Paravertebral muscle fat segmentation method, device and equipment based on magnetic resonance T2 axial position map and medium

Abstract

The invention provides a paravertebral muscle fat segmentation method, device and equipment based on a magnetic resonance T2 axial position map and a medium, and relates to the technical field of medical image processing. The method comprises the following steps: acquiring image data, wherein the image data is a T2 axial position sequence lumbar vertebra image of a target object acquired by hospital magnetic resonance imaging equipment; acquiring labeling information of the paravertebral muscle area to obtain image data labeled with the paravertebral muscle area; performing image preprocessing on the image data marked with the paravertebral muscle areas, and performing image enhancement on each paravertebral muscle area to obtain paravertebral muscle area enhanced images; based on a PSO-Kapur threshold segmentation algorithm, carrying out image segmentation on the paravertebral muscle region enhanced image, and determining a muscle fat segmentation result of each paravertebral muscle; and superposing the muscle fat segmentation result of each paravertebral muscle to the T2 axial position sequence lumbar vertebra image, and outputting a visual segmentation result graph. Therefore, intelligent segmentation of paravertebral muscle fat can be realized.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of medical image processing, and in particular to a method, device, equipment and medium for paraspinal muscle fat segmentation based on magnetic resonance T2 axial images. Background Art

[0002] As the global population grows and aging issues become increasingly prominent, low back pain is considered the second most common disease after the common cold and the main cause of disability. The main cause of low back pain is degenerative changes in the lumbar paraspinal muscles, which are usually caused by fatty infiltration of the paraspinal muscles. Therefore, measuring the degree of fatty infiltration of the paraspinal muscles can assist in the diagnosis of low back pain and effectively prevent a series of lumbar specific and non-specific diseases and other symptoms.

[0003] Measuring the cross-sectional area of ​​the fat region is the core of measuring the degree of paraspinal muscle fat infiltration, but there is currently little research on the optimization algorithm for paraspinal muscle fat segmentation. Summary of the invention

[0004] A first aspect of an embodiment of the present invention provides a paraspinal muscle fat segmentation method based on a magnetic resonance T2 axial image, the method comprising:

[0005] Acquire image data, wherein the image data is a T2 axial sequence lumbar spine image of the target object acquired by a magnetic resonance imaging device in a hospital;

[0006] Acquire the annotation information of the paraspinal muscle region, and obtain the image data with the paraspinal muscle region annotated;

[0007] Performing image preprocessing on the image data marked with the paraspinal muscle regions, performing image enhancement on each paraspinal muscle region, and obtaining an enhanced image of the paraspinal muscle region;

[0008] Performing image segmentation on the paraspinal muscle region enhanced image based on the PSO-Kapur threshold segmentation algorithm to determine the muscle fat segmentation results of each paraspinal muscle;

[0009] The muscle fat segmentation results of each paraspinal muscle are superimposed on the T2 axial sequence lumbar spine image, and a visual segmentation result map is output.

[0010] Optionally, acquiring image data includes:

[0011] Acquire an L4-L5 regional image, wherein the L4-L5 regional image is obtained based on a T2 axial sequence lumbar spine image of the target object combined with a sagittal image;

[0012] Get annotation information of the paraspinal muscle area, including:

[0013] Obtain annotation information for each paraspinal muscle region in the L4-L5 region image, wherein each paraspinal muscle region includes at least one of the following: left erector spinae, left multifidus, left psoas major, right erector spinae, right multifidus, right psoas major, and an intervertebral disc region.

[0014] Optionally, image preprocessing is performed on the image data with paraspinal muscle regions annotated thereon, and image enhancement is performed on each paraspinal muscle region, including:

[0015] Calculate the segmentation area rectangle of each paraspinal muscle area using the marked paraspinal muscle area;

[0016] Based on the rectangular frames of each segmented area, the images in each segmented area are retained, the grayscale values ​​of pixels outside the segmented area are set to 0, and the CLAHE image enhancement algorithm is used to highlight the local details and image contrast of the image in the segmented area.

[0017] Optionally, a CLAHE image enhancement algorithm is used to highlight local details and image contrast of the image within the segmented area, including:

[0018] For an image within a segmented region:

[0019] dividing the image into a plurality of overlapping or non-overlapping tiles;

[0020] Calculate the contrast function of each tile to adjust the pixel value of each tile;

[0021] Merge adjacent adjusted tiles to avoid borders;

[0022] The histogram of the merged image block is cropped; when the value of the cropped histogram segment is higher than the preset contrast value, the pixels corresponding to the segment are evenly distributed to other segments to limit the contrast and balance the image;

[0023] Bilinear interpolation is applied to the cropped and merged images to remove artifacts at tile boundaries.

