A method for constructing a three-dimensional electromagnetic simulation breast model based on an improved OTSU algorithm

Abstract

The present application relates to a kind of based on the method for constructing three-dimensional electromagnetic simulation breast model of improved OTSU algorithm, comprising the following steps: (1) by interpolation, stretch breast MRI image to match FDTD grid;(2) the image after stretching is carried out median filtering, then morphological open operation is completed pre-processing;(3) binaryzation pre-processing MRI image, preliminary breast boundary is obtained to the binaryzation image traversal, select the largest connected domain as breast tissue boundary and from the breast picture after stretching breast tissue is intercepted;(4) each breast MRI slice image is stacked and interpolation is obtained three-dimensional pixel data;(5) according to breast boundary, complete air, thoracic cavity, skin setting, using improved OTSU algorithm, complete breast internal tissue division;(6) to different tissue is endowed with corresponding electrical characteristic value, obtains three-dimensional breast electromagnetic simulation model.

Description

Technical Field

[0001] This invention belongs to the field of biomedical detection technology and relates to a method for constructing a three-dimensional electromagnetic simulation breast model based on an improved OTSU algorithm. Background Technology

[0002] Breast cancer has become the leading malignant tumor threatening women's health, with its incidence rate rising annually worldwide. Early detection helps patients receive timely and appropriate treatment, improving survival rates. Currently, breast cancer detection methods mainly include mammography, computed tomography (CT), breast ultrasound, magnetic resonance imaging (MRI), and infrared thermography. These methods all have certain drawbacks, such as potential radiation risks preventing repeated examinations in a short period, damage from breast compression, high misdiagnosis rates, and high costs. Ultra-wideband microwave breast tumor detection technology, due to the advantages of ultra-wideband microwaves—strong tissue penetration, large information carrying capacity, and low radiation power—can achieve non-destructive testing with no ionizing radiation, low cost, and high accuracy, making it a viable method for breast tumor detection. This method uses an antenna array to emit ultra-wideband electromagnetic waves to irradiate the target object, receives scattered signals containing tumor information, and processes these scattered signals to achieve breast tumor imaging. Currently, this technology is still in the experimental research stage and has not yet reached the stage of becoming a clinical detection method.

[0003] The experimental research phase of ultra-wideband microwave breast tumor imaging technology requires the construction of a breast numerical model for electromagnetic simulation as a comprehensive platform for computational research and method optimization. Early simulation studies typically used two-dimensional models or relatively simple and symmetrical three-dimensional models, such as cylindrical, hemispherical, and square models. These models lack a realistic breast outline and do not consider the internal glandular tissue. This invention accurately extracts breast tissue based on boundary recognition from clinical MRI images, uses an improved OTSU algorithm to precisely segment the internal breast tissue, and assigns corresponding electrical property values ​​to various tissues. Morphologically, it realistically reflects the shape and internal tissues of the breast, constructing a more realistic three-dimensional electromagnetic simulation breast model that is beneficial for advancing experimental research on ultra-wideband microwave breast tumor imaging technology. Summary of the Invention

[0004] The purpose of this invention is to provide a method for constructing a three-dimensional electromagnetic simulation breast model based on an improved OTSU algorithm. The technical solution of this invention is as follows:

[0005] (1) Each breast MRI slice was stretched by interpolation to match the FDTD grid;

[0006] (2) Median filtering is applied to the stretched MRI images to remove noise and smooth the data, which is beneficial for the extraction of the boundary lines of breast tissue. Morphological opening operation is performed on the sliced ​​images by selecting the size of the disk-shaped structural element to disconnect the connection between the thoracic cavity and breast tissue, thus completing the preprocessing of the MRI images.

[0007] (3) MRI images after binarization preprocessing; traversing the binarized images pixel by pixel to extract the initial breast boundary; extracting the largest connected component in the initial boundary as the final breast tissue boundary, filling the boundary image with binary to obtain a mask, and multiplying it pixel by pixel with the stretched breast image to obtain the breast tissue portion;

[0008] (4) Stack the MRI slices containing only breast tissue after masking extraction in sequence and perform bicubic interpolation to obtain the three-dimensional pixel data of the breast.

