CN105389815A - Breast image registration method and device - Google Patents

Abstract

The invention discloses a breast image registration method, comprising: inputting an image to be registered in a symmetrical view; obtaining breast contour and body position information of the image to be registered; The registration point; the images are registered according to the registration point. Also disclosed is a breast image registration device. The method and device provided by the invention can improve image registration quality under the premise of controlling calculation amount and satisfying real-time registration requirements.

Description

A breast image registration method and device

【Technical field】

The invention relates to the field of medical image processing, in particular to a registration method and device for breast images.

【Background technique】

Mammography uses professional imaging equipment to take pictures of the left and right breasts in different positions to form breast images. Due to the uncertainty of the shooting position, there will be differences in position and size between the breast images of the left and right view pairs. This difference will affect the comparative analysis results. In order to quickly locate the lesion area, the doctor hopes to register the left and right view images of the same body position to a symmetrical position as much as possible, so as to facilitate comparative analysis.

To meet this requirement, in the prior art, methods for registering and aligning breast images mainly include:

1. The method based on image gray level distribution: take one mammary gland image as a reference image, and use the other image as a floating image to traverse the reference image, and find the position with the largest cross-correlation information feature (entropy) of the two images as the matching image. On point, this method has to deal with two breast images of large size, the amount of calculation is very large, and it is difficult to meet the real-time requirements.

2. The method based on the geometric characteristics of the image organization area: that is, select points such as centroid and centroid of the mammary gland image as alignment (registration) points (such as Chinese patent application CN201310732363. The methods disclosed in "Registration Method and Registration System"), the accuracy of this type of method is poor, and the existing errors often cause visual differences, and the registration effect is not ideal.

Therefore, there is a need for a new breast image registration method and device, which can improve image registration quality under the premise of controlling the calculation amount and meeting the real-time registration requirements.

【Content of invention】

The invention solves the problems that the existing mammary gland image registration method has a large amount of calculation, cannot meet the real-time requirement and has poor registration effect.

In order to solve the above problems, the present invention provides a breast image registration method, which includes: inputting an image to be registered in a symmetrical view; obtaining breast contour and body position information of the image to be registered; A registration point is determined on the contour; images are registered according to the registration point.

Optionally, obtaining the breast contour includes: segmenting the breast tissue image in the image to be registered; and obtaining the breast contour by using a chain code tracking method.

Optionally, the segmentation is threshold segmentation.

Optionally, the method further includes: removing invalid boundaries in the breast contour.

Optionally, said determining the registration points on the breast contour of the symmetric view according to the body position information includes: selecting a point in the breast contour farthest from the symmetry axis as a registration point or selecting a fitting circle with the smallest radius of the breast contour The center of the circle is used as the registration point.

Optionally, said determining the registration points on the breast contour of the symmetric view according to the body position information includes: performing a first judgment, judging whether the body position information of the image to be registered is the first body position: if so, using the first A method determines the registration point; if not, a second method is used to determine the registration point.

Optionally, the first body position is a head-to-tail position; the first method includes: selecting a point in the breast contour farthest from the symmetry axis as a registration point.

Optionally, the second method includes: performing a second judgment, judging whether the radius of the fitting circle with the smallest radius of the breast contour is smaller than the diameter of the boundary circle of the nipple tissue; if so, selecting the center of the fitting circle with the smallest radius as a registration point; if not, then reduce the fitting circle step size of the breast contour, re-determine the minimum radius fitting circle and execute the second judgment, if the step size is reduced to less than a preset threshold , the radius of the fitting circle with the minimum radius is still not less than the diameter of the boundary circle of the nipple tissue, then the point in the breast contour farthest from the symmetry axis is selected as the registration point.

Optionally, the registration of the image according to the registration point includes: taking a registration point as a reference registration point, and moving the image including the rest of the registration points, and the moving distance is the distance between the rest of the registration points and the reference registration point. The difference between the longitudinal coordinates of the punctual point.

Correspondingly, the present invention also provides a breast image registration device for implementing the above-mentioned breast image registration method, including: an image input unit for inputting images to be registered; an image processing unit for processing the images to be registered The image is processed to obtain the breast contour and body position information of the image to be registered; the image registration unit is used to determine registration points on the breast contour of the symmetrical view according to the body position information, and perform image processing based on the registration points Registration.