[0024] Optionally, performing image segmentation on the paraspinal muscle region enhanced image based on a PSO-Kapur threshold segmentation algorithm includes:

[0025] The enhanced images of each paraspinal muscle region were segmented using the PSO-Kapur algorithm;

[0026] Kapur entropy is used as the objective function to maximize the objective function, and the particle swarm optimization algorithm is used to optimize the running speed of Kapur multi-threshold segmentation to obtain a value close to the optimal solution;

[0027] Among them, the threshold number of Kapur multi-threshold segmentation is 2.

[0028] Optionally, the Kapur entropy is used as the objective function to maximize the objective function, including:

[0029] Use multiple thresholds to divide the image into different segments. The following is the Kapur entropy for each threshold interval:

[0030] ......

[0032] ......

[0034]

[0035] Among them, p i is the probability of event i occurring, H represents Kapur entropy, a, b, c…j…m represent thresholds, p a , p b ,…,p m The grayscale probability of different threshold intervals is represented by a multi-threshold method;

[0036] In order to select the optimal threshold, the following objective function should be maximized:

[0037]

[0038] H i (th) represents the Kapur entropy in the i-th threshold interval when the threshold is th;

[0039] To extend the multi-level threshold expression, Kapur’s objective function is defined as:

[0040] f Kapur (TH) = f Kapur (th i ),i=1,2,3,…,k;

[0041] Among them, different thresholds are represented by TH[th1, th2, th3, ..., thk-1], and i corresponds to a specific threshold interval.

[0042] Optionally, a particle swarm optimization algorithm is used to optimize the running speed of Kapur multi-threshold segmentation to obtain a value close to the optimal solution, including:

[0043] Generate initial position and initial velocity;

[0044] Evaluate the objective function for each particle;

[0045] Update the single optimal solution and the global optimal solution, the single optimal solution is the single optimal particle, and the global optimal solution is the global optimal particle

[0046] Update the velocity v and position x of each particle;

[0047] Repeat the particle swarm optimization process until the termination condition is reached;

[0048] In the particle swarm optimization algorithm, the particle velocity update formula is as follows:

[0049] v j (k+1)=wv j (k)+c1r1(p j (k)-x j (k))+c2r2(g(k)-x j (k)),j=1,2,…,np.

[0050] In the particle swarm optimization algorithm, the particle position update formula is as follows:

[0051] x j (k+1)=x j (k)+v j (k);

[0052] Among them, v j (k) and x j (k) are the velocity and position of the jth particle at the kth iteration, respectively, and p j g(k) and g(k) are the single best particle and the global best particle at the kth iteration, respectively; w is the initial weight parameter, c1 is the individual cognitive parameter, and c2 is the social cognitive parameter; r1 and r2 are random numbers between 0 and 1 with uniform distribution.

[0053] A second aspect of an embodiment of the present invention provides a paraspinal muscle fat segmentation device based on magnetic resonance T2 axial image, the device comprising:

[0054] An image data acquisition module, used to acquire image data, wherein the image data is a T2 axial sequence lumbar spine image of the target object acquired by a magnetic resonance imaging device in a hospital;

[0055] The labeling information acquisition module is used to acquire the labeling information of the paraspinal muscle area and obtain the image data with the paraspinal muscle area labeled;

[0056] An image preprocessing module is used to perform image preprocessing on the image data marked with the paraspinal muscle area, perform image enhancement on each paraspinal muscle area, and obtain an enhanced image of the paraspinal muscle area;

[0057] A segmentation module, used for performing image segmentation on the paraspinal muscle region enhanced image based on a PSO-Kapur threshold segmentation algorithm to determine the muscle fat segmentation result of each paraspinal muscle;

[0058] The output module is used to superimpose the muscle fat segmentation results of each paraspinal muscle on the T2 axial sequence lumbar spine image and output a visual segmentation result map.

[0059] Optionally, the image data acquisition module is specifically used to acquire L4-L5 regional images, wherein the L4-L5 regional images are obtained based on the target object's T2 axial sequence lumbar spine images combined with sagittal images;

[0060] The annotation information acquisition module is specifically used to obtain annotation information for each paraspinal muscle area in the L4-L5 area image, and each paraspinal muscle area includes at least one of the following: left erector spinae, left multifidus, left psoas major, right erector spinae, right multifidus, right psoas major and intervertebral disc area.

[0061] Optionally, the image preprocessing module includes:

[0062] The paraspinal muscle region segmentation submodule is used to calculate the segmentation region rectangular frame of each paraspinal muscle region using the marked paraspinal muscle region;

[0063] The enhancement submodule is used to retain the image in each segmented area based on the rectangular frame of each segmented area, set the grayscale value of pixels outside the segmented area to 0, and use the CLAHE image enhancement algorithm to highlight the local details and image contrast of the image in the segmented area.

