CN108831532B

CN108831532B - Method and system for processing nuclear medicine thyroid gland imaging image

Info

Publication number: CN108831532B
Application number: CN201810618274.5A
Authority: CN
Inventors: 张春丽; 成彧; 王荣福
Original assignee: Peking University First Hospital
Current assignee: Peking University First Hospital
Priority date: 2018-06-15
Filing date: 2018-06-15
Publication date: 2022-03-29
Anticipated expiration: 2038-06-15
Also published as: CN108831532A

Abstract

The application discloses a method for processing a nuclear medicine thyroid imaging image and a corresponding system, the method first converts an original image acquired by a nuclear medicine imaging device into a jpeg format, then, an HTML component is used for carrying out digital processing on the image in the jpeg format, preprocessing such as edge and character removing is carried out on the image after the digital processing, the thyroid gland boundary and the double-leaf thyroid gland parting line are extracted according to the gray value of each pixel point in the image after the preprocessing, therefore, accurate double-leaf thyroid gland morphology and single-leaf thyroid gland morphology are obtained, accurate double-leaf morphology parameters and double-leaf brightness parameters and single-leaf morphology parameters and single-leaf brightness parameters are obtained, the problems that data are inaccurate and the like caused by that a doctor manually delimits a thyroid gland region through work experience are solved, and meanwhile, the realization of artificial intelligent interpretation of a nuclear medicine thyroid gland image in the future is facilitated.

Description

Method and system for processing nuclear medicine thyroid gland imaging image

Technical Field

The application belongs to the field of nuclear medicine, and particularly relates to a method and a system for processing a thyroid imaging image in nuclear medicine.

Background

Nuclear medicine imaging is performed by injecting a radionuclide-labeled imaging agent (radiopharmaceutical) into a human body, distributing the radiopharmaceutical in a tissue or an organ to be examined, and displaying the form and function of the tissue or the organ to be examined through in vitro imaging. The thyroid gland image is prepared from^99mTc]The sodium pertechnetate solution is introduced into the patient by oral administration or intravenous injection^99mTc]The sodium pertechnetate is concentrated in the thyroid gland, so that the thyroid gland is imaged, and whether the form and the function of the thyroid gland are normal or not is judged according to the form, the size and the radioactivity distribution uniformity displayed by a thyroid gland nuclear medicine imaging image.

With the rapid development of artificial intelligence, the application research of artificial intelligence in medical image analysis is rapidly developed in recent years. At present, the artificial intelligent image analysis research on thyroid is mainly focused on the analysis of B-ultrasonic images, and the analysis research on thyroid images in nuclear medicine is relatively less. The detection method of the thyroid nuclide image segmentation of nuclear medicine by combining super-resolution reconstruction and KFCM (Kernel-based FCM, clustering algorithm) on the thyroid nuclide image is not reported in documents.

Thyroid gland development is a relatively extensive examination carried out by nuclear medicine, whether a thyroid gland development image is normal or not is mainly observed by a clinician by naked eyes at present, so that a great amount of time is spent by the clinician, the judgment result is subjective, the judgment result of different physicians can be different for the same thyroid gland development image, and the correct judgment depends on the experience of the clinician. With the rapid development of the artificial intelligence technology, the application of the artificial intelligence technology in medicine has also been developed in recent years, such as intelligent diagnosis of benign and malignant nodules in the lung of CT images and intelligent analysis of pathological results, and the application of the artificial intelligence technology in medical image analysis is an important development direction of future medicine. In the aspect of nuclear medicine thyroid imaging, an artificial intelligence technology is utilized to process thyroid images to obtain relevant parameters in the thyroid images for reference of doctors in image analysis, so that not only can a large amount of labor of doctors be saved, but also the diagnosis results of the doctors can be more objective, and meanwhile, the artificial intelligence interpretation of the nuclear medicine thyroid images in future can be conveniently realized.

Disclosure of Invention

The application provides a method for processing a thyroid imaging image in nuclear medicine, which aims to solve the problems that parameters extracted by a thyroid nuclear medicine image processing method in the prior art are inaccurate and the like.

The present application aims to provide the following aspects:

in a first aspect, the present application provides a method for processing a nuclear medicine thyroid imaging image, the method comprising: obtaining a nuclear medicine thyroid imaging image; preprocessing a thyroid imaging image in nuclear medicine; extracting a double-leaf thyroid nuclide imaging image and a single-leaf thyroid nuclide imaging image from the preprocessed nuclear medicine thyroid nuclide imaging image; and acquiring a double-leaf morphological parameter and a double-leaf brightness parameter of the double-leaf thyroid nuclide development image and a single-leaf morphological parameter and a single-leaf brightness uniformity parameter of each thyroid nuclide development image.

The applicant finds that a nuclear medicine thyroid imaging image is preprocessed to obtain a digitized image, a double-leaf thyroid nuclide imaging image and a single-leaf thyroid nuclide imaging image can be accurately extracted from the nuclear medicine thyroid imaging image according to the gray value of each pixel point in the digitized image, and more accurate thyroid contour lines and double-leaf thyroid parting lines can be obtained, so that accurate double-leaf morphological parameters and double-leaf brightness parameters, namely single-leaf morphological parameters and single-leaf brightness parameters, are provided, and the problems that a doctor manually defines a thyroid region by virtue of work experience to cause inaccurate data and the like are solved.

In one implementable manner, the pre-processing comprises: converting an original image acquired by the nuclear medicine imaging equipment in the DICOM format into a jpeg format, for example, converting the original image acquired by the nuclear medicine imaging equipment in the DICOM format into the jpeg format by using Java language, and then performing digital processing on the image in the jpeg format by using an HTML component; removing frame images around the digital nuclear medicine thyroid imaging image to obtain a borderless image; and removing the character image in the borderless image.

Optionally, the step of performing digital processing on the image in the jpeg format by using the HTML component includes converting an original image acquired by the nuclear medicine imaging device into a digital image formed by a pixel point array; and adjusting the width and height of the page window to the width and height of the digital image, wherein the HTML component comprises canvas, and the used computer languages are Java language and JavaScript language. The inventor finds that the HTML component comprises canvas, the used computer language is Java language and JavaScript language, and specific functions can be packaged into specific methods in the plug-in, if other people need to develop other functions of nuclear medicine image processing on the basis of the system, only the plug-in needs to be expanded, only the grammar of JavaScript and jQuery and some related knowledge of image processing need to be mastered, and how to realize functions of uploading and importing images and the like in a background is not concerned, so that a great deal of development time can be saved; secondly, the plug-in can be directly used through a small amount of jQuery codes, and the function of opening the box and using the plug-in is achieved; an HTML component and Java language and JavaScript language can be compatible with two operational characters of $ and jQuery, so that the programming flexibility can be improved; fourthly, global dependence can be avoided, specifically, if a global variable is used in the function, statements in the function body can access by bypassing the function parameter and the return value, and the independence of the function is damaged, so that the function depends on the global variable; and fifthly, the damage of a third party can be avoided, so that the safety and the stability of the image processing system are ensured. The embodiment of the application can also realize the intelligent judgment of the normal image of the thyroid imaging in nuclear medicine by adopting the C + + language, but the Java language has the advantages that the Java can better support Web server-side application, the Java language is more popular than the C + + language, the programming is simpler, the Java comprises standard libraries for completing specific tasks, the libraries are simple and easy to use, the C + + language depends on nonstandard libraries provided by other manufacturers, and the Java language does not have the concept of pointers compared with the C + + language, so that the system problem possibly caused by operating pointers in the C + + language is effectively prevented, and the program is safer.

Optionally, removing border images around the digitized nuclear medicine thyroid imaging image to obtain a borderless image includes: determining the boundaries of the four borders and the imaging part in the nuclear medicine thyroid imaging image respectively; and deleting four images outside the boundary to obtain a borderless image, wherein the outside is the side close to the outer edge of the nuclear medicine thyroid imaging image. The inventor finds that the non-image information around the nuclear medicine thyroid imaging image, such as a scroll bar and patient information, can greatly interfere with the extraction of the thyroid nuclide imaging image, and the removal of the non-image information before extracting the information of interest from the nuclear medicine thyroid imaging image can effectively improve the image processing speed.

Optionally, the removing the text image in the borderless image includes: determining a character area in the borderless image, wherein the character area is an area outside a closed rectangular area formed on the borderless image by two groups of transverse character pixel rows and two groups of longitudinal character pixel columns, and the outside is one side close to the edge of the borderless image; searching a text image in the text area, wherein the text image is a pixel point with a gray value larger than a preset text gray value; and modifying the gray value of the character image into a preset character correction gray value.