Compared with the prior art, the present invention has the following beneficial effects:

The body position information of the breast image is fully utilized in combination with the characteristics of the imaging contour for registration, which reduces the amount of image registration calculations and ensures the registration quality. The solution of the present invention can also be implemented in a more optimal manner, by selecting different alignment methods for images of different body positions, and performing fault-tolerant processing on abnormal images, so as to better adapt to the specificity of images of different patients and ensure the efficiency of the algorithm.

【Description of drawings】

Fig. 1 is a schematic flow chart of an embodiment of the present invention;

Fig. 2 is a schematic diagram of the process of registration of mammary gland images in head and tail positions in an embodiment of the present invention;

Fig. 3 is a schematic diagram of the registration process of the medial and lateral oblique mammary gland images in an embodiment of the present invention;

Fig. 4 is a schematic diagram of a registration point determination process in an embodiment of the present invention.

【detailed description】

In order to make the above objects, features and advantages of the present invention more comprehensible, specific implementations of the present invention will be described in detail below in conjunction with the accompanying drawings.

In the following description, many specific details are set forth in order to fully understand the present invention, but the present invention can also be implemented in other ways than those described here, so the present invention is not limited by the specific embodiments disclosed below.

As shown in FIG. 1 , in an embodiment of the breast image registration method provided by the present invention, step S1 is first performed to input a breast image to be registered. In order to quickly locate the lesion area, the user hopes to be able to register the symmetrical views (referring to the views obtained in the symmetrical photography direction based on the same body position, such as the left and right side views) of the same body position to as symmetrical positions as possible to facilitate comparative analysis. In this embodiment, the input mammary gland image is illustrated by taking the mammary gland image of the left and right views based on the same body position as an example, as can be seen in FIG. The images to be registered and the images to be registered in the left and right views shown in FIG. The breast image may be a breast image obtained by mammography using imaging equipment such as a molybdenum target, a molybdenum-rhodium dual target, and a molybdenum-tungsten dual target. The way of image input can be directly transmitted and input after the medical imaging equipment completes the imaging, and can also be input in the way of controlling and reading the storage device. The format of the image may be a breast image conforming to the DICOM (Digital Imaging and Communications in Medicine, Digital Imaging and Communications in Medicine) standard.

Continuing to refer to FIG. 1 , after the image to be corrected is input, step S2 is performed to obtain breast contour and body position information of the image to be registered. Body position information refers to the body position information of the subject when the mammogram is taken. Among them, the medial and lateral oblique position and the head and tail position are the body positions routinely used in all mammography. Correspondingly, the left side view of the head and tail image can be defined as the LCC image. The right view is the RCC image; the left view of the medial-lateral oblique image is the LMLO image, and the right view is the RMLO image. Body position information and view information can be obtained through user input or image analysis. For example, for breast images conforming to the DICOM standard, body position information and view information can be obtained by reading DICOM information of the image.

The breast contour refers to the boundary information of the breast tissue in the breast image. It can be obtained by segmenting the breast image to be registered. For example, after the image to be registered is input, the background pixel value of the image to be registered is judged. If the background pixel value is zero, the non-zero pixel point can be directly read as the breast tissue image; if the background pixel value is non-zero, then the The image is segmented by threshold segmentation. Since the distribution of breast tissue is relatively concentrated and the difference from the background is large, the maximum between-class variance method (OSTU, also known as Otsu method) can be preferably used for segmentation. After the mammary gland image is segmented, the boundary of the mammary gland tissue (that is, breast contour) can be obtained by chain code (Freeman code) tracking, where the mammary gland contour of the left view is defined as chain code L, and the mammary gland contour of the right view is chain Code R.

