising the following steps:

[0043] Step 1: confirm the breast X-ray image associated with the associated patient, and lock the abnormal area from the breast X-ray image based on the grayscale values ​​associated with different points in the breast X-ray image. Specifically, in the breast X-ray film, the grayscale value is used to determine the lesion area; the breast X-ray film is essentially based on the difference in the absorption degree of X-rays by different tissues, and this absorption difference is manifested as different grayscale values ​​on the image; the grayscale value usually ranges from 0 (black, indicating an area where X-rays are almost completely penetrated, such as air or fat tissue) to 255 (white, indicating an area where X-rays are largely absorbed, such as bones or high-density lesions);

[0044] The specific sub-steps of locking the abnormal area are as follows:

[0045] S11, based on the mammary X-ray image confirmed this time, determine the grayscale values ​​associated with different points in the image, and calibrate the different grayscale values ​​associated with different points as HD i , where i represents the different points in this image;

[0046] S12, the different gray values ​​HD associated with different points i , and check with the preset interval, where the endpoint values ​​of the preset interval are preset values, which are formulated by relevant operators based on experience, generally 180-255, and the gray value HD i Satisfaction: HD i ∈The relevant points in the preset interval are marked as abnormal points, otherwise, no calibration is performed;

[0047] S13, marking the area covered by several groups of abnormal points as an abnormal area, and marking it in the breast X-ray image, wherein the abnormal points generally appear continuously, and when there is a lesion or abnormality in a certain area, the corresponding area covered by several abnormal points is the corresponding lesion area, that is, the confirmed abnormal area;

[0048] Step 2: Based on the abnormal area marked in the mammary X-ray image, the gradient characteristics of several points in the abnormal area are confirmed in combination with the mammary X-ray image, and the gradient points are locked from the gradient characteristics determined in sequence, thereby locking the overall regional contour of the abnormal area. Specifically, each different abnormal area has a different regional contour, and the gradient characteristics of the pixel points of the regional contour are relatively obvious. It can be directly confirmed based on the gradient data between the corresponding pixel points, thereby directly locking the overall regional contour of the different abnormal areas;

[0049] Among them, the specific sub-steps of locking the overall area outline of this abnormal area are:

[0050] S21. Based on the determined abnormal area and the several points existing in the abnormal area, the several points existing in the corresponding single group of abnormal areas are marked as pending points, and the pending point is taken as the center point, and the adjacent points around the center point are taken as subsidiary points. There are eight groups of subsidiary points, and the center point is located at the center position. The pending points and the determined subsidiary points are taken as the point sequence of the pending points, and a group of gray value sequences is determined based on the determined point sequence and the gray values ​​corresponding to different points: Where D k is the gray value associated with the determined point to be determined, and H1-H8 are the gray values ​​corresponding to the surrounding points of the point to be determined;

[0051] S22, based on the gray value sequence associated with the undetermined point, determine the horizontal vertical gradient Gx associated with the undetermined point k And the vertical gradient Gy k :

[0052] Where Gx k =(-1)×H1+0×H2+1×H3+(-2)×H4+0×D k +2×H5+(-1)×H6+0×H7+1×H8. Specifically, the operator template associated with the horizontal direction is:

[0053] Among them, Gy k =(-1)×H1+(-2)×H2+(-1)×H3+0×H4+0×D k +0×H5+1×H6+2×H7+1×H8. Specifically, the operator template associated with the vertical direction is:

[0054] use Determine the gradient feature G associated with this undetermined point k ;

[0055] S23, based on the different gradient features G associated with different undetermined points in this abnormal area k , G k The undetermined point position of >Y1 is calibrated as the contour point position, where Y1 is the preset value, and its specific value is determined by the operator based on experience. k No calibration is performed on the undetermined points ≤Y1. Specifically, when the gradient feature is more obvious, that is, when the corresponding gradient feature is larger, the corresponding total gradient feature value will be larger. Under normal circumstances, the gradient feature will not change significantly.

