Method and device for determining tumor edge for medical image
By combining MRI and CT images and using segmentation and relative scale change functions to correct the CT images, the problem of difficult tumor edge determination was solved, and accurate tumor edge identification was achieved.
Patent Information
- Application Number
- CN202410940017.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-15
- Publication Date
- 2024-11-08
AI Technical Summary
Existing technologies struggle to accurately determine tumor margins, especially in MR imaging where the boundaries between normal and tumor tissues intersect, resulting in CT images that do not provide sufficient detail.
By combining MRI and CT images, segmentation methods, gradient changes, and relative scale change functions are used to correct CT images to accurately locate tumor edges, and similarity and difference values are combined to determine tumor edges in MR images.
It enables precise identification of tumor edges, improving the accuracy and precision of tumor edge identification.
Smart Images

Figure CN118918131A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical imaging technology, and in particular to a method and device for determining a tumor margin in medical imaging. Background Art
[0002] Image tumor recognition and segmentation technology are of great significance in disease diagnosis and preoperative prevention.
[0003] Although MR (magnetic resonance imaging) imaging can show relatively rich details of tumors, for some tumor areas, the boundaries between normal tissue and tumor tissue are intertwined, making it difficult to accurately determine the edge of the tumor; although CT images can assist in determining the edge contour of the tumor to a certain extent, they do not show rich details enough. How to improve the recognition of tumors has become a technical problem that needs to be solved urgently. Summary of the invention
[0004] Based on this, an object of the present invention is to provide a method and device for determining tumor margins in medical images, so as to accurately identify the margins of tumors based on MRI images and CT images.
[0005] In a first aspect, an embodiment of the present invention provides a method for determining a tumor edge in a medical image, the method comprising: S102: segmenting an initial grayscale CT image Pc and an initial grayscale MR image Pm based on a preset segmentation method to obtain a corresponding plurality of feature regions; S104: determining the coordinates of a point column of a contour of the feature region based on a gradient change in pixel values of pixel points of the feature region; S106: calculating a relative scale change function S between the initial grayscale CT image Pc and the initial grayscale MR image Pm based on the coordinates of the point column of the contour of the feature region, wherein the relative scale change function S is divided into three scales: transverse, longitudinal and angular; S108: correcting the initial grayscale CT image Pc according to the relative scale change function S to obtain a corrected CT image Pc'; S110: determining the tumor edge of the initial grayscale MR image Pm based on the preset similarity and difference values between the tumor-like regions in Pm and Pc' and referring to the edge of the tumor in the corrected CT image Pc'.
[0006] Further, S102 includes: S102-2: for Pc and Pm, respectively establish a two-dimensional coordinate system of the image with the upper left corner of the image as the coordinate center point, the horizontal side as the X-axis, and the vertical side as the Y-axis; S102-4: based on the established two-dimensional coordinate system, evenly divide Pc and Pm into j feature areas respectively to obtain Pc1, Pc2, Pc3...Pcj and Pm1, Pm2, Pm3...Pmj.
[0007] Further, S104 includes: S104-2: respectively utilizing the gradient changes of the pixel values of the pixel points of the inner contour and outer contour of the images Pc1, Pc2, Pc3...Pcj to obtain j groups of outer contour point sequences; S104-4: similarly, respectively utilizing the gradient changes of the pixel values of the pixel points of the inner contour and outer contour of the images Pm1, Pm2, Pm3...Pmj to obtain j groups of outer contour point sequences; S104-6: by proportionally expanding and reducing the number of contour points, the number of p1, p2, p3...pj, q1, q2, q3,...qj is unified to a.
[0008] Further, S106 includes: S106-2: the coordinates of the point sequence of the outer contour corresponding to the first group of feature areas, obtaining the first scale transformation function S1 (x, y, θ); S106-4: similarly, the scale change functions of the other three feature areas can be calculated: S2 (x, y, θ), S3 (x, y, θ), S4 (x, y, θ).