As shown in Figure 2(b) and Figure 3(b), the breast contour obtained through the above steps may also contain some invalid boundary information. Boundaries are removed. In one embodiment of the present invention, it can be carried out by judging whether the breast boundary point is the image boundary or defining the effective height of the breast image, for example: defining the image boundary point coordinates as (x, y), if the boundary point coordinates have any value If it is equal to the corresponding coordinate value of the image boundary point, then the breast tissue boundary point is considered invalid, and it is filtered out. For another example, the effective height range of the MLO bit image can be set to be one-third of the image height to four-fifths of the image height, and the effective height range of the CC bit image can be set to one-third of the image height to two-thirds of the image height. The breast image boundaries outside the effective height range can be considered as invalid boundary information and filtered out. As shown in Fig. 2(c) and Fig. 3(c), after the invalid boundary is removed, the effective contour information of the breast image is obtained.

Continuing to refer to FIG. 1 , after obtaining the breast contour and body position information of the image to be registered, step S3 is performed to determine the registration point on the breast contour according to the body position information.

Breast images with different body position information have different breast contour features. For example, the breast contour of the CC image is hemispherical and symmetrically distributed up and down, and the nipple tissue is centered. Therefore, the nipple contour in the symmetrical view LCC and RCC is the farthest from the symmetry axis; For the non-CC position (such as MLO position) body image, the nipple tissue is located in the lower middle of the image, and the boundary of the cut nipple area is circular, and there are obvious inflection points with other boundaries. Therefore, by judging the body position information of the image to be registered, different registration methods can be used for registration of breast images of different body positions, so as to improve the efficiency and accuracy of registration. For example, the first registration method can be selected for CC breast images, and the second registration method can be selected for non-CC body positions. More specifically, for CC breast images, since the breast contour is symmetrically distributed up and down in a hemisphere, the nipple tissue is centered and farthest from the symmetry axis of the left and right views, so the point in the breast contour farthest from the symmetry axis is used as the registration point. Registration (as shown in Figure 2(d)); and for non-CC breast images, the center of the fitting circle with the smallest radius of the breast contour can be used as the registration point for registration (as shown in Figure 3(d)).

Fig. 4 shows the process of determining the registration points of non-CC mammary gland images. First, the breast contour is fitted with a circle, such as setting the step size, and using this step length to cycle through the selection of points on the breast contour Fitting circle, take the circle with the smallest radius among all fitting circles as the fitting circle with the smallest radius.

For example, the initial step size can be set to be DELTA (its value can preferably approach or be equal to the radius of the nipple tissue boundary circle), and the current point p, p+DELTA and p+DELTA*2 fits the circle, after the cycle is completed, the radius of the circle with the smallest radius is minRadius.

Then, judge the minimum radius, if the minimum radius minRadius<the first threshold (the first threshold can be 2 times the initial step value, that is, approaching or equal to the diameter of the papillary tissue boundary circle), then select the minimum radius The corresponding center of the circle is the registration point.

If the minimum radius minRadius is greater than or equal to the first threshold, it is considered that the found minimum radius circle is invalid, and the step size value is reduced (the reduction can be performed in a proportional or differential manner according to the actual situation and taking into account the calculation efficiency, for example, each time decreases 0.75 times the value of the previous step, or 0.1 times the initial step each time), and judge whether the new step value is less than the second threshold (the second threshold can be set to 0.5 times the initial step), if not less than For the second threshold, return to re-determine the minimum radius circle and its radius, and compare the radius with the first threshold again. If the new step size is reduced to be smaller than the minimum radius circle determined by the second threshold and is still invalid, use the point in the breast contour farthest from the symmetry axis (ie, the point with the largest symmetry axis distance) as the registration point.

Continuing to refer to FIG. 1 , after the registration point is determined, step S4 is performed to register the images according to the registration point. Specifically, one registration point can be used as a reference registration point, and the image containing the rest of the registration points can be moved. The moving distance is the difference between the remaining registration points and the longitudinal coordinates of the reference registration point. The registration point that is closer to the center of the registration image (or the symmetry axis of the left and right views) is the reference registration point, and the image containing the registration point that is farther away is moved, and the moving distance is the distance between the registration point and the longitudinal coordinate of the reference registration point. Difference.

The images after registration are shown in Figure 2(e) and Figure 3(e).

The breast image registration method provided by the present invention makes full use of the body position information of the breast image combined with the characteristics of the imaging contour for registration, selects different registration methods according to different body position information, and ensures the accuracy of registration while reducing the calculation amount of image registration. quasi-quality. In addition, fault-tolerant processing can be performed on image abnormalities to better adapt to the specificity of different patient images and ensure algorithm efficiency.