[0056] S24, confirming a number of contour points in the abnormal area, and connecting adjacent contour points based on the determined contour points to confirm the overall area contour belonging to the abnormal area (the gradient features generated by the points associated with the edge contour of the corresponding area are relatively large, and because there is an obvious difference in grayscale value, the edge contour associated with the corresponding abnormal area can be determined based on the adjacent contour points, thereby locking the overall area contour of the abnormal area);

[0057] Step 3: Based on the overall regional contour determined by the abnormal region, a burr segment is selected from the overall regional contour of the abnormal region, and based on the proportion of the burr segment in the overall regional contour, it is evaluated whether the abnormal region belongs to a focal region, wherein the specific sub-steps of the evaluation are:

[0058] S31, marking the overall area contour determined by the abnormal area as the pending contour, determining the contour points in the pending contour in sequence, and identifying the point angle JJ associated with the corresponding contour point from the determined contour points q , where q represents different contour points, a group of contour points are randomly selected, two groups of contour points adjacent to this contour point are connected, and two groups of angles between the two groups of connecting lines are determined (a group of inner angles and a group of outer angles), and the angle with the smallest value is selected from the two groups of angles as the point angle of this contour point. Specifically, in the actual processing process, there are corresponding angles between the corresponding point connecting lines. Because the contour around the focal area is generally burr-shaped, the point angles associated with the corresponding burrs are generally acute angles. If the internal angle is selected, when the corresponding burr tends to bulge inward, then the corresponding internal angle is an obtuse angle, which will result in recognition errors. Therefore, the minimum angle between the two groups of angles is selected as the point angle of the corresponding contour point, so as to lock the focal area;

[0059] S32, confirm the different point angles associated with different contour points in turn, and satisfy: JJ qThe contour points ≤Y2 are marked as burr points, where Y2 is a preset value. The specific value is determined by the operator based on experience, usually 45°, which does not meet JJ q No calibration is performed on the contour points ≤Y2, and the two sets of lines associated with the burr points are taken as burr segments;

[0060] S33, determining the total length ZL of the burr segments in the undetermined contour, and then determining the total length ZD of the undetermined contour, using ZL÷ZD=ZB to determine the proportion value ZB of the burr segments, and marking the abnormal area with ZB≥40% as the focal area, otherwise no marking is performed. Specifically, when the area is abnormal, there will be a large number of burr segments, and the corresponding area with a large proportion of burr segments belongs to the focal area, which needs to be paid great attention to;

[0061] Step 4: Based on the focal area determined in the breast X-ray image of the associated patient, confirm the breast X-ray image taken last time for the corresponding associated patient, and use the same method to simultaneously determine the focal area from the breast X-ray image taken last time, and compare and analyze the two confirmed groups of focal areas to assess whether there is any abnormal change in the associated patient, and display it in real time, that is, assess the degree of change based on the overall change of the area and burrs in the focal area. If the deterioration is serious, it needs to be taken seriously and displayed in real time. The specific sub-steps of the assessment are as follows:

[0062] S41, based on the patient associated with the current breast X-ray image, confirm the most recent breast X-ray image of the patient from the historical data and mark it as the main image (that is, the image taken last time), and process the confirmed main image in the same manner as steps 1 to 3, confirm the focal area associated with the main image and record it as the main focal area;

[0063] S42, based on the overall edge contour of the main focal area, confirm the center point of the main focal area, place the overall edge contour in a two-dimensional coordinate system, and based on different two-dimensional coordinates associated with different contour points, average the two-dimensional coordinates of several groups of contour points to determine the average coordinates, and the determined average coordinates are the location of the center point of the corresponding main focal area;

[0064] Then, based on the confirmed main image, a set of horizontal reference lines is generated (the horizontal reference lines are parallel to the image plate and the MLO compression plate. When performing breast photography, the patient needs to place the breast between the image plate and the MLO compression plate before taking the photo). The horizontal reference lines are gradually moved upward from the bottom of the main image, and the intersection segments of the horizontal reference lines and the main image during the movement are recorded. The longest intersection segment is marked as the feature segment.