[0009] Further, S106-2 includes: 1) a horizontal scale change sequence S1(x): S1(x) = Pm1(x)-Pc1(x)...Formula 1); wherein Pm1(x) represents a data set of horizontal coordinates of all points in Pm1; Pc(x) is a data set of horizontal coordinates of all points in Pc1, and S1(x) is a data set after the two are subtracted; 2) a vertical scale change sequence S1(y): S1(y) = Pm1(y)-Pc1(y)...Formula 2); wherein Pm1(y) represents a data set of coordinates of all points in Pm1; Pc1(y) is a data set of vertical coordinates of all points in Pc1, and S1(y) is a data set after the two are subtracted. Set; 3) Angle change sequence S1(θ): θc = arctan(|Pc1(y)| / |Pc1(x)|)......Formula 3); θm = arctan(|Pm1(y)| / |Pm1(x)|)......Formula 4); S1(θ) = θc-θm......Formula 5); 4) Combine S1(x), S1(y), S1(θ) to obtain the regional scale change function S1(x, y, θ) of the first feature area: S1(x, y, θ) = (S1(x) / pc1(x)+1)×S1(θ)+(S1(y) / Pc1(y)+1)×S1(θ)......Formula 6).
[0010] Further, S108 includes: calibrating the four feature areas of the initial grayscale CT image according to S1 (x, y, θ), S2 (x, y, θ), S3 (x, y, θ), and S4 (x, y, θ), and stitching them after calibration to obtain a corrected CT image Pc'.
[0011] Further, S110 includes: S110-2: selecting k tumor-like regions in Pm to obtain Pm1, Pm2...Pmk; S110-4: confirming the corresponding tumor-like regions Pc1', Pc2'...Pck' in Pc1' according to Pm1, Pm2...Pmk; S110-6: performing similarity calculation on k groups of tumor-like regions, and eliminating tumor-like regions corresponding to similarity values greater than a preset similarity threshold; S110-8: binarizing the retained tumor-like regions based on a preset binarization threshold to obtain an MR binary image Pmb and a CT binary image Pcb'; S110-10: performing differential calculation on the MR binary image and the CT binary image to obtain a difference value Diff; S110-12: using the CT binary image Pcb' corresponding to the difference value Diff greater than the preset difference threshold as a drawing basis to redraw the tumor edge in the initial grayscale MR image Pm.
[0012] Further, S110-6 includes: S110-6-2: Calculate the similarity Q1 between Pm1 and Pc1' according to the Pearson similarity coefficient formula: Among them, Hm and Hc are the grayscale values of the coordinates of Pm1 and Pc1'' respectively; μ1 and μ2 are the pixel means of Pm1 and Pc1' respectively; σ1 and σ2 are the grayscale standard deviations of Pm1 and Pc1' respectively, and E is the covariance of Pm1 and Pc1''; S110-6-4: Similarly, calculate the similarity coefficients Q2, Q3, ..., Qk between other tumor-like regions; S110-6-6: Eliminate e groups of tumor-like regions that are greater than or equal to the preset segmentation threshold, and retain f groups of tumor-like regions that are less than the preset segmentation threshold.
[0013] In a second aspect, the present invention provides a device for determining the tumor edge of medical images, the device comprising: a segmentation module, for segmenting an initial grayscale CT image Pc and an initial grayscale MR image Pm based on a preset segmentation method to obtain a corresponding plurality of feature areas; a coordinate confirmation module, for determining the coordinates of a point array of a contour of a feature area according to a gradient change of pixel values of pixel points of the feature area; a relative scale calculation module, for calculating a relative scale change function S between the initial grayscale CT image Pc and the initial grayscale MR image Pm based on the coordinates of the point array of the contour of the feature area, wherein the relative scale change function S is divided into three scales: transverse, longitudinal and angular; a correction module, for correcting the initial grayscale CT image Pc according to the relative scale change function S to obtain a corrected CT image Pc'; an edge confirmation module, for determining the tumor edge of the initial grayscale MR image Pm based on preset similarities and difference values between tumor-like areas in Pm and Pc', with reference to the edge of the tumor in the corrected CT image Pc'.