The breast image registration method in the above embodiments of the present invention can be implemented in a computer-readable medium such as computer software, hardware or a combination of computer software and hardware. For hardware implementation, the embodiments described in this disclosure may be implemented on one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DAPDs), Programmable Logic Devices (PLDs) , field programmable gate array (FPGA), processor, controller, microcontroller, microprocessor, other electronic means for performing the functions described above, or a selected combination of the above means. In some cases, such embodiments may be implemented by a controller.

For a software implementation, the embodiments described in this disclosure may be implemented through separate software modules, such as procedures and functions, each of which implements one or more of the methods described herein. function and operation. Software codes can be implemented by application software written in a suitable programming language, stored in memory, and executed by a controller or processor. For example, a breast image registration device proposed according to an embodiment of the present invention includes: an image input unit, configured to input a symmetrical view image to be registered; an image processing unit, configured to process the image to be registered to obtain the breast contour and body position information of the image to be registered; the image registration unit is configured to determine registration points on the breast contour of the symmetrical view according to the body position information, and register the images according to the registration points.

Although the present invention has been disclosed as above with preferred embodiments, it is not intended to limit the present invention, and any person skilled in the art can use the methods disclosed above and technical content to analyze the present invention without departing from the spirit and scope of the present invention. Possible changes and modifications are made in the technical solution. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention, which do not depart from the content of the technical solution of the present invention, all belong to the technical solution of the present invention. protected range.

Claims (10)

1. A mammary gland image registration method, is characterized in that, comprises: Input the image to be registered of the symmetrical view; Obtain breast contour and body position information of the image to be registered; According to the body position information, the registration points are respectively determined on the breast contour of the symmetrical view; The images are registered according to the registration points. 2. The breast image registration method according to claim 1, wherein obtaining the breast contour comprises: Segment the breast tissue image in the image to be registered; The mammary gland contours were obtained using the chaincode tracking method. 3. The breast image registration method according to claim 2, wherein the segmentation is threshold segmentation. 4. breast image registration method according to claim 2, is characterized in that, also comprises: Removes invalid boundaries in breast contours. 5. The breast image registration method according to claim 1, wherein said determining the registration points respectively on the breast contours of the symmetric view according to the body position information comprises: selecting the farthest distance from the symmetry axis among the breast contours point as the registration point or select the center of the fitting circle with the minimum radius of the breast contour as the registration point. 6. The breast image registration method according to claim 1, wherein said determining the registration points on the breast contour of the symmetrical view according to the body position information comprises: Executing the first judgment, judging whether the body position information of the image to be registered is the first body position: If so, then use the first method to determine the registration point; If not, the registration point is determined using a second method. 7. breast image registration method according to claim 6, is characterized in that, The first body position is head and tail position; The first method includes: selecting a point in the breast contour farthest from the symmetry axis as a registration point. 8. The breast image registration method according to claim 7, wherein the second method comprises: Executing a second judgment, judging whether the radius of the fitting circle with the minimum radius of the breast contour is smaller than the diameter of the nipple tissue boundary circle; If so, select the center of the fitting circle with the smallest radius as the registration point; If not, reduce the step size of the fitting circle of the breast contour, re-determine the minimum radius fitting circle and execute the second judgment, if the step size is reduced to less than the preset threshold, the minimum radius fitting If the radius of the circle is still not smaller than the diameter of the nipple tissue boundary circle, the point in the breast contour farthest from the symmetry axis is selected as the registration point. 9. The breast image registration method according to claim 1, wherein said registering the images according to the registration points comprises: Taking one registration point as a reference registration point, moving the image including the remaining registration points, the moving distance is the difference between the remaining registration points and the longitudinal coordinates of the reference registration point. 10. A breast image registration device, comprising: An image input unit, configured to input images to be registered for symmetrical views; An image processing unit, configured to process the image to be registered to obtain breast contour and body position information of the image to be registered; The image registration unit is configured to respectively determine registration points on the breast contour of the symmetrical view according to the body position information, and register the images according to the registration points.