Optionally, the distance between the horizontal text pixel row and the borderless image edge may be a preset value; the distance between the longitudinal text pixel row and the borderless image edge can also be a preset value.

In one implementation, the extracting the bilobalt thyroid nuclide imaging image from the preprocessed nuclear medicine thyroid imaging image comprises: extracting a rectangular region comprising a bilobalt thyroid nuclide imaging image from the preprocessed nuclear medicine thyroid imaging image; and extracting a bileaflet thyroid nuclide imaging image from the rectangular region.

Wherein extracting a rectangular region including a bilobalt thyroid species imaging image from the preprocessed nuclear medicine thyroid imaging image comprises: carrying out filtering processing on the preprocessed nuclear medicine thyroid imaging image; searching filtering pixel points in the filtered image, wherein the filtering pixel points are pixel points with the maximum gray value in the filtered image; acquiring the position coordinates of the filtering pixel points in the filtered image; and searching thyroid boundary lines and thyroid boundary columns line by line or line by line in four directions from the filtering pixel points to the upper direction, the lower direction, the left direction and the right direction.

Optionally, sequentially searching the thyroid gland boundary line and the thyroid gland boundary column line by line in four directions, namely up, down, left, and right, by the filtering pixel point comprises: counting the gray values of all pixel points in the current row or the current column; searching a conversion pixel point with a gray value smaller than a conversion gray preset value in the current column or the current row; carrying out linear transformation on the gray value of the transformation pixel point to obtain a transformation pixel point; sorting the gray values of the variable-value pixel points and the gray values of the pixel points which are not converted in the current row or the current column to obtain a sorting array; selecting standard pixel points in the sequencing array, wherein the standard pixel points are located at preset positions in the sequencing array; calculating the difference value between the gray value of each pixel point in the sequencing array and the gray value of the standard pixel point in sequence from the first pixel point in the sequencing array; if the difference value is smaller than a stop calculation preset value, stopping calculating the difference value of the gray values of the pixel point behind the pixel point and the standard pixel point; obtaining the position of a stopping pixel point, wherein the stopping pixel point is the first pixel point in the sequencing array, and the difference value of the first pixel point is smaller than a stopping preset value; acquiring the length of a pixel point group, wherein the length of the pixel point group is the number of a current row or a current column stop calculating pixel point and all previous pixel points; calculating the gray level approximation degree, wherein the gray level approximation degree is the ratio of the length of a pixel group to the length of a current row or a current column, and the length of the current row or the current column is the total number of pixels in the current row or the current column; determining a thyroid gland boundary row and a thyroid gland boundary column, wherein the thyroid gland boundary row is a candidate pixel row closest to a filtering pixel point, the thyroid gland boundary column is a candidate pixel column closest to the filtering pixel point, the gray level approximation degree of the candidate pixel row is greater than or equal to a preset segmentation value, and the gray level approximation degree of the candidate pixel column is greater than or equal to a preset segmentation value.

According to the thyroid gland boundary determining method and device, the thyroid gland boundary is determined in the preset range according to the gray value of each pixel point, the integrity of the boundary can be guaranteed, and the accuracy of the boundary can be guaranteed.

In one implementation, extracting a bilobalt thyroid nuclide imaging image from the rectangular region includes: determining a neck horizontal centerline in the rectangular region; dividing a gland region in the rectangular region, wherein the gland region is a region below a horizontal central line of a neck in the rectangular region; if the filtering pixel point is not in the gland region, updating the filtering pixel point, wherein the updated filtering pixel point is the pixel point which obtains the maximum gray value after filtering again in the gland region; if the filtering pixel point is in the gland region, but the distance between the filtering pixel point and the gland region is smaller than a preset distance value, updating the gland region, and updating the filtering pixel point in the updated gland region, wherein the updated filtering pixel point is the pixel point with the maximum gray value in the updated gland region as the filtering pixel point after filtering processing is carried out again on the updated gland region; marking out the contour line of the double-leaf thyroid nuclide imaging image in the updated gland region by adopting a threshold segmentation method; and extracting a double-leaf thyroid nuclide development image from the gland region, wherein the double-leaf thyroid nuclide development image is the contour line of the double-leaf thyroid nuclide development image and an internal image thereof.

Optionally, if the filtering pixel point is in the gland region, but the distance between the filtering pixel point and the gland region is less than a preset distance value, updating the gland region specifically includes: searching for a judgment pixel point in each pixel point row from the top row to the U-th row of the original gland area, wherein the judgment pixel is the pixel point with the maximum gray value in the current pixel point row, calculating the average value of the gray values of all the judgment pixel points, calculating the ratio of the average value to the gray value of the filtering pixel point, and deleting the top row to the U-th row of the gland area if the ratio is greater than a deletion preset value to form an updated gland area, wherein U is a preset value.

Optionally, the determining the horizontal centerline of the neck in the rectangular region comprises: calculating the nuclide width of each row of pixel point rows in the rectangular region line by line, wherein the nuclide width is the width of a neck in the rectangular region; determining a neck horizontal center line, wherein the neck center line is a pixel point row with the minimum nuclide width and is larger than a preset width value.

Optionally, the delineating the contour line of the bilobed thyroid nuclide imaging image from the glandular region using a thresholding segmentation method comprises: calculating a segmentation threshold; determining an initial pixel point row, wherein the initial pixel point row is a pixel point row where a filtering pixel point is located; searching two boundary pixel points in the initial pixel point row, wherein the boundary pixel points are adjacent to a plurality of continuous background pixel points in the pixel point row, and the gray value of the background pixel points is smaller than a segmentation threshold; sequentially searching boundary pixel points in each row of pixel point rows in the gland region; and extracting the contour line of the double-leaf thyroid nuclide imaging image from the gland region, wherein the contour line of the double-leaf thyroid nuclide imaging image is the connecting line of all the boundary pixel points.

Optionally, the calculating the segmentation threshold comprises: sorting the gray values of all pixel points in the gland region from low to high to obtain a one-dimensional array; selecting a reference pixel point from a preset position in the one-dimensional array; and calculating a segmentation threshold value by using the gray value of the filtering pixel point and the gray value of the reference pixel point.

Further, the grey value of the filtering pixel point and the grey value of the reference pixel point are used for calculating a segmentation threshold according to the following formula I:

x-w 1 x1+ w2 x2 formula I

Wherein X represents a division threshold, W1 represents a weight of a reference pixel, X1 represents a gray scale value of the reference pixel, W2 represents a weight of a filter pixel, X2 represents a gray scale value of the filter pixel, and W1+ W2 is 1.

In one implementation, the sequentially finding boundary pixels in each column of pixel columns in the gland region comprises: acquiring a current gray value, wherein the current gray value is the gray value of pixel points in the current pixel point row in the same row as the boundary pixel points in the previous pixel point row; if the current gray value is larger than or equal to the segmentation threshold, selecting a boundary pixel point outside the pixel point; if the current gray value is smaller than the segmentation threshold, selecting a boundary pixel point at the inner side of the pixel point; and if the distance between the upper boundary pixel point of the current pixel point row and the outermost pixel point in the upper boundary pixel points in the upper pixel point row is more than or equal to one line, selecting a longitudinal boundary pixel point on the current pixel point row, wherein the longitudinal boundary pixel point is all pixel points between two boundary pixel points. The applicant finds that the integrity of the thyroid image contour can be ensured by acquiring the longitudinal boundary pixel points in the current pixel point row.

In one implementation, extracting a single-lobe thyroid species imaging image from the preprocessed nuclear medicine thyroid imaging image includes: if the double-leaf thyroid nuclide imaging image meets a preset drawing condition, drawing a thyroid double-leaf boundary in the double-leaf thyroid nuclide imaging image; and extracting a single-lobe thyroid nuclide development image from the double-lobe thyroid nuclide development image, wherein the single-lobe thyroid nuclide development image is a closed figure and an internal image thereof, which are formed by the boundary line of the thyroid double lobes and the contour line of the double-lobe thyroid nuclide development image on the left side or the right side of the thyroid double lobes in a surrounding manner.

And the preset drawing condition is that the upper edge difference height ratio is greater than or equal to a preset upper edge difference height ratio value.