[0014] The beneficial effects of the embodiments of the present invention are as follows:
[0015] The present invention provides a method and device for determining the tumor margin of medical images; the method includes: segmenting the initial grayscale CT image Pc and the initial grayscale MR image Pm based on a preset segmentation method to obtain a plurality of corresponding feature regions; determining the coordinates of the point column of the contour of the feature region based on the gradient change of the pixel values of the pixel points of the feature region; calculating the relative scale change function S between Pc and Pm based on the coordinates of the point column of the contour of the feature region; correcting the initial grayscale CT image Pc according to the relative scale change function S to obtain the corrected CT image Pc'; based on the preset similarity and difference values between the tumor-like regions in Pm and Pc', referring to the edge of the tumor in Pc', determining the tumor margin of Pm. This technology combines MR images and CT images to more accurately identify the tumor margin.
[0016] Other features and advantages of the present invention will be set forth in the following description, or some features and advantages may be inferred or unambiguously determined from the description, or may be learned by implementing the above-mentioned technology of the present invention.
[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are specifically listed below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0019] Figure 1 is a flow chart of a method for determining tumor margins in medical imaging;
[0020] Figure 2 This is a CT image of a patient's brain;
[0021] Figure 3 This is an MR image of the brain of the same patient;
[0022] Figure 4 This is the CT image of the patient's brain after labeling;
[0023] Figure 5 Schematic diagram of segmentation into four feature regions. DETAILED DESCRIPTION
[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.
[0025] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. 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.
[0026] Embodiment 1
[0027] The present invention provides a method for determining tumor margins in medical images. The flowchart of the method is as follows: Figure 1 As shown, including:
[0028] S102: Segment the initial grayscale CT image Pc and the initial grayscale MR image Pm based on a preset segmentation method to obtain corresponding multiple feature regions.
[0029] Specifically, Figure 2 is the initial grayscale CT image; Figure 3 is the initial grayscale MR image.
[0030] Specifically, the initial grayscale CT image is denoted as Pc, and the initial grayscale MR image is denoted as Pm.
[0031] S102 includes:
[0032] S102-2: For Pc and Pm, a two-dimensional coordinate system of the image is established with the upper left corner of the image as the coordinate center point, the horizontal side as the X-axis, and the vertical side as the Y-axis.
[0033] Specifically, due to the differences in resolution and position between Pc and Pm, the coordinate values of lesions in the same position area are different in Pc and Pm, so it is necessary to perform position matching calibration on the CT and MR images.
[0034] S102-4: Based on the established two-dimensional coordinate system, Pc and Pm are evenly divided into j feature regions respectively to obtain Pc1, Pc2, Pc3...Pcj and Pm1, Pm2, Pm3...Pmj.
[0035] Specifically, Figure 5This is a schematic diagram of segmentation into 4 feature regions, i.e., j=4. It can also be segmented into 6, 8, etc. feature regions. The more regions are segmented and calculated, the smaller the error caused by the image distortion itself, and the more accurate the recognition.
[0036] S104: Determine the coordinates of the point sequence of the contour of the feature area according to the gradient change of the pixel values of the pixel points of the feature area.
[0037] Specifically, this step is to obtain the coordinates of the outer contour of the feature area according to the inner contour. This step can make the correction process more accurate.
[0038] S104-2: Obtain j groups of outer contour point sequences using the gradient changes of the pixel values of the inner contour and outer contour pixel points of the images Pc1, Pc2, Pc3, ..., Pcj, respectively.