[0009] (5) The obtained complete boundary contour is divided into skin boundary line and chest cavity contour line. The area outside the skin boundary line is set as air, the area outside the chest cavity contour line is set as chest cavity, and the skin boundary line and its small neighboring pixel area are set as skin. The breast tissue is first divided into glands and fat by calculating the threshold using the improved OTSU algorithm. Then, the glands and fat are further divided into four types of breast tissue: gland 1, gland 2, fat 1 and fat 2.

[0010] (6) Due to the difference in the proportion of glandular tissue and fat in breast tissue, the variances of fat and glandular tissue differ significantly. Using the OTSU algorithm would cause the calculated threshold to be biased towards the tissue with larger variance. Therefore, the OTSU algorithm was improved by introducing a correction factor K into the inter-class variance. The correction factor K is the roughly calculated proportion of fat tissue in breast tissue. The specific formulas for selecting the inter-class variance and the threshold are as follows:

[0011] Between-class variance:

[0012] Threshold selection:

[0013] The inter-class variances corresponding to all k values ​​are calculated, and the k value corresponding to the maximum inter-class variance is the threshold T.

[0014] (7) Assign corresponding electrical property values ​​to the four types of breast tissue, air and pleural cavity obtained by division to obtain a three-dimensional electromagnetic simulation model of the breast. Detailed Implementation

[0015] The invention will now be described with reference to specific examples and accompanying drawings:

[0016] (1) MRI source image stretching: such asFigure 1 As shown, the acquired clinical MRI source image has a resolution of 512x512, corresponding to an actual length of 30x30cm. Calculations show that each pixel corresponds to a grid size of 0.586x0.586mm. To match the 0.5x0.5mm grid size used by FDTD, an image interpolation function was used to stretch the MRI source image to 600×600 pixels, as shown below. Figure 2 As shown;

[0017] (2) Preprocessing before breast boundary extraction: Two-dimensional median filtering is performed on the stretched breast image, with the neighborhood size set to 3x3, meaning that each output pixel is the median value of the corresponding pixel in the input image's 3x3 neighborhood. This smooths the data and removes noise from the image, which is beneficial for subsequent breast tissue boundary extraction. Then, a disk-shaped structuring element with a radius of 3 is created to perform morphological opening operations on the image, disconnecting the connection between the thoracic cavity and breast tissue, thus completing the preprocessing of the MRI image, such as... Figure 3 As shown;

[0018] (3) Boundary and MRI breast tissue extraction: MRI images after binarization preprocessing; preliminary boundaries are obtained through the binarized image. The specific steps are: create a zero-value matrix of the same size as the binarized image to store the preliminary boundaries, and traverse the binary logical matrix pixel by pixel; when a true value is encountered in the binary matrix, assign the values ​​of the remaining 8 pixels in the 3x3 neighborhood of that pixel to the corresponding positions in the zero-value matrix. After the traversal is completed, the preliminary boundaries can be obtained, such as... Figure 4 As shown; the largest connected component in the initial boundary is extracted as the final breast tissue boundary. Figure 5 A schematic diagram of the final breast boundary drawn in the stretched breast diagram; the boundary... Figure 2 Value filling is used to obtain a mask, which is then multiplied pixel-by-pixel with the stretched breast image to extract the breast tissue portion, such as... Figure 6 As shown;

[0019] (4) MRI slice stacking interpolation: The MRI slices containing only breast tissue after masking extraction are stacked sequentially. In this case, the patient had a total of 26 MRI slice images, which were stacked to obtain 600x600x26 three-dimensional pixel data. The spacing between each MRI slice is 5mm. To match the FDTD three-dimensional grid with a size of 0.5x0.5x0.5mm, interpolation is required in the stacking direction. Each sagittal slice with a size of 600x26 in the three-dimensional pixel data is interpolated to a size of 600x260 using bicubic interpolation, such as... Figure 7 As shown; after interpolation, a 600x600x260 three-dimensional pixel data of the breast is obtained. Figure 8 This is a three-dimensional breast after stacking and interpolation;

[0020] (5) Breast tissue segmentation: The obtained complete boundary contour is segmented into skin boundary line and chest cavity contour line. The area outside the skin boundary line is set as air, and the area outside the chest cavity contour line is set as chest cavity. All pixels on the skin boundary line and all pixels in its 3x3x3 neighborhood are set as skin. The breast tissue is divided into glands G1 and fat F1 by calculating the threshold using the improved OTSU algorithm. Gland G1 is divided into glands G11 and G12 by calculating the threshold using the improved OTSU algorithm. Fat F1 is divided into fat F11 and F12 by calculating the threshold using the improved OTSU algorithm. Four types of breast tissue are obtained, such as... Figure 9 As shown;