Optionally, calculating the upper edge difference height ratio comprises: drawing the contour line of the double-leaf thyroid nuclide imaging image in a rectangular coordinate system; drawing an average dividing line of the double-leaf thyroid nuclide imaging image in the rectangular coordinate system; selecting a central pixel point row near the average dividing line, wherein the central pixel point row is a pixel point row with the minimum vertical coordinate of an upper boundary pixel point in the double-leaf thyroid nuclide imaging image; acquiring a longitudinal coordinate of a central pixel point, wherein the central pixel point is an upper boundary pixel point of a central pixel point row, and the longitudinal coordinate of the central pixel point is the longitudinal coordinate of the central pixel point in the rectangular coordinate system; two rows of upper edge pixel point rows are selected on two sides of the average dividing line, and the upper edge pixel point row is the pixel point row with the maximum vertical coordinate of the upper edge pixel point; respectively acquiring vertical coordinates of two upper edge pixel points, wherein the upper edge pixel points are upper boundary pixel points of an upper edge pixel point row, and the vertical coordinates of the upper edge pixel points are vertical coordinates of the upper boundary pixel points; respectively calculating upper edge difference values, wherein the upper edge difference values are the difference values of the vertical coordinates of the central pixel point and the vertical coordinates of the two upper edge pixel points; and calculating an upper edge difference height ratio, wherein the upper edge difference height ratio is the ratio of the upper edge difference value to the height of the double-leaf thyroid nuclide development image, and the height of the double-leaf thyroid nuclide development image is the difference value of the vertical coordinates of the pixel point of the highest upper edge and the pixel point of the lowest lower edge in the double-leaf thyroid nuclide development image.

Optionally, the drawing the thyroid bilobal demarcation line in the bilobal thyroid nuclide imaging image comprises: calculating the sum of gray values of all pixel points in the central pixel point row; respectively calculating the sum of gray values of all pixel points in each row of alternative pixel point rows, wherein the alternative pixel point rows are a plurality of rows near the left side and the right side of the central pixel point row; selecting a candidate pixel point row, wherein the candidate pixel point row is a pixel point row with the minimum sum of pixel gray values in the candidate pixel point row; calculating the ratio of the gray sums, wherein the ratio of the gray sums is the ratio of the sum of all the gray values of the pixels in the candidate pixel point row to the sum of all the gray values of the pixels in the central pixel point row; if the ratio of the gray sums is larger than or equal to the preset ratio of the gray sums, determining the position of the thyroid double-lobe boundary, wherein the position of the thyroid double-lobe boundary is a central pixel point row; and if the ratio of the gray sums is smaller than the preset ratio of the gray sums, drawing a thyroid double-lobe boundary line, wherein the thyroid double-lobe boundary line is a candidate pixel point row.

In one implementation, the acquiring bileaf morphological parameters of a bileaf thyroid nuclide imaging image includes: calculating the total width of the double-leaf thyroid nuclide development image and the height of the double-leaf thyroid nuclide development image, wherein the total width of the double-leaf thyroid nuclide development image is the total number of pixel point rows in the double-leaf thyroid nuclide development image, the height of the double-leaf thyroid nuclide development image is the difference between the ordinate of the upper edge pixel and the ordinate of the lower edge pixel, and the lower edge pixel is the pixel with the minimum lower boundary ordinate.

In one implementation, acquiring bileaf intensity parameters for a bileaf thyroid species imaging image includes: acquiring the maximum gray value of a double-leaf thyroid nuclide development image, wherein the maximum gray value of the double-leaf thyroid nuclide development image is the gray value of a pixel point with the maximum gray value in the double-leaf thyroid nuclide development image; respectively obtaining the maximum gray value of each single-leaf thyroid gland development image and the average gray value of the single-leaf thyroid gland development image, wherein the maximum gray value of the single-leaf thyroid gland development image is the gray value of a pixel point with the maximum gray value in the single-leaf thyroid gland development image, and the average gray value of the single-leaf thyroid gland development image is the average value of the gray values of all the pixel points in the single-leaf thyroid gland development image; and calculating the ratio of the maximum gray values of the two single-leaf thyroid imaging images in the double-leaf thyroid nuclide imaging image and the ratio of the average gray values of the two single-leaf thyroid imaging images.

The single-leaf morphological parameters for acquiring the imaging image of each leaf of the thyroid nuclide comprise: acquiring the width, height and area of each single-leaf thyroid gland development image, wherein the width of each single-leaf thyroid gland development image is the total number of pixel point rows in each single-leaf thyroid gland development image, the height of each single-leaf thyroid gland development image is the difference between the ordinate of an upper edge pixel point and the ordinate of a lower edge pixel point of a thyroid nuclide development image, and the area of each single-leaf thyroid gland development image is calculated by the total number of pixel points in the single-leaf thyroid gland development image; calculating the aspect ratio of the single-leaf thyroid imaging image; calculating the ratio of the area of the thyroid gland of each lobe to the area of the neck visualization image, the area of the neck visualization image being calculated as the square of the width of the neck visualization image; the ratio of the width, the ratio of the height, the ratio of the area of the two thyroid imaging images and the ratio of the area of each thyroid imaging image to the area of the neck imaging image are respectively calculated.

Acquiring the single-leaf morphological parameters of each thyroid imaging image further comprises: uniformly dividing the single-leaf thyroid imaging image into m rows of image strips, wherein m is an odd number, and each image strip comprises a plurality of pixel points; calculating the area of a kernel image in each image strip; and calculating the nuclide image area ratio of each image strip in the single-leaf thyroid imaging image, wherein the nuclide image area ratio is the ratio of the area of the pixel image in the current image strip to the reference area of the image strip, the reference area of the image strip is the area of the nuclide image in the reference image strip, and the reference image strip is the image strip positioned in the center of the single-leaf thyroid imaging image.

The brightness parameters for obtaining the thyroid imaging image of each leaf include: uniformly dividing the single-leaf thyroid imaging image into m 'rows of image strips, wherein m' is an odd number, and each image strip comprises a plurality of pixel points; calculating the average gray value of each image strip; calculating the gray ratio of all image strips in the single-leaf thyroid gland development image, wherein the gray ratio of the image strips is the ratio of the average gray value of the current image strip to the reference gray value of the image strips, the reference gray value of the image strips is the average gray value of the reference image strips, and the reference image strips are the image strips located in the center of the single-leaf thyroid development image; extending the reference image strip towards the outer side in any direction, and uniformly dividing the single-leaf thyroid imaging image into n 'rows, wherein n' is an odd number to form an m '× n' block image block; calculating the average gray value of each image block; and calculating the gray ratio of all image blocks in the single-leaf thyroid gland imaging image, wherein the gray ratio of the image blocks is the ratio of the average gray value of the current image block to the reference gray value of the image blocks, the reference gray value of the image blocks is the average gray value of the reference image blocks, and the reference image blocks are the image blocks positioned in the centers of all the image strips.

Compared with the prior art, the technical scheme provided by the application includes that an original image acquired by nuclear medicine imaging equipment is converted into a jpeg format, then an HTML (hypertext markup language) component is used for carrying out digital processing on the jpeg format image, preprocessing such as edge removal and character removal is carried out on the digital processed image, and thyroid boundary and a thyroid parting line of a double leaf are extracted according to the gray value of each pixel point in the preprocessed image, so that a thyroid gland body form and a thyroid gland body form with accurate forms are obtained, accurate double leaf form parameters and double leaf brightness parameters and single leaf form parameters and single leaf brightness parameters are obtained, and the problems that data are inaccurate and the like due to the fact that a doctor manually defines a thyroid region through work experience are solved.

The present application further provides a nuclear medicine thyroid imaging image detection system, the system comprising: the image acquisition unit is used for acquiring a nuclear medicine thyroid imaging image; the image preprocessing unit is used for preprocessing the nuclear medicine thyroid imaging image;

the nuclide image extraction unit is used for extracting a double-leaf thyroid nuclide development image and a single-leaf thyroid nuclide development image from the preprocessed nuclear medicine thyroid development image; and the parameter acquisition unit is used for acquiring the double-leaf morphological parameters and the double-leaf brightness parameters of the double-leaf thyroid nuclide imaging image and the single-leaf morphological parameters and the single-leaf brightness uniformity parameters of each thyroid imaging image.

Drawings

Fig. 1 is a schematic flowchart of a method for processing a nuclear medicine thyroid gland imaging image by a data processing device according to an embodiment of the present application;

FIG. 2 shows a digitized nuclear medicine thyroid imaging image obtained from the pre-processing;

fig. 3 shows the position of the normal thyroid in the neck.

Detailed Description

The following describes embodiments of the present application with reference to the drawings in the embodiments of the present application.