[0039] Specifically, the point sequences of the outer contour of the CT image are recorded as: Pc1(T1, T2, T3...Tp1), p1 is the number of outer contour points of Pc1; Pc2(T1, T2, T3...Tp2), p2 is the number of outer contour points of Pc2; Pc3(T1, T2, T3...Tp3), p3 is the number of outer contour points of Pc3...Pcj(T1, T2, T3...Tpj), pj is the number of outer contour points of Pcj, and T represents the two-dimensional coordinates (x, y) of each point in the outer contour.
[0040] S104-4: Similarly, use the gradient changes of the pixel values of the pixel points of the inner contour and outer contour of the images Pm1, Pm2, Pm3...Pmj to obtain the point sequences of j groups of outer contours.
[0041] Specifically, the point series of the outer contour of the MR image are recorded as: Pm1 (T1, T2, T3...Tq1), q1 is the number of outer contour points of Pm1; Pm2 (T1, T2, T3...Tq2), q2 is the number of outer contour points of Pm2; Pm3 (T1, T2, T3...Tq3), q3 is the number of outer contour points of Pm3...Pmj (T1, T2, T3...Tqj), qj is the number of outer contour points of Pmj, and T represents the two-dimensional coordinates (x, y) of each point in the outer contour.
[0042] S104-6: By proportionally expanding and reducing the number of contour points, the number of p1, p2, p3, ..., pj, q1, q2, q3, ..., qj is unified to a.
[0043] Specifically, the number of all collected outer contour points is the same, so that subsequent calculations can be performed. At this point, the corresponding 2*j groups of outer contour coordinates T can be obtained, with a coordinate points in each group.
[0044] S106: Calculate the relative scale change function S between the initial grayscale CT image Pc and the initial grayscale MR image Pm based on the coordinates of the point sequence of the contour of the feature area, wherein the relative scale change function S is divided into three scales: transverse, longitudinal and angular.
[0045] S106-2: Obtain the first scale transformation function S1 (x, y, θ) from the coordinates of the point sequence of the outer contour corresponding to the first set of feature regions.
[0046] The specific process is as follows:
[0047] 1) Horizontal scale change sequence S1(x):
[0048] S1(x)=Pm1(x)-Pc1(x)......Formula 1);
[0049] Specifically, Pm1(x) represents the data set of the horizontal coordinates of all points in Pm1; Pc(x) is the data set of the horizontal coordinates of all points in Pc1, and S1(x) is the data set after the two are subtracted.
[0050] 2) Longitudinal scale change sequence S1(y):
[0051] S1(y)=Pm1(y)-Pc1(y)...Formula 2);
[0052] Specifically, Pm1(y) represents the data set of coordinates of all points in Pm1; Pc1(y) is the data set of ordinates of all points in Pc1, and S1(y) is the data set after subtracting the two.
[0053] 3) Angle change sequence S1(θ):
[0054] θc=arctan(|Pc1(y)| / |Pc1(x)|)...Formula 3);
[0055] θm=arctan(|Pm1(y)| / |Pm1(x)|)......Formula 4);
[0056] S1(θ)=θc-θm...Formula 5).
[0057] 4) Combine S1(x), S1(y), and S1(θ) to obtain the regional scale change function S1(x, y, θ) of the first feature area:
[0058] S1(x,y,θ)=(S1(x) / pc1(x)+1)×S1(θ)+(S1(y) / Pc1(y)+1)×S1(θ)...Formula 6).
[0059] Specifically, S1 (x, y, θ) is a data set, indicating that the edge coordinates of each organ (in this embodiment, the brain) have a corresponding scale change value, also called a correction value.
[0060] S106-4: Similarly, the scale change functions of the other three feature areas can be calculated: S2(x, y, θ), S3(x, y, θ), and S4(x, y, θ).
[0061] S108: Correcting the initial grayscale CT image Pc according to the relative scale change function S to obtain a corrected CT image Pc'.
[0062] Specifically, the coordinate values of the initial grayscale CT image are multiplied by the corresponding scale transformation function S to obtain new coordinate values, and the new coordinate value combination is the corrected CT image Pc'.