[0064] Optionally, the enhancer module is specifically used to:

[0065] For an image within a segmented region:

[0066] dividing the image into a plurality of overlapping or non-overlapping tiles;

[0067] Calculate the contrast function of each tile to adjust the pixel value of each tile;

[0068] Merge adjacent adjusted tiles to avoid borders;

[0069] The histogram of the merged image block is cropped; when the value of the cropped histogram segment is higher than the preset contrast value, the pixels corresponding to the segment are evenly distributed to other segments to limit the contrast and balance the image;

[0070] Bilinear interpolation is applied to the cropped and merged images to remove artifacts at tile boundaries.

[0071] Optionally, the image segmentation module includes:

[0072] The enhanced image segmentation submodule is used to segment the enhanced images of each paraspinal muscle region using the PSO-Kapur algorithm;

[0073] The solution submodule is used to maximize the objective function by taking Kapur entropy as the objective function, and to optimize the running speed of Kapur multi-threshold segmentation by using particle swarm optimization algorithm to obtain a value close to the optimal solution;

[0074] Among them, the threshold number of Kapur multi-threshold segmentation is 2.

[0075] Optionally, the solution submodule is specifically used for:

[0076] Use multiple thresholds to divide the image into different segments. The following is the Kapur entropy for each threshold interval:

[0077] ......

[0079] ......

[0081]

[0082] Among them, ρ i is the probability of event i occurring, H represents Kapur entropy, a, b, c…j…m represent thresholds, p a , p b ,…,p m The grayscale probability of different threshold intervals is represented by a multi-threshold method;

[0083] In order to select the optimal threshold, the following objective function should be maximized:

[0084]

[0085] H i (th) represents the Kapur entropy in the i-th threshold interval when the threshold is th;

[0086] To extend the multi-level threshold expression, Kapur’s objective function is defined as:

[0087] f Kapur (TH) = f Kapur (th i ),i=1,2,3,…,k;

[0088] Among them, different thresholds are represented by TH[th1, th2, th3, ..., thk-1], and i corresponds to a specific threshold interval.

[0089] Optionally, the solution submodule is further used for:

[0090] Generate initial position and initial velocity;

[0091] Evaluate the objective function for each particle;

[0092] Update the single optimal solution and the global optimal solution, the single optimal solution is the single optimal particle, and the global optimal solution is the global optimal particle

[0093] Update the velocity v and position x of each particle;

[0094] Repeat the particle swarm optimization process until the termination condition is reached;

[0095] In the particle swarm optimization algorithm, the particle velocity update formula is as follows:

[0096] v j (k+1)=wv j (k)+c1r1(p j (k)-x j (k))+c2r2(g(k)-x j (k)),j=1,2,…,np.

[0097] In the particle swarm optimization algorithm, the particle position update formula is as follows:

[0098] x j (k+1)=x j (k)+v j (k);

[0099] Among them, v j (k) and x j (k) are the velocity and position of the jth particle at the kth iteration, respectively, and p j g(k) and g(k) are the single best particle and the global best particle at the kth iteration, respectively; w is the initial weight parameter, c1 is the individual cognitive parameter, and c2 is the social cognitive parameter; r1 and r2 are random numbers between 0 and 1 with uniform distribution.

[0100] A third aspect of an embodiment of the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when executed, the processor implements the steps in the paraspinal muscle fat segmentation method based on magnetic resonance T2 axial imaging as described in the first aspect of the present invention.

[0101] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the paraspinal muscle fat segmentation method based on magnetic resonance T2 axial images as described in the first aspect of the present invention.

[0102] The paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image provided by the embodiment of the present invention can automatically segment fat areas for each paraspinal muscle area after determining the marked paraspinal muscle area. In addition, the segmentation method provided by the embodiment of the present invention has been tested by Bland-Altman consistency test, proving that the segmentation result is highly consistent with the manual segmentation. In addition, the running time of the segmentation algorithm proposed by the embodiment of the present invention is lower than that of other methods in the related art, and the segmentation efficiency is much higher than that of manual segmentation. BRIEF DESCRIPTION OF THE DRAWINGS

[0103] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments of the present invention will be briefly introduced below. Obviously, the accompanying drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0104] Figure 1 A flowchart of the steps of a paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image provided by an embodiment of the present invention is shown;

[0105] Figure 2 The image data with paraspinal muscle regions marked thereon obtained in the paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image provided by an embodiment of the present invention is shown;