In order to better understand the method and system for processing a nuclear medicine thyroid gland imaging image disclosed in the embodiments of the present application, a description will be first given of a hardware scenario to which the embodiments of the present application are applicable. The present application is applied to a nuclear medicine working system, please refer to fig. 1, as shown in fig. 1, the nuclear medicine working system includes a nuclear medicine imaging system 100 and an image information system (workstation) 200, data can be communicated between the nuclear medicine imaging device 100 and the image information system 200, or at least, the nuclear medicine imaging device 100 can send a collected image to the image information processing system, and the image information system 200 can also receive and process an image sent by the nuclear medicine imaging device 100. The nuclear medicine imaging device 100 is used for acquiring nuclear medicine images. Currently, nuclear medicine imaging devices mainly include a Single-Photon Emission Computed Tomography (SPECT) device and a Positron Emission Tomography (PET) device, and the image information system 200 is configured to process images acquired by the nuclear medicine imaging device 100. Generally, the images acquired by the nuclear medicine imaging apparatus 100 are original digital images, and the width and height of each acquired image may be slightly different. The visual information system 200 may be a computer or other device for data processing. The image processing in the video information system 200 can be visualized, and the size of the window of the visualized image on the system 200 may be the same as or different from the size of the display screen. The method and system provided by the embodiment of the application are applied to the image information system 200.

The nuclear medicine image is mainly formed by introducing a radioactive labeled medicine into a specific area (such as thyroid) in a human body and combining an image of the specific area acquired by a tracing technology and a radioactive detection and scanning imaging technology.

Fig. 1 is a schematic flowchart of a method for processing a nuclear medicine thyroid gland imaging image by a data processing device according to an embodiment of the present application, where as shown in fig. 1, the method includes:

and S100, acquiring a nuclear medicine thyroid imaging image.

The nuclear medicine thyroid gland imaging image is an image which is acquired by a nuclear medicine means and is related to the thyroid gland, and the format of the image is an original digital image format. The width and height of each acquired thyroid nuclear medicine image may be slightly different due to equipment and the like.

And S200, preprocessing the nuclear medicine thyroid imaging image.

The data processing device can preprocess the acquired nuclear medicine thyroid image to obtain a digital image without interference information such as frames, characters and the like, and the specific steps can be as follows:

(1) the raw images acquired by the nuclear medicine imaging device in the DICOM format are converted into the jpeg format, and optionally, the raw images acquired by the nuclear medicine imaging device in the DICOM format can be converted into the jpeg format by using the Java language.

(2) Digitizing jpeg formatted images using HTML components

Converting an original image acquired by nuclear medicine imaging equipment into a digital image formed by a pixel point array by utilizing an HTML (hypertext markup language) component, wherein the HTML component comprises canvas, and the used computer languages are Java language and JavaScript language; and adjusting the width and height of the page window to the width and height of the digital image to ensure that the width and height of the canvas assembly are the same as those of the digital image obtained by digital processing so as to facilitate subsequent image processing, and then calling a drawImage () method to draw a jpeg format image in the canvas assembly. After the image is drawn, the image in the canvas component is digitized, a getImageData () method is called to read the pixel data in the canvas component, and a result imgData is returned.

For each pixel in the digitized nuclear medicine thyroid imaging image, there are 4 aspects of information: the imgData [ i ] is red, imgData [ i +1]: green, imgData [ i +2]: blue, imgData [ i +3]: opacity, so traversing imgData requires that the increment of subscript i in each cycle be set to 4, where i represents the position of the current digitized pixel point in the imgData array. Since the image to be processed is a gray image, and the gray value of each pixel in the digitized nuclear medicine thyroid imaging image generally satisfies imgData [ i ] ═ imgData [ i +1] ═ imgData [ i +2] and alpha ═ 255, the gray value of each pixel is imgData [ i ] (between 0 and 255).

For example, the data processing device acquires an original image with width × height of 1024 × 1024, and processes the original image into a digitized nuclear medicine thyroid gland visualization image with 1024 × 1024 rows and 1024 columns of pixels by using the HTML component, where the grayscale value of the ith pixel is imgData [4 × i ].

(3) Removing the frame image around the digital nuclear medicine thyroid imaging image to obtain a borderless image

In the embodiment of the application, the digital nuclear medicine thyroid gland imaging image can be preliminarily cut, so that all or a part of the periphery of the digital nuclear medicine thyroid gland imaging image which is useless for obtaining related parameters is cut, the useless images are characterized in that in the same pixel point row or the same pixel point column, the number of pixel point pairs with the gray value of the next pixel point close to or equal to that of the previous pixel point is more, by taking a pixel point behavior example, in the same pixel point row, the gray value of the 2 nd pixel point is found to be close to the gray value of the 1 st pixel point from left to right, the gray value of the 3 rd pixel point is also close to that of the 2 nd pixel point, and so on, many pixel point pairs in the pixel point row indicate that the pixel point row does not have the characteristics of the thyroid gland image and has the characteristics of a frame, as can be gathered in particular from fig. 2. Fig. 2 shows a digital nuclear medicine thyroid gland imaging image obtained by preprocessing, and as shown in fig. 2, almost all pixel points on a frame of the digital nuclear medicine thyroid gland imaging image obtained by preprocessing are gray-white, that is, the gray values of adjacent pixel points are almost unchanged, while the gray values of adjacent pixel points in the same pixel point row or the same pixel point column are changed greatly in an inverted thyroid gland image part.

The step of removing the frame image around the digital nuclear medicine thyroid imaging image to obtain a borderless image may specifically be:

1) determining the boundaries of the four borders and the imaging part in the digital nuclear medicine thyroid imaging image respectively, which specifically comprises the following steps:

determining a border pixel group corresponding to each edge of the digital nuclear medicine thyroid gland imaging image;

respectively counting the gray values of all pixel points in each group of frame pixel groups;

respectively counting the number of pixels of which the gray value difference between two adjacent pixels in each group of frame pixel groups is smaller than a first preset gray value;

calculating the proportion of pixels with the gray value difference of adjacent pixels in the two adjacent frame pixel groups smaller than a first gray preset value in the pixel group;

if the proportion of the pixel points with the gray value difference of two adjacent pixel points of each group of frame pixel groups smaller than the first preset gray value in the group of frame pixel groups is larger than or equal to the preset proportion value, cutting the frame pixel groups and all the pixel points outside the frame pixel groups, and stopping at the row or the column;

and if the ratio of the pixel points with the gray value difference of the adjacent pixel points smaller than the first preset gray value in the group of frame pixel groups is smaller than the preset ratio, moving one row or one column to the outer side of the image to calculate the next group, and cutting off all the pixel points outside the boundary to obtain the borderless image when the movement is stopped in the up, down, left and right directions.

2) And deleting four images outside the boundary to obtain a borderless image, wherein the outside is the side close to the outer edge of the nuclear medicine thyroid imaging image.

For example, 2 rows of pixel points and 2 columns of pixel points are determined on the digitized nuclear medicine thyroid gland imaging image, wherein the 1 st row of pixel points is separated from the top row of pixel points of the digitized nuclear medicine thyroid gland imaging image by a row of pixel points, the 2 nd row of pixel points is separated from the bottom row of pixel points of the digitized nuclear medicine thyroid gland imaging image by b rows of pixel points, the 1 st column of pixel points is separated from the leftmost column of pixel points of the digitized nuclear medicine thyroid gland imaging image by c columns of pixel points, and the 2 nd column of pixel points is separated from the rightmost column of pixel points of the digitized nuclear medicine thyroid gland imaging image by d columns of pixel points, wherein a, b, c and d are preset values, and may be the same or different, and a, b, c and d may be set empirically.

In another embodiment, deleting four images outside the boundary, and obtaining the borderless image may further include:

determining 2 rows of pixel points and 2 columns of pixel points on the digital nuclear medicine thyroid gland imaging image, wherein the number of pixel point rows in the 1 st row, which are separated from the top row of pixel point rows of the digital nuclear medicine thyroid gland imaging image, accounts for a '% of the total number of pixel points in the digital nuclear medicine thyroid gland imaging image, the number of pixel point rows in the 2 nd row, which are separated from the bottom row of pixel point rows of the digital nuclear medicine thyroid gland imaging image, accounts for b '% of the total number of pixel points in the digital nuclear medicine thyroid gland imaging image, the number of pixel point rows in the 1 st row, which are separated from the leftmost column of pixel point rows of the digital nuclear medicine thyroid gland imaging image, accounts for c '% of the total number of pixel points in the digital nuclear medicine thyroid gland imaging image, and the number of pixel point rows in the 2 nd column, which are separated from the rightmost column of pixel point rows of the digital nuclear medicine thyroid gland imaging image, accounts for the digital nuclear medicine thyroid imaging image D '% of the total column number of the pixel points in the medical thyroid gland imaging image, wherein a ', b ', c ' and d ' are preset values which can be the same or different and can be set according to experience;

sequentially calculating the difference value between the gray value of the next pixel point in the current line or the current column and the gray value of the previous pixel point from the leftmost pixel point of the pixel point lines in the 2 lines and the uppermost pixel point in the pixel point columns in the 2 columns, and if the difference value is less than the preset clipping value and exceeds a preset proportion, marking the current line or the current column as a candidate clipping line or a candidate clipping column;

judging whether the current line/column is a candidate cutting line/column or not according to the steps line by line/column by column on the edge of the digital nuclear medicine thyroid gland imaging image by taking a pixel point line or a pixel point column as a step length;

and if the continuous e 'rows/columns are candidate clipping rows/columns, deleting the first candidate clipping row/column and the pixel point rows/columns outside the first candidate clipping row/column, wherein the outside is the side close to the edge of the digital nuclear medicine thyroid imaging image, and e' is a preset value.