[0065] Record the intersection of the feature segment and the left side of the main image (that is, the location of the areola). The left and right sides of the main image have been calibrated in advance. Take the left intersection as the starting point and the center point of the main focal area as the end point to confirm a set of position vectors, which are recorded as the main position vectors.

[0066] S43, taking the current breast X-ray image as a secondary image, taking the focal area determined in the secondary image as the secondary focal area, determining the horizontal reference line in the secondary image in the same manner, and synchronously determining the left intersection point in the secondary image, combining the initial point of the main position vector with the currently confirmed left intersection point based on the confirmed main position vector, locking the position where the end point of the main position vector is located and recording it as a characteristic position: identifying whether there is a secondary focal area to which the characteristic position belongs, if so, taking the secondary focal area to which the characteristic position belongs as a verification area for the main focal area, if not, directly generating an error signal, indicating that there is an incorrect posture during shooting, resulting in failure to find the corresponding comparison area;

[0067] S44, comparing the confirmed primary focal region and secondary focal region: preferentially calibrating the area of ​​the primary focal region as M1, calibrating the area of ​​the secondary focal region as M2, and calibrating the total length of the burr section of the primary focal region as ZL1, and calibrating the total length of the burr section of the secondary focal region as ZL2, using: HD = (M2-M1) × C1 + (ZL2-ZL1) × C2 to confirm the check value HD, wherein C1 and C2 are both preset fixed coefficient factors, and their specific values ​​are determined by the operator based on experience, comparing the check value HD with the preset value Y3, wherein the specific value of Y3 is determined by the operator based on experience, if HD>Y3, then generating a patient change abnormal signal, otherwise, then continuously monitoring;

[0068] Specifically, when the changes are abnormal, the area corresponding to the same group of focal areas and the surrounding associated burr segments will gradually increase, resulting in a large difference in the verification values. Therefore, when such a situation occurs, it is generally necessary to generate relevant abnormal signals for display, so that relevant medical staff can take relevant patient treatment measures in a timely manner.

[0069] Second embodiment

[0070] Combination Figure 2 , a breast cancer analysis and prediction system based on artificial intelligence, comprising:

[0071] The abnormal area confirmation terminal confirms the breast X-ray image associated with the associated patient, and based on the grayscale values ​​associated with different points in the breast X-ray image, locks the abnormal area from the breast X-ray image;

[0072] Region contour confirmation end, based on the abnormal region calibrated in the mammogram, combines the mammogram to confirm the gradient features of several points within the abnormal region, locks the gradient points from the sequentially determined gradient features, and thus locks the overall region contour of this abnormal region;

[0073] Focal area locking end, based on the overall region contour determined for the abnormal region, selects the spiculated segments from the overall region contour of the abnormal region, and evaluates whether this abnormal region belongs to the focal area based on the proportion of the spiculated segments in the overall region contour;

[0074] Patient abnormality verification end, based on the focal area determined in the mammogram of the associated patient this time, confirms the mammogram taken by the associated patient last time, synchronously determines the focal area in the same way from the mammogram taken last time, and conducts a comparative analysis on the two sets of confirmed focal areas, evaluates whether there are abnormal changes in this associated patient, and displays it in real time.

[0075] Some of the data in the above formula are numerically calculated after removing their dimensions, and the content not described in detail in this specification belongs to the prior art well known to those skilled in the art.