[0021] (6) Due to the difference in the proportion of glandular tissue and fat in breast tissue, the variances of fat and glandular tissues differ significantly. Using the OTSU algorithm would cause the calculated threshold to be biased towards the tissue with higher variance. Therefore, the OTSU algorithm was improved by introducing a correction factor K into the inter-class variance. By drawing a histogram of breast tissue, two peaks and their corresponding pixel values ​​were obtained. The average of these two pixel values ​​was used as the threshold for roughly segmenting the breast and glandular tissues. This threshold was then used to calculate the proportion of fat tissue in breast tissue to obtain the correction factor K. The specific formulas for selecting the inter-class variance and the threshold are as follows:

[0022] Between-class variance:

[0023] Threshold selection:

[0024] The inter-class variances corresponding to all k values ​​are calculated, and the k value corresponding to the maximum inter-class variance is the threshold T.

[0025] (7) Assigning electrical properties: Assigning corresponding electrical property values ​​to the four types of breast tissue, air, and pleural cavity obtained from the division according to Table 1, to obtain a three-dimensional electromagnetic simulation model of the breast, such as... Figure 10 As shown.

[0026] Table 1. Electrical properties of different tissues in the breast model.

[0027] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application. Attached Figure Description

[0028] Figure 1 Breast-derived image from clinical MRI.

[0029] Figure 2 Image of a stretched breast.

[0030] Figure 3 Breast image after median filtering and morphological opening operation.

[0031] Figure 4 Preliminary breast boundaries.

[0032] Figure 5 Extract the final boundary and plot it on the stretched breast map.

[0033] Figure 6 Extracting breast tissue.

[0034] Figure 7 Sagittal slice image after interpolation.

[0035] Figure 8 3D breast after stacking and interpolation.

[0036] Figure 9 Image of the breast after tissue division.

[0037] Figure 10 Breast tissue is endowed with electrical properties.

Claims

1. A method for constructing a three-dimensional electromagnetic simulation breast model based on an improved OTSU algorithm, characterized in that, The method comprises the following steps: (1) stretching the breast MRI image by interpolation to match the grid of finite difference time domain (FDTD); (2) performing median filtering on the MRI image to remove noise in the image and smooth the data to facilitate extraction of edge features in the image; performing morphological opening operation on the slice image by selecting a disc-shaped structural element of a certain size to disconnect the connection between the chest cavity and the breast tissue; (3) binarizing the preprocessed MRI image; extracting a preliminary breast boundary by traversing the binarized image pixel by pixel; extracting the maximum connected domain as the breast tissue boundary; performing binary filling on the boundary image to obtain a mask, and multiplying the stretched breast image pixel by pixel to obtain the breast tissue part; (4) stacking the MRI slice images containing only breast tissue after mask extraction, and performing bicubic interpolation to obtain three-dimensional pixel data of the breast; (5) dividing the obtained complete boundary into a skin boundary line and a chest cavity contour line, setting the pixels outside the skin boundary line as air, setting the pixels outside the chest cavity contour line as the chest cavity, and setting the skin boundary line and its small neighborhood pixels as skin; in order to accurately divide the breast tissue, the maximum inter-class variance method (OTSU) is improved; the breast tissue is divided into gland and fat by calculating the threshold value through the improved OTSU algorithm; the gland and fat are further divided into gland 1, gland 2, fat 1 and fat 2 by calculating the threshold value through the improved OTSU algorithm; (6) assigning corresponding electrical characteristic values to different tissues to obtain a three-dimensional breast electromagnetic simulation model; In step (5), the OTSU algorithm is improved, and a correction factor K is introduced into the inter-class variance; the specific formula for selecting the threshold value is: Interclass variance: σ 2 = p A (k)(m A (k)-m G ) 2 + Kp B (k)(m B (k)-m G ) 2 ; Threshold selection T = MAXσ 2 ; The k value corresponding to the maximum inter-class variance is the threshold value T.

Images