[0063] Specifically, S108 includes: calibrating the four feature areas of the initial grayscale CT image according to S1 (x, y, θ), S2 (x, y, θ), S3 (x, y, θ), and S4 (x, y, θ), and stitching them after calibration to obtain a corrected CT image Pc'.
[0064] S110: Based on the preset similarities and difference values between the tumor-like regions in Pm and Pc′, and with reference to the tumor edge in the corrected CT image Pc′, determine the tumor edge of the initial grayscale MR image Pm.
[0065] S110-2: Select k tumor-like regions in Pm to obtain Pm1, Pm2...Pmk.
[0066] Specifically, the tumor-like region can be rectangular and can be manually selected or machine-recognized. The purpose is to frame the area where the tumor is preliminarily determined to exist. The doctor manually selects the area where the doctor believes there is a tumor, i.e., the tumor-like region.
[0067] S110-4: According to Pm1, Pm2...Pmk, confirm the corresponding tumor-like regions Pc1', Pc2'...Pck' in Pc1'.
[0068] Specifically, since the coordinate systems of Pm and Pc' have been matched and calibrated, the same segmentation method can be used to crop the corrected CT image Pcb' to obtain the corresponding images Pc1', Pc2'...Pck'. At this point, k groups of tumor-like regions can be obtained.
[0069] S110-6: Calculate the similarity of the k groups of tumor-like regions, and eliminate the tumor-like regions corresponding to the similarity greater than a preset similarity threshold.
[0070] S110-6-2: Calculate the similarity Q1 between Pm1 and Pc1' according to the Pearson similarity coefficient formula.
[0071]
[0072] Among them, Hm and Hc are the grayscale values of the coordinates of Pm1 and Pc1'' respectively; μ1 and μ2 are the pixel means of Pm1 and Pc1' respectively; σ1 and σ2 are the grayscale standard deviations of Pm1 and Pc1' respectively, and E is the covariance of Pm1 and Pc1''.
[0073] S110-6-4: Similarly, similarity coefficients Q2, Q3, ..., Qk between other tumor-like regions are calculated.
[0074] S110-6-6: Eliminate the e group of tumor-like regions that are greater than or equal to the preset segmentation threshold, and retain the f group of tumor-like regions that are less than the preset segmentation threshold.
[0075] Specifically, e+f=k.
[0076] S110 - 8 : Based on a preset binarization threshold, the retained tumor-like region is binarized to obtain an MR binarized image Pmb and a CT binarized image Pcb′.
[0077] S110-10: Perform a difference calculation on the MR binary image and the CT binary image to obtain a difference value Diff.
[0078] Diff=|Rmb-Rcb'|......Formula 8);
[0079] Wherein, Rmb is the pixel mean value of the MR binary image Pmb, and Rcb' is the pixel mean value of the CT binary image Pcb'.
[0080] Specifically, by retaining several groups of tumor-like regions, we can get several difference values Diff
[0081] S110 - 12 : Using the CT binary image Pcb′ corresponding to the difference value Diff greater than the preset difference threshold as a drawing basis to redraw the tumor edge in the initial grayscale MR image Pm.
[0082] Specifically, the white dots in the CT binary image Pcb" in the depiction basis are used as the contour of the tumor, and the black dots are non-contour areas, so that the edge of the tumor in the nuclear magnetic resonance image can be accurately depicted.
[0083] This technology combines MR images and CT images to more accurately identify tumor edges.
[0084] Embodiment 2
[0085] An embodiment of the present invention provides a device for determining a tumor margin in a medical image, comprising:
[0086] A segmentation module, used to segment the initial grayscale CT image Pc and the initial grayscale MR image Pm based on a preset segmentation method to obtain corresponding multiple feature regions;
[0087] The coordinate confirmation module is used to determine the coordinates of the point sequence of the contour of the feature area according to the gradient change of the pixel values of the pixel points of the feature area.