Setting the digital nuclear medicine thyroid gland imaging image as an image with 800 pixel point rows × 1000 pixel point columns, presetting that the difference value of more than 5% in each row/column is close to be a candidate clipping row/column, presetting that the candidate clipping row/column is searched from a position 10% away from the edge of the digital nuclear medicine thyroid gland imaging image, assuming that the difference value in the pixel point rows/columns of 5 rows/columns is larger than the preset value, namely, line clipping, and taking a clipping top horizontal edge as an example (firstly searching for pixel point rows) to explain the scheme of the embodiment:

firstly, determining the initial pixel point behavior of 800 multiplied by 10 percent to be 80, namely, starting from the 80 th row of pixel point rows from the top of the digital nuclear medicine thyroid gland imaging image, calculating the difference value of the gray values of two adjacent pixel points in each row of pixel point rows, if the number of the difference values smaller than the preset value in the 80 th row of pixel point rows is 10 and the number of the difference values smaller than the preset value in the 80 th row of pixel point rows is 50, calculating the difference value of the gray values of two adjacent pixel points in the 79 th row of pixel point rows from the top of the digital nuclear medicine thyroid gland imaging image, if the number of the difference values smaller than the preset value in the 79 th row of pixel point rows is still smaller than 50, continuing to calculate the difference value of the gray values of two adjacent pixel points in the 78 th row from the top of the digital nuclear medicine thyroid gland imaging image, and so on until starting from the 70 th row from the top of the digital nuclear medicine thyroid imaging image, and (3) continuously arranging 5 rows of pixel point rows, wherein the number of the difference values of the gray values of two adjacent pixel points in each row of pixel point rows is more than 50, deleting all the pixel point rows from the 1 st row to the 70 th row from the top of the digital nuclear medicine thyroid imaging image, and finishing the cutting of the top horizontal edge. The other edges are cut according to the same type.

(4) Removing character image in borderless image

After the edge removing processing is carried out on the image, some character information which is close to white and is useless for analysis in the borderless image can be removed, and the characteristic of the pixel point for expressing the character information is that the pixel point is close to white, namely the gray value is close to 255 and is close to 4 edges of the image after the preliminary cutting.

The removing of the text image in the borderless image may specifically include:

determining a character area in the borderless image, wherein the character area is an area outside a closed rectangular area formed on the borderless image by two groups of transverse character pixel rows and two groups of longitudinal character pixel columns, and the outside is one side close to the edge of the borderless image;

searching a text image in the text area, wherein the text image is a pixel point with a gray value larger than a preset text gray value;

and modifying the gray value of the character image into a preset character correction gray value.

For example, a certain pixel point exists in the edge-removed image, the ratio of the distance between the position of the pixel point and the upper boundary of the edge-removed image (the number of pixel points between the pixel point row where the pixel point is located and the upper boundary) to the height of the edge-removed image (the total number of pixel point rows of the edge-removed image) is less than or equal to a preset ratio, and when the difference between the gray value of the pixel point and 255 is less than or equal to a preset character gray value, the pixel point is considered to be used for expressing the character information, and the gray value of the pixel point is reset to 0.

In the above example, assuming that an image of a borderless image obtained after the edge deletion is 700 pixel point rows × 800 pixel point columns, the gray value of a certain pixel point in a pixel point at a distance of 1% (the preset proportion is 2%, 1% refers to the 7 th pixel point row from the upper boundary of the borderless image) from the upper boundary of the borderless image is 240, and the difference between 255 and 240 is 15 less than the preset character gray value (30), the gray value of the pixel point is reset to 0.

S300, extracting a double-leaf thyroid nuclide imaging image and a single-leaf thyroid nuclide imaging image from the preprocessed nuclear medicine thyroid nuclide imaging image.

Preprocessing a nuclear medicine thyroid imaging image to obtain an image formed by a pixel matrix with known gray value of each pixel, and extracting a thyroid nuclide imaging image from the image, wherein the method specifically comprises the following steps:

(1) extracting a rectangular region comprising a bilobalt thyroid species imaging image from the preprocessed nuclear medicine thyroid species imaging image.

Extracting a rectangular region including a bilobalt thyroid species imaging image from the preprocessed nuclear medicine thyroid species imaging image may specifically include:

filtering the preprocessed nuclear medicine thyroid imaging image, for example, selecting a template with a proper size to perform mean filtering on a borderless image, taking a pixel point (x, y) as an example, taking the average value of the gray levels of N pixel points of a neighborhood delta xy of the pixel point (x, y) as the gray level of the pixel point (x, y), filtering noise in the image from the surface, and keeping the detail characteristics of the image;

searching filtering pixel points in the filtered image, wherein the filtering pixel points are pixel points with the maximum gray value in the filtered image;

acquiring the position coordinates of the filtering pixel points in the filtered image; and searching thyroid boundary lines and thyroid boundary columns line by line or line by line in four directions of the filtering pixel points.

Optionally, searching the thyroid gland boundary row and the thyroid gland boundary column by the filtering pixel points in four directions, namely, up, down, left and right, line by line or column by column sequentially comprises: counting the gray values of all pixel points in the current row or the current column; searching a conversion pixel point with a gray value smaller than a conversion gray preset value in the current column or the current row; carrying out linear transformation on the gray value of the transformation pixel point to enable the transformation pixel point to be more sensitive to a darker area in an image and obtain a variable value pixel point; sorting the gray values of the variable-value pixel points and the gray values of the pixel points which are not converted in the current row or the current column to obtain a sorting array; selecting standard pixel points in the sequencing array, wherein the standard pixel points are located at preset positions in the sequencing array; calculating the difference value between the gray value of each pixel point in the sequencing array and the gray value of the standard pixel point in sequence from the first pixel point in the sequencing array; if the difference value is smaller than a stop calculation preset value, stopping calculating the difference value of the gray values of the pixel point behind the pixel point and the standard pixel point; obtaining the position of a stopping pixel point, wherein the stopping pixel point is the first pixel point in the sequencing array, and the difference value of the first pixel point is smaller than a stopping preset value; acquiring the length of a pixel point group, wherein the length of the pixel point group is the number of a current row or a current column stop calculating pixel point and all previous pixel points; calculating the gray level approximation degree, wherein the gray level approximation degree is the ratio of the length of a pixel group to the length of a current row or a current column, and the length of the current row or the current column is the total number of pixels in the current row or the current column; determining a thyroid gland boundary row and a thyroid gland boundary column, wherein the thyroid gland boundary row is a candidate pixel row closest to a filtering pixel point, the thyroid gland boundary column is a candidate pixel column closest to the filtering pixel point, the gray level approximation degree of the candidate pixel row is greater than or equal to a preset segmentation value, and the gray level approximation degree of the candidate pixel column is greater than or equal to a preset segmentation value.