[0076] The above embodiments are only used to illustrate the technical method of the present invention and not to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical method of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A breast cancer analysis and prediction method based on artificial intelligence, characterized in that: The following steps are involved: Step 1: confirming the breast X-ray image associated with the associated patient, and based on the grayscale values ​​associated with different points in the breast X-ray image, locking the abnormal area from the breast X-ray image; Step 2: Based on the abnormal area marked in the mammary X-ray image, the gradient characteristics of several points in the abnormal area are confirmed in combination with the mammary X-ray image, and the gradient points are locked from the gradient characteristics determined in sequence, thereby locking the overall area contour of the abnormal area; Step 3: based on the overall regional contour determined by the abnormal region, select a burr segment from the overall regional contour of the abnormal region, and assess whether the abnormal region belongs to a focal region based on the proportion of the burr segment in the overall regional contour; Step 4. Based on the focal area determined in the breast X-ray image of the associated patient, confirm the breast X-ray image of the corresponding associated patient taken at the last time, and simultaneously determine the focal area from the breast X-ray image taken at the last time using the same method, and compare and analyze the two confirmed groups of focal areas to assess whether there are any abnormal changes in this associated patient, and display them in real time.

2. The method for analyzing and predicting breast cancer based on artificial intelligence according to claim 1, characterized in that: In step 1, the specific sub-steps of locking the abnormal area are: S11, based on the mammary X-ray image confirmed this time, determine the grayscale values ​​associated with different points in the image, and calibrate the different grayscale values ​​associated with different points as HD i , where i represents the different points in this image; S12, the different gray values ​​HD associated with different points i , check with the preset interval, and set the gray value HD i Satisfaction: HD i ∈The relevant points in the preset interval are marked as abnormal points, otherwise, no calibration is performed; S13, marking the area covered by the groups of abnormal points as an abnormal area, and marking it in the breast X-ray image.

3. The method for analyzing and predicting breast cancer based on artificial intelligence according to claim 1, characterized in that: In step 2, the specific sub-steps of locking the overall area contour of the abnormal area are: S21. Based on the determined abnormal area and the several points existing in the abnormal area, the several points existing in the corresponding single group of abnormal areas are marked as pending points, and the pending point is taken as the center point, and the adjacent points around the center point are taken as subsidiary points. There are eight groups of subsidiary points, and the center point is located at the center position. The pending points and the determined subsidiary points are taken as the point sequence of the pending points, and a group of gray value sequences is determined based on the determined point sequence and the gray values ​​corresponding to different points: Where D k is the gray value associated with the determined point to be determined, and H1-H8 are the gray values ​​corresponding to the surrounding points of the point to be determined; S22, based on the gray value sequence associated with the undetermined point, determine the horizontal vertical gradient Gx associated with the undetermined point k And the vertical gradient Gy k : Where Gx k =(-1)×H1+0×H2+1×H3+(-2)×H4+0×D k +2×H5+(-1)×H6+0×H7+1×H8; Among them, Gy k =(-1)×H1+(-2)×H2+(-1)×H3+0×H4+0×D k +0×H5+1×H6+2×H7+1×H8; use Determine the gradient feature G associated with this undetermined point k ; S23, based on the different gradient features G associated with different undetermined points in this abnormal area k , G k >Y1 is the undetermined point to be calibrated as the contour point, where Y1 is the preset value and G k No calibration is performed for the undetermined points ≤Y1; S24, confirming a number of contour points in the abnormal area, and connecting adjacent contour points based on the determined contour points to confirm the overall area contour belonging to the abnormal area.

4. The method for analyzing and predicting breast cancer based on artificial intelligence according to claim 1, characterized in that: In step 3, the specific sub-steps for assessing whether the abnormal area belongs to the focal area are: S31, marking the overall area contour determined by the abnormal area as the pending contour, determining the contour points in the pending contour in sequence, and identifying the point angle JJ associated with the corresponding contour point from the determined contour points q , where q represents different contour points. A group of contour points is randomly selected, two groups of contour points adjacent to this contour point are connected, two groups of angles between the two groups of connecting lines are determined, and the angle with the smallest value is selected from the two groups of angles as the point angle of this contour point; S32, confirm the different point angles associated with different contour points in turn, and satisfy: JJ q ≤Y2 contour points are marked as burr points, where Y2 is the preset value and does not meet JJ q No calibration is performed on the contour points ≤Y2, and the two sets of lines associated with the burr points are taken as burr segments; S33, determining the total length ZL of the burr segment in the undetermined contour, and then determining the total length ZD of the undetermined contour, using ZL÷ZD=ZB to determine the proportion ZB of the burr segment, and marking the abnormal area with ZB≥40% as the focal area.