[0088] The relative scale calculation module is used to calculate the relative scale change function S between the initial grayscale CT image Pc and the initial grayscale MR image Pm based on the coordinates of the point column of the contour of the feature area, wherein the relative scale change function S is divided into three scales: horizontal, longitudinal and angular.
[0089] The correction module is used to correct the initial grayscale CT image Pc according to the relative scale change function S to obtain the corrected CT image Pc'.
[0090] The edge confirmation module is used to determine the tumor edge of the initial grayscale MR image Pm based on the similarity and difference values between the preset tumor-like regions in Pm and Pc' and with reference to the edge of the tumor in the corrected CT image Pc'.
[0091] The device for determining the tumor margin of a medical image provided in an embodiment of the present invention has the same implementation principle and technical effects as those of the aforementioned method for determining the tumor margin of a medical image. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.
[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for determining tumor margins in medical images, characterized in that: The method comprises: S102: Segmenting the initial grayscale CT image Pc and the initial grayscale MR image Pm based on a preset segmentation method to obtain corresponding multiple feature regions; S104: determining the coordinates of a point column of a contour of the feature area according to a gradient change of pixel values of pixel points of the feature area; S106: Calculating a relative scale change function S between the initial grayscale CT image Pc and the initial grayscale MR image Pm based on the coordinates of the point sequence of the contour of the feature region, wherein the relative scale change function S is divided into three scales: transverse, longitudinal and angular; S108: Correcting the initial grayscale CT image Pc according to the relative scale change function S to obtain a corrected CT image Pc'; S110: Based on the preset similarities and difference values between the tumor-like regions in Pm and Pc′, and with reference to the tumor edge in the corrected CT image Pc′, determine the tumor edge of the initial grayscale MR image Pm.
2. The method for determining tumor margins in medical images according to claim 1, characterized in that: S102 includes: S102-2: For Pc and Pm, a two-dimensional coordinate system of the image is established with the upper left corner of the image as the coordinate center point, the horizontal side as the X axis, and the vertical side as the Y axis; S102-4: Based on the established two-dimensional coordinate system, Pc and Pm are evenly divided into j feature regions respectively to obtain Pc1, Pc2, Pc3...Pcj and Pm1, Pm2, Pm3...Pmj.
3. The method for determining tumor margins in medical images according to claim 2, characterized in that: S104 includes: S104-2: Obtain j groups of outer contour point sequences by using the gradient changes of the pixel values of the inner contour and the outer contour of the images Pc1, Pc2, Pc3, ..., Pcj, respectively; S104-4: Similarly, using the gradient changes of the pixel values of the inner contours and the outer contours of the images Pm1, Pm2, Pm3, ..., Pmj, obtain j groups of outer contour point sequences; S104-6: By proportionally expanding and reducing the number of contour points, the number of p1, p2, p3, ..., pj, q1, q2, q3, ..., qj is unified to a.
4. The method for determining tumor margins in medical images according to claim 3, characterized in that: S106 includes: S106-2: Obtain the first scale transformation function S1 (x, y, θ) from the coordinates of the point sequence of the outer contour corresponding to the first set of feature regions; S106-4: Similarly, the scale change functions of the other three feature areas can be calculated: S2(x, y, θ), S3(x, y, θ), and S4(x, y, θ).
5. The method for determining tumor margins in medical images according to claim 4, characterized in that: S106-2 includes: 1) Horizontal scale change sequence S1(x): S1(x)=Pm1(x)-Pc1(x)......Formula 1); Among them, Pm1(x) represents the data set of the horizontal coordinates of all points in Pm1; Pc(x) is the data set of the horizontal coordinates of all points in Pc1, and S1(x) is the data set after the two are subtracted; 2) Longitudinal scale change sequence S1(y): S1(y)=Pm1(y)-Pc1(y)...Formula 2); Among them, Pm1(y) represents the data set of coordinates of all points in Pm1; Pc1(y) is the data set of ordinates of all points in Pc1, and S1(y) is the data set after subtracting the two; 3) Angle change sequence S1(θ): θc=arctan(|Pc1(y)| / |Pc1(x)|)...Formula 3); θm=arctan(|Pm1(y)| / |Pm1(x)|)......Formula 4); S1(θ)=θc-θm...Formula 5); 4) Combine S1(x), S1(y), and S1(θ) to obtain the regional scale change function S1(x, y, θ) of the first feature area: S1(x,y,θ)=(S1(x) / pc1(x)+1)×S1(θ)+(S1(y) / Pc1(y)+1)×S1(θ)...Formula 6).