For example, continuing with the above example, assume that the current image has no text information and is still 700 pixel dot rows × 800 pixel dot columns. The filtering pixels are located in the 300 th row (counted from the upper boundary) and the 400 th row (counted from the left boundary), the rows where the filtering pixels are located are searched, namely the pixels with the gray values smaller than the preset gray value of the transform in the 300 th row, the gray values of the pixels are subjected to linear transform to obtain variable-value pixels, then the gray values of the variable-value pixels and the gray values of the pixels which are not subjected to transform in the 300 th row are sequenced together, and a sequencing array is obtained; selecting standard pixel points in the sequencing array, wherein the standard pixel points are located at 50% of the leftmost pixel points of the distance sequencing array (the preset position is that the 800 × 50% of the first pixel point on the left side is 400 pixel points); calculating the difference value between the gray value of each pixel point in the sequencing array and the gray value of the standard pixel point in sequence from the first pixel point on the left side of the sequencing array; if the difference value is smaller than a stop calculation preset value, stopping calculating the difference value of the gray values of the pixel point behind the pixel point and the standard pixel point; acquiring the position of a stop pixel point (300, 300 means that the 300 th pixel point is counted from the first pixel point on the left side of a sequencing array), wherein the stop pixel point is the first pixel point in the sequencing array, and the difference value is smaller than a stop preset value; acquiring the length (300) of a pixel point group, wherein the length of the pixel point group is the number of a current line stop calculation pixel point and all previous pixel points; calculating a gray level approximation (300/800), wherein the gray level approximation is the ratio of the length of the pixel group to the length of the current line, and the length of the current line is the total number of pixels in the current line; assuming that the segmentation preset value is 500/800, and the gray approximation degree of the current row is 300/800 and is smaller than 500/800, the gray approximation degree of the pixel point row adjacent to the filtering pixel point is calculated until the gray approximation degree of the 100 th row is respectively larger than 500/800, the upper thyroid gland boundary row is determined as the 100 th row, and similarly, the lower boundary row and the left and right boundary rows can be determined.

(2) And extracting a bileaflet thyroid nuclide imaging image from the rectangular region.

Determining a neck horizontal centerline in the rectangular region; dividing a gland region in the rectangular region, wherein the gland region is a region below a horizontal central line of a neck in the rectangular region; if the filtering pixel point is not in the gland region, updating the filtering pixel point, wherein the updated filtering pixel point is the pixel point which obtains the maximum gray value after filtering again in the gland region; if the filtering pixel point is in the gland region but the distance between the filtering pixel point and the upper boundary of the gland region is smaller than a preset distance value, updating the gland region, and updating the filtering pixel point in the updated gland region, wherein the updated filtering pixel point is the pixel point with the largest gray value in the updated gland region after the updated gland region is subjected to filtering processing again; marking out the contour line of the double-leaf thyroid nuclide imaging image in the updated gland region by adopting a threshold segmentation method; and extracting a double-leaf thyroid nuclide development image from the gland region, wherein the double-leaf thyroid nuclide development image is the contour line of the double-leaf thyroid nuclide development image and an internal image thereof.

Since the area of the thyroid gland with the highest brightness should be below the horizontal line in the center of the neck to conform to the physiological configuration, see fig. 3, fig. 3 shows the position of the normal thyroid gland in the neck, where 1 is thyroid cartilage, 2 is thyroid, 3 is trachea, 4 is sternum, and 5 is clavicle. As can be seen from fig. 3, since the normal thyroid is located below the horizontal line of the center of the neck, the area with high brightness above the horizontal line of the center of the neck is not the thyroid, and therefore, these pixel points must be removed from the map to avoid interference.

Alternatively, determining the nuclide width may include: for example, starting from any row of pixel point rows in the rectangular region, sequentially searching whether a certain pixel point exists from left to right, wherein the gray value of the pixel is greater than the nuclide brightness preset value, and the gray values of f% pixel points in all the pixel points on the right side of the pixel point are also greater than the nuclide brightness preset value, wherein f is a preset value, and if so, determining that the pixel point is a left boundary point; and then, in the row, a right boundary point is searched from right to left by the same method, and the number of the left boundary point, the right boundary point and all pixel points between the two points is the nuclide width.

Further, the minimum value of the width of the nuclide and the pixel point row where the nuclide is located are determined, and if the width of the nuclide in the pixel point row is larger than or equal to a preset value of the width of the nuclide, the position where the pixel point row is located is considered to be the vertical position of the center of the neck.

Further, if the distance between the filtering pixel point and the horizontal central line of the neck is smaller than a preset distance value, after dividing the gland region in the rectangular region, before dividing the double-lobed thyroid nuclide imaging image in the gland region by using a threshold segmentation method, the method further comprises: acquiring gray values of a plurality of pixels at the top of the gray value in each row of pixel rows from the center line of the neck part downwards line by line; calculating the average gray value of a plurality of pixel points with the highest gray value in the row; calculating the ratio of the average gray value to the gray value of the filtering pixel point; and if the ratio is larger than the deletion preset value, deleting the row and updating the gland area.

For example, if the filtering pixel is closer to the central line of the neck, it indicates that the filtering pixel may not be in the thyroid, at this time, several pixels with the maximum gray value in each row of pixels need to be sequentially solved from the 1 st row below the central position of the neck, for example, the average value of the gray values of the 10 pixels with the maximum gray values is obtained by analyzing the normal thyroid image, the value (10) is an empirical value obtained by continuously calculating the ratio of the average gray value of the 10 pixels to the gray value of the filtering pixel, and if the ratio is greater than the deletion preset value, the 1 st row below the central position of the neck needs to be removed, the maximum value of the image gray and the position coordinates of the maximum value in the image are re-solved by changing the range. The inventors have found that updating the gland region by deleting rows having a ratio greater than the deletion preset value enables a more accurate determination of the location of the thyroid gland.

x-w 1 x1+ w2 x2 formula I

For example, continuing the above example, assuming that the image after the update of the gland region is a 450 pixel row × 600 pixel column, sorting the gray values of all the pixels in the gland region from low to high to obtain a one-dimensional array, determining the gray value of a reference pixel in the one-dimensional array (for example, a pixel in the one-dimensional array that is 60% away from the minimum gray value and 40% away from the maximum gray value, in this example, a 600 × 60% — 360 pixels), and performing weighted average by using the gray value of the reference pixel and the gray value of the filtering pixel, optionally, the weighted average may be calculated according to the following method:

let the gray value taken from the sorted array be x1 with weight of w1, the maximum value of the thyroid gray is x2 with weight of w2, 0< w1<1 and 0< w2<1 and w1+ w2 ═ 1, and the weighted average value x ═ w1 × 1+ w2 × 2

The weight determination principle may be: when the gray value in the sorted array is taken out to be higher, the weight of the gray value is slightly increased, and when the gray value in the sorted array is taken out to be lower, the weight of the gray value is slightly decreased. For example, the weight is 0.8 when the extracted gradation value is less than 20, 0.81 when it is greater than or equal to 20 and less than 30, 0.82 when it is greater than or equal to 30 and less than 40, and 0.83 when it is greater than or equal to 40.

Because the shape of the thyroid gland double-lobe is similar to an ellipse, there are only upper and lower boundaries of the thyroid gland in the same pixel point column of the thyroid gland image under normal conditions, and the method of the embodiment improves the threshold segmentation method, so that the segmentation threshold is related to both the position and the gray value of the pixel point in the image at the same time, and specifically, the threshold segmentation method may include:

continuing with the above example, after updating, the current location of the filtering pixel point is listed as column 200, first two boundary points above and below the thyroid gland in column 200 are found, and then boundary points above and below the thyroid gland in columns 199 and 201 are found. Taking the example of finding the upper and lower boundary points in the 199 th column as an example, if it is determined that the upper boundary point is the pixel point in the 250 th row in the 200 th column and the lower far point is the pixel point in the 350 th row, the gray value of the pixel point in the 250 th row in the 199 th column is first found, if the gray value of the pixel point is greater than the segmentation threshold, the gray value of the pixel point in the 249 th row is found, whether the gray value of the pixel point in the 249 th row is the upper boundary point is determined, similarly, when the lower boundary point in the 199 th column is found, the gray value of the pixel point in the 350 th row in the 199 th column is first found, whether the pixel point is the lower boundary point is determined, and if the gray value of the pixel point is less than the segmentation threshold, whether the 349 th row is the lower boundary point is found. Taking column 201 as an example, if the upper boundary point of column 201 is determined to be the pixel point of row 240 according to the above method, all the pixel points between row 240 and row 250 on column 201 are marked as upper boundary points.

(3) Extracting single-lobe thyroid nuclide imaging image from preprocessed nuclear medicine thyroid imaging image

1) If the double-leaf thyroid nuclide imaging image meets a preset drawing condition, drawing a thyroid double-leaf boundary in the double-leaf thyroid nuclide imaging image;

2) And extracting a single-lobe thyroid nuclide development image from the double-lobe thyroid nuclide development image, wherein the single-lobe thyroid nuclide development image is a closed figure and an internal image thereof, which are formed by the boundary line of the thyroid double lobes and the contour line of the double-lobe thyroid nuclide development image on the left side or the right side of the thyroid double lobes in a surrounding manner.

S400, acquiring double-leaf morphological parameters and double-leaf brightness parameters of double-leaf thyroid nuclide imaging images and single-leaf morphological parameters and single-leaf brightness uniformity parameters of each double-leaf thyroid nuclide imaging image.