5. The method for analyzing and predicting breast cancer based on artificial intelligence according to claim 4, characterized in that: In the step S33, no calibration is performed for the abnormal area where ZB is less than 40%.

6. The method for analyzing and predicting breast cancer based on artificial intelligence according to claim 4, characterized in that: In step 4, the specific sub-steps of assessing whether the associated patient has abnormal changes are as follows: S41, based on the patient associated with the current breast X-ray image, confirm the most recent breast X-ray image of the patient from the historical data and mark it as the main image, and process the confirmed main image in the same manner as steps 1 to 3, confirm the focal area associated with the main image and record it as the main focal area; S42, based on the overall edge contour of the main focal area, confirm the center point of the main focal area, place the overall edge contour in a two-dimensional coordinate system, and based on different two-dimensional coordinates associated with different contour points, average the two-dimensional coordinates of several groups of contour points to determine the average coordinates, and the determined average coordinates are the location of the center point of the corresponding main focal area; Then, based on the confirmed main image, a set of horizontal reference lines is generated, and the horizontal reference lines are gradually moved upward from the bottom of the main image. The intersection segments of the horizontal reference lines and the main image during the movement are recorded, and the longest intersection segment is marked as a feature segment. Record the intersection point between the feature segment and the main image on the left side. The left and right sides of the main image have been calibrated in advance. Take the left intersection point as the starting point and the center point of the main focal area as the end point to confirm a set of position vectors, which are recorded as the main position vectors. S43, using the current breast X-ray image as a secondary image, using the focal area determined in the secondary image as a secondary focal area, determining a horizontal reference line in the secondary image in the same manner, and synchronously determining a left intersection point in the secondary image, combining the initial point of the main position vector with the currently confirmed left intersection point based on the confirmed main position vector, locking the position of the end point of the main position vector and recording it as a characteristic position: identifying whether the characteristic position has a secondary focal area, if so, using the secondary focal area to which the characteristic position belongs as a verification area for the main focal area, and directly generating an error signal if not. S44. Compare the confirmed primary focus area and secondary focus area: preferentially calibrate the area of ​​the primary focus area as M1, and the area of ​​the secondary focus area as M2; calibrate the total length of the burr segment of the primary focus area as ZL1, and calibrate the total length of the burr segment of the secondary focus area as ZL2; use: HD = (M2-M1) × C1 + (ZL2-ZL1) × C2 to confirm the verification value HD, wherein C1 and C2 are both preset fixed coefficient factors; compare the verification value HD with the preset value Y3; if HD>Y3, generate an abnormal patient change signal; otherwise, continue monitoring.

7. A breast cancer analysis and prediction system based on artificial intelligence, the system is operated according to the breast cancer analysis and prediction method based on artificial intelligence according to any one of claims 1 to 6, characterized in that: include: The abnormal area confirmation terminal confirms the breast X-ray image associated with the associated patient, and based on the grayscale values ​​associated with different points in the breast X-ray image, locks the abnormal area from the breast X-ray image; The region contour confirmation end confirms the gradient features of several points in the abnormal region based on the abnormal region marked in the mammary X-ray image and in combination with the mammary X-ray image, and locks the gradient points from the gradient features determined in sequence, thereby locking the overall region contour of the abnormal region; The focal region locking end selects a burr segment from the overall regional contour of the abnormal region based on the overall regional contour determined by the abnormal region, and evaluates whether the abnormal region belongs to the focal region based on the proportion of the burr segment in the overall regional contour; The patient abnormality verification end confirms the last breast X-ray image of the corresponding associated patient based on the focal area determined in the breast X-ray image of the associated patient, and simultaneously determines the focal area from the last breast X-ray image in the same way, and verifies and analyzes the two confirmed focus areas to assess whether there are any abnormal changes in the associated patient, and displays them in real time.

Images