6. The method for determining tumor margins in medical images according to claim 4, characterized in that: S108 includes: The four feature areas of the initial grayscale CT image are calibrated according to S1(x, y, θ), S2(x, y, θ), S3(x, y, θ), and S4(x, y, θ). After calibration, they are spliced to obtain the corrected CT image Pc'.
7. The method for determining tumor margins in medical images according to claim 4, characterized in that: S110 includes: S110-2: Select k tumor-like regions in Pm to obtain Pm1, Pm2...Pmk; S110-4: confirm the corresponding tumor-like regions Pc1', Pc2'...Pck' in Pc1' according to Pm1, Pm2...Pmk; S110-6: Calculating the similarity of the k groups of tumor-like regions, and eliminating the tumor-like regions corresponding to those with a similarity greater than a preset similarity threshold; S110-8: Based on a preset binarization threshold, the retained tumor-like region is binarized to obtain an MR binarization image Pmb and a CT binarization image Pcb'; S110-10: performing a difference calculation on the MR binary image and the CT binary image to obtain a difference value Diff; S110 - 12 : Using the CT binary image Pcb′ corresponding to the difference value Diff greater than the preset difference threshold as a drawing basis to redraw the tumor edge in the initial grayscale MR image Pm.
8. The method for determining tumor margins in medical images according to claim 7, characterized in that: S110-6 includes: S110-6-2: Calculate the similarity Q1 between Pm1 and Pc1' according to the Pearson similarity coefficient formula: Among them, Hm and Hc are the grayscale values of the coordinates of Pm1 and Pc1'' respectively; μ1 and μ2 are the pixel means of Pm1 and Pc1' respectively; σ1 and σ2 are the grayscale standard deviations of Pm1 and Pc1' respectively, and E is the covariance of Pm1 and Pc1''; S110-6-4: Similarly, calculate the similarity coefficients Q2, Q3, ..., Qk between other tumor-like regions; S110-6-6: Eliminate the e group of tumor-like regions that are greater than or equal to the preset segmentation threshold, and retain the f group of tumor-like regions that are less than the preset segmentation threshold.
9. A device for determining tumor margins in medical images, characterized in that: The device comprises: A segmentation module, used to segment the initial grayscale CT image Pc and the initial grayscale MR image Pm based on a preset segmentation method to obtain corresponding multiple feature regions; A coordinate confirmation module, used to determine the coordinates of the point column of the contour of the feature area according to the gradient change of the pixel values of the pixel points of the feature area; A relative scale calculation module is used to calculate the relative scale change function S between the initial grayscale CT image Pc and the initial grayscale MR image Pm based on the coordinates of the point array of the contour of the feature area, wherein the relative scale change function S is divided into three scales: transverse, longitudinal and angular; A correction module, used for correcting the initial grayscale CT image Pc according to the relative scale change function S to obtain a corrected CT image Pc'; The edge confirmation module is used to determine the tumor edge of the initial grayscale MR image Pm based on the similarity and difference values between the preset tumor-like regions in Pm and Pc' and with reference to the edge of the tumor in the corrected CT image Pc'.
Citation Information
Cited By
Stomach tumor cell image extraction method and system
CN119273687A
Preoperative planning system for tumor interventional therapy based on artificial intelligence image recognition
CN122025017A