(1) Double-leaf morphological parameters for obtaining double-leaf thyroid nuclide imaging image

Calculating the total width of the double-leaf thyroid nuclide development image and the height of the double-leaf thyroid nuclide development image, wherein the total width of the double-leaf thyroid nuclide development image is the total number of pixel point rows in the double-leaf thyroid nuclide development image, the height of the double-leaf thyroid nuclide development image is the difference between the ordinate of the upper edge pixel and the ordinate of the lower edge pixel, and the lower edge pixel is the pixel with the minimum lower boundary ordinate.

(2) Double-leaf brightness parameter for obtaining double-leaf thyroid nuclide imaging image

Acquiring the maximum gray value of a double-leaf thyroid nuclide development image, wherein the maximum gray value of the double-leaf thyroid nuclide development image is the gray value of a pixel point with the maximum gray value in the double-leaf thyroid nuclide development image; respectively obtaining the maximum gray value of each single-leaf thyroid gland development image and the average gray value of the single-leaf thyroid gland development image, wherein the maximum gray value of the single-leaf thyroid gland development image is the gray value of a pixel point with the maximum gray value in the single-leaf thyroid gland development image, and the average gray value of the single-leaf thyroid gland development image is the average value of the gray values of all the pixel points in the single-leaf thyroid gland development image; and calculating the ratio of the maximum gray values of the two single-leaf thyroid imaging images in the double-leaf thyroid nuclide imaging image and the ratio of the average gray values of the two single-leaf thyroid imaging images.

(3) Single-leaf morphological parameters for obtaining per-leaf thyroid nuclide imaging image

1) Obtaining width of single-leaf thyroid imaging image per leaf

The width of each leaf of the single-leaf thyroid gland imaging image is the total number of pixel point columns in each leaf of the single-leaf thyroid gland imaging image,

2) obtaining height of single-leaf thyroid imaging image per leaf

The height of each single-leaf thyroid gland development image is the difference between the ordinate of the upper edge pixel and the ordinate of the lower edge pixel, and the area of the single-leaf thyroid gland development image is calculated by the total number of the pixels in the single-leaf thyroid gland development image;

3) acquiring the area of each single-leaf thyroid imaging image,

calculating the aspect ratio of the single-leaf thyroid imaging image; calculating the ratio of the area of the thyroid gland with a single lobe to the area of the neck visualization image, wherein the area of the neck visualization image is calculated by the square of the width of the neck visualization image; the ratio of the width, the ratio of the height, the ratio of the area of the two thyroid imaging images and the ratio of the area of each thyroid imaging image to the area of the neck imaging image are respectively calculated.

4) Obtaining the gradient ratio of each thyroid imaging image

Uniformly dividing the single-leaf thyroid imaging image into m rows of image strips, wherein m is an odd number, and each image strip comprises a plurality of pixel points; calculating the area of a kernel image in each image strip; calculating a gradient ratio of the single-leaf thyroid gland development image, wherein the gradient ratio is a nuclide image area ratio of each image strip in the single-leaf thyroid gland development image, the nuclide image area ratio is a ratio of the area of a pixel image in a current image strip to a reference image strip area, the reference image strip area is the area of the nuclide image in the reference image strip, and the reference image strip is an image strip located in the center of the single-leaf thyroid gland development image.

(4) Obtaining the brightness uniformity parameter of each thyroid imaging image

Uniformly dividing the single-leaf thyroid imaging image into m 'rows of image strips, wherein m' is an odd number, and each image strip comprises a plurality of pixel points; calculating the average gray value of each image strip; calculating the gray ratio of all image strips in the single-leaf thyroid gland development image, wherein the gray ratio of the image strips is the ratio of the average gray value of the current image strip to the reference gray value of the image strips, the reference gray value of the image strips is the average gray value of the reference image strips, and the reference image strips are the image strips located in the center of the single-leaf thyroid development image; extending the reference image strip towards the outer side in any direction, and uniformly dividing the single-leaf thyroid imaging image into n 'rows, wherein n' is an odd number to form an m '× n' block image block; calculating the average gray value of each image block; and calculating the gray ratio of all image blocks in the single-leaf thyroid gland imaging image, wherein the gray ratio of the image blocks is the ratio of the average gray value of the current image block to the reference gray value of the image blocks, the reference gray value of the image blocks is the average gray value of the reference image blocks, and the reference image blocks are the image blocks positioned in the centers of all the image strips.

Compared with the prior art, the technical scheme provided by the application includes that an original image collected by nuclear medicine imaging equipment in a DICOM format is converted into a jpeg format by using Java language, the jpeg format image is subjected to digital processing by using an HTML component, the digital processed image is subjected to preprocessing such as edge removal and character removal, the thyroid boundary and the bilobalt thyroid gland parting line are extracted according to the gray value of each pixel point in the preprocessed image, the accurate bilobalt thyroid gland morphology and the accurate single-lobe thyroid gland morphology are obtained, the accurate bilobalt morphology parameters and the accurate double-lobe brightness parameters are obtained as the single-lobe morphology parameters and the single-lobe brightness parameters, and the problems that data are inaccurate and the like due to the fact that a doctor manually defines a thyroid region through work experience are solved.

The application also provides a system for implementing the above method for processing nuclear medicine thyroid imaging images, the system comprising: the image acquisition unit is used for acquiring a nuclear medicine thyroid imaging image; the image preprocessing unit is used for preprocessing the nuclear medicine thyroid imaging image;

the nuclide image extraction unit is used for extracting a double-leaf thyroid nuclide development image and a single-leaf thyroid nuclide development image from the preprocessed nuclear medicine thyroid development image; and the parameter acquisition unit is used for acquiring the double-leaf morphological parameters and the double-leaf brightness parameters of the double-leaf thyroid nuclide imaging image and the single-leaf morphological parameters and the single-leaf brightness parameters of each thyroid imaging image.

The present application has been described in detail with reference to specific embodiments and illustrative examples, but the description is not intended to limit the application. Those skilled in the art will appreciate that various equivalent substitutions, modifications or improvements may be made to the presently disclosed embodiments and implementations thereof without departing from the spirit and scope of the present disclosure, and these fall within the scope of the present disclosure. The protection scope of this application is subject to the appended claims.

Claims

1. A method of nuclear medicine thyroid imaging image processing, the method comprising:

obtaining a nuclear medicine thyroid imaging image;

preprocessing a thyroid imaging image in nuclear medicine;

extracting a double-leaf thyroid nuclide imaging image and a single-leaf thyroid nuclide imaging image from the preprocessed nuclear medicine thyroid nuclide imaging image;

acquiring double-leaf morphological parameters and double-leaf brightness parameters of a double-leaf thyroid nuclide development image and single-leaf morphological parameters and single-leaf brightness uniformity parameters of each thyroid imaging image, wherein,

the extracting of the bilobalt thyroid nuclide imaging image from the preprocessed nuclear medicine thyroid imaging image comprises the following steps:

extracting a rectangular region comprising a bilobalt thyroid nuclide imaging image from the preprocessed nuclear medicine thyroid imaging image;

extracting a bileaflet thyroid nuclide imaging image from the rectangular region,

further, extracting a bilobalid thyroid nuclide imaging image from the rectangular region comprises:

determining a horizontal center line of the neck;

dividing a gland region in the rectangular region, wherein the gland region is a region below a horizontal central line of a neck in the rectangular region;

if the filtering pixel point is not in the gland region, updating the filtering pixel point, wherein the updated filtering pixel point is the pixel point which obtains the maximum gray value after filtering again in the gland region;

if the filtering pixel point is in the gland region but the distance between the filtering pixel point and the upper boundary of the gland region is smaller than a preset distance value, updating the gland region, and updating the filtering pixel point in the updated gland region, wherein the updated filtering pixel point is the pixel point with the largest gray value in the updated gland region after the updated gland region is subjected to filtering processing again;

marking out the contour line of the double-leaf thyroid nuclide imaging image in the updated gland region by adopting a threshold segmentation method;

and extracting a double-leaf thyroid nuclide development image from the gland region, wherein the double-leaf thyroid nuclide development image is the contour line of the double-leaf thyroid nuclide development image and an internal image thereof.

2. The method of claim 1, wherein preprocessing the nuclear medicine thyroid imaging image comprises:

the method comprises the steps of converting an original image in a DICOM format acquired by nuclear medicine imaging equipment into an image in a jpeg format;

carrying out digital processing on the image in the jpeg format by utilizing an HTML component;

removing frame images around the digital nuclear medicine thyroid imaging image to obtain a borderless image;

and removing the character image in the borderless image.

3. The method of claim 1, wherein extracting a rectangular region comprising a bilobed thyroid species imaging image from the preprocessed nuclear medicine thyroid species imaging image comprises:

carrying out filtering processing on the preprocessed nuclear medicine thyroid imaging image;

acquiring the position coordinates of the filtering pixel points in the filtered image;

and searching thyroid boundary lines and thyroid boundary columns line by line or line by line in four directions from the filtering pixel points to the upper direction, the lower direction, the left direction and the right direction.

4. The method of claim 1, wherein extracting a single-lobe thyroid species imaging image from the preprocessed nuclear medicine thyroid species imaging image comprises:

drawing a thyroid double-lobe boundary in the double-lobe thyroid nuclide imaging image;

and extracting a single-lobe thyroid nuclide development image from the double-lobe thyroid nuclide development image, wherein the single-lobe thyroid nuclide development image is a closed figure and an internal image thereof, which are formed by the boundary line of the thyroid double lobes and the contour line of the double-lobe thyroid nuclide development image on the left side or the right side of the thyroid double lobes in a surrounding manner.

5. The method as claimed in claim 4, wherein the mapping of the thyroid double-lobular demarcation in the thyroid nuclide imaging image comprises:

drawing the contour line of the double-leaf thyroid nuclide imaging image in a rectangular coordinate system;

drawing an average dividing line of the thyroid nuclide imaging image in the thyroid nuclide imaging image;

selecting a central pixel point row near the average dividing line, wherein the central pixel point row is a pixel point row with the minimum vertical coordinate of an upper boundary pixel point in a thyroid nuclide imaging image;

acquiring a longitudinal coordinate of a central pixel point, wherein the longitudinal coordinate of the central pixel point is the longitudinal coordinate of an upper boundary pixel point of the central pixel point row in a thyroid nuclide imaging image;

two rows of upper edge pixel point rows are selected on two sides of the average dividing line, and the upper edge pixel point row is the pixel point row with the maximum vertical coordinate of the upper edge pixel point;

respectively acquiring vertical coordinates of two upper edge pixel points, wherein the vertical coordinates of the upper edge pixel points are vertical coordinates of upper boundary pixel points of the upper edge pixel point row;

respectively calculating upper edge difference values, wherein the upper edge difference values are the difference values of the vertical coordinates of the central pixel point and the vertical coordinates of the two upper edge pixel points;

calculating an upper edge difference height ratio, wherein the upper edge difference height ratio is the ratio of the upper edge difference value to the height of the thyroid nuclide development image;

if the upper edge difference height ratio is greater than or equal to the upper edge difference height ratio preset value, then,

calculating the sum of gray values of all pixel points in the central pixel point row;

respectively calculating the sum of gray values of all pixel points in each row of alternative pixel point rows, wherein the alternative pixel point rows are a plurality of rows near the left side and the right side of the central pixel point row;

selecting a candidate pixel point row, wherein the candidate pixel point row is a pixel point row with the minimum sum of pixel gray values in the candidate pixel point row; calculating the ratio of the gray sums, wherein the ratio of the gray sums is the ratio of the sum of all the gray values of the pixels in the candidate pixel point row to the sum of all the gray values of the pixels in the central pixel point row;

if the ratio of the gray sums is larger than or equal to the preset ratio of the gray sums, determining the position of the thyroid double-lobe boundary, wherein the position of the thyroid double-lobe boundary is a central pixel point row;

and if the ratio of the gray sums is smaller than the preset ratio of the gray sums, drawing a thyroid double-lobe boundary line, wherein the thyroid double-lobe boundary line is a candidate pixel point row.

6. The method of claim 1,

the double-lobe morphological parameters for obtaining the double-lobe thyroid nuclide imaging image comprise:

calculating the total width of the double-leaf thyroid nuclide development image and the height of the double-leaf thyroid nuclide development image, wherein the total width of the double-leaf thyroid nuclide development image is the total number of pixel point rows in the double-leaf thyroid nuclide development image, the height of the double-leaf thyroid nuclide development image is the difference between the ordinate of the upper edge pixel and the ordinate of the lower edge pixel, and the lower edge pixel is the pixel with the minimum lower boundary ordinate; and/or

The double-leaf brightness parameters for obtaining the double-leaf thyroid nuclide imaging image comprise:

acquiring the maximum gray value of a double-leaf thyroid nuclide development image, wherein the maximum gray value of the double-leaf thyroid nuclide development image is the gray value of a pixel point with the maximum gray value in the double-leaf thyroid nuclide development image;

respectively obtaining the maximum gray value of each single-leaf thyroid gland development image and the average gray value of the single-leaf thyroid gland development image, wherein the maximum gray value of the single-leaf thyroid gland development image is the gray value of a pixel point with the maximum gray value in the single-leaf thyroid gland development image, and the average gray value of the single-leaf thyroid gland development image is the average value of the gray values of all the pixel points in the single-leaf thyroid gland development image;

and calculating the ratio of the maximum gray values of the two single-leaf thyroid imaging images in the thyroid nuclide imaging image and the ratio of the average gray values of the two single-leaf thyroid imaging images.

7. The method of claim 1, wherein obtaining single leaf morphological parameters for each leaf thyroid imaging image comprises:

acquiring the width, height and area of each single-leaf thyroid gland development image, wherein the width of each single-leaf thyroid gland development image is the total number of pixel point rows in each single-leaf thyroid gland development image, the height of each single-leaf thyroid gland development image is the difference between the ordinate of an upper edge pixel point and the ordinate of a lower edge pixel point of a thyroid nuclide development image, and the area of each single-leaf thyroid gland development image is calculated by the total number of pixel points in the single-leaf thyroid gland development image;

calculating the aspect ratio of the single-leaf thyroid imaging image;

calculating the ratio of the area of the thyroid gland with a single lobe to the area of the neck visualization image, wherein the area of the neck visualization image is calculated by the square of the width of the neck visualization image;

respectively calculating the width ratio, height ratio and area ratio of the two thyroid imaging images and the ratio of the area of each thyroid imaging image to the area of the neck imaging image; and/or

Acquiring the single-leaf morphological parameters of each thyroid imaging image further comprises:

uniformly dividing the single-leaf thyroid imaging image into m rows of image strips, wherein m is an odd number, and each image strip comprises a plurality of pixel points;

calculating the area of a kernel image in each image strip;

calculating the nuclide image area ratio of each image strip in the single-leaf thyroid imaging image, wherein the nuclide image area ratio is the ratio of the area of a pixel image in the current image strip to the reference area of the image strip, the reference area of the image strip is the area of a nuclide image in the reference image strip, and the reference image strip is an image strip positioned in the center of the single-leaf thyroid imaging image; and/or

The parameters for obtaining the brightness uniformity of the thyroid imaging image of each leaf comprise:

uniformly dividing the single-leaf thyroid imaging image into m 'rows of image strips, wherein m' is an odd number, and each image strip comprises a plurality of pixel points;

calculating the average gray value of each image strip;

calculating the gray ratio of all image strips in the single-leaf thyroid gland development image, wherein the gray ratio of the image strips is the ratio of the average gray value of the current image strip to the reference gray value of the image strips, the reference gray value of the image strips is the average gray value of the reference image strips, and the reference image strips are the image strips located in the center of the single-leaf thyroid development image;

extending the reference image strip towards the outer side in any direction, and uniformly dividing the single-leaf thyroid imaging image into n 'rows, wherein n' is an odd number to form an m '× n' block image block;

calculating the average gray value of each image block;

and calculating the gray ratio of all image blocks in the single-leaf thyroid gland imaging image, wherein the gray ratio of the image blocks is the ratio of the average gray value of the current image block to the reference gray value of the image blocks, the reference gray value of the image blocks is the average gray value of the reference image blocks, and the reference image blocks are the image blocks positioned in the centers of all the image strips.

8. A nuclear medicine thyroid imaging image processing system, characterized in that the system comprises:

the image acquisition unit is used for acquiring a nuclear medicine thyroid imaging image;

the image preprocessing unit is used for preprocessing the nuclear medicine thyroid imaging image;

the nuclide image extraction unit is used for extracting a double-leaf thyroid nuclide development image and a single-leaf thyroid nuclide development image from the preprocessed nuclear medicine thyroid development image;

a parameter acquiring unit for acquiring a bileaf morphological parameter and a bileaf brightness parameter of a bileaf thyroid nuclide imaging image and a single-leaf morphological parameter and a single-leaf brightness uniformity parameter of each bileaf thyroid nuclide imaging image, wherein,

the nuclide image extraction unit is specifically configured to:

determining a horizontal center line of the neck;