[0106] Figure 3 The figure shows various paraspinal muscle regions obtained based on an image enhancement algorithm in a paraspinal muscle fat segmentation method based on a magnetic resonance T2 axial image provided by an embodiment of the present invention;

[0107] Figure 4 A Bland-Altman result diagram showing the muscle fat segmentation results of each paraspinal muscle obtained in the paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image provided by an embodiment of the present invention

[0108] Figure 5 FA-Otsu, PSO-Otsu, FA-Kapur, and PSO-Kapur iteration curves are shown;

[0109] Figure 6 The running time of the optimization algorithms of Otsu, K-Means, SpectralClustering, Birch, Otsu multi-threshold segmentation (k=2), Kapur multi-threshold segmentation (k=2), FA_Kapur multi-threshold segmentation (k=2), and PSO_Kapur multi-threshold segmentation (k=2) are shown;

[0110] Figure 7A schematic diagram of a visualized segmentation result outputted from a paraspinal muscle fat segmentation method based on a magnetic resonance T2 axial image provided by an embodiment of the present invention is shown;

[0111] Figure 8 The structure block diagram of the paraspinal muscle fat segmentation device based on magnetic resonance T2 axial image provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0112] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0113] In an embodiment of the present invention, a method for segmenting paraspinal muscle fat based on magnetic resonance T2 axial images is provided. Figure 1 As shown, it shows a flowchart of the steps of a paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image according to an embodiment of the present invention, the method comprising:

[0114] S101, acquiring image data, wherein the image data is T2 axial sequence lumbar spine images of a target object acquired by a magnetic resonance imaging device in a hospital.

[0115] S102, obtaining the labeling information of the paraspinal muscle region, and obtaining the image data with the labeling of the paraspinal muscle region.

[0116] In an embodiment of the present invention, the method can be executed by a server, and the server can correspond to a client, through which a user can input a T2 axial sequence lumbar spine image of a target object acquired by a hospital magnetic resonance imaging device. The server can also be directly connected to the hospital magnetic resonance imaging device for communication, and the image acquired by the hospital magnetic resonance imaging device is directly sent to the server.

[0117] In an embodiment of the present invention, after acquiring the image data, the server can display the image data to professional technicians (e.g., doctors) through the client, so that the professional technicians can mark the paraspinal muscle area based on the image data. The client can also receive the marking information of the paraspinal muscle area provided by the professional technicians.

[0118] Thus, the server can obtain the image data with the paraspinal muscle region annotated based on the annotation of the paraspinal muscle region input by the professional technician, which can be achieved by the professional technician drawing a circle on the image.

[0119] Specifically, in step S101, acquiring image data includes: acquiring L4-L5 region images, wherein the L4-L5 region images are obtained based on T2 axial sequence lumbar vertebrae images of the target object combined with sagittal images.

[0120] Specifically, the image data can also be displayed to professional technicians so that they can determine the image in the L4-L5 region based on their experience and the sagittal image. The image in the L4-L5 region is then displayed to professional technicians so that they can annotate the paraspinal muscle region based on the L4-L5 region image. The client can also receive the annotated information of the paraspinal muscle region provided by the professional technicians.

[0121] In this case, step S102 includes: obtaining annotation information for each paraspinal muscle area in the L4-L5 regional image, wherein each paraspinal muscle area includes at least one of the following: left erector spinae muscle LES, left multifidus muscle LMM, left psoas major muscle LPM, right erector spinae muscle RES, right multifidus muscle RMM, right psoas major muscle RPM, and intervertebral disc area ID. Figure 2 The figure shows the image data with paraspinal muscle regions marked thereon obtained in the paraspinal muscle fat segmentation method based on magnetic resonance T2 axial images according to an embodiment of the present invention.

[0122] In the embodiment of the present invention, the acquired annotation information may be obtained by annotation and correction by multiple professional technicians.

[0123] S103, performing image preprocessing on the image data with the paraspinal muscle regions marked thereon, performing image enhancement on each paraspinal muscle region, and obtaining a paraspinal muscle region enhanced image.

[0124] In the present embodiment, any mature image enhancement algorithm in the relevant technology (for example, histogram equalization, adaptive histogram equalization, contrast enhancement, sharpening filtering, etc.) can be used to enhance the image of each paraspinal muscle region to obtain an enhanced image of the paraspinal muscle region.

[0125] Specifically, the embodiment of the present invention adopts CLAHE (Contrast Limited Adaptive Histogram Equalization) image enhancement algorithm to achieve image enhancement of each paraspinal muscle region, enhance the contrast of the image of each paraspinal muscle region and improve the image details. The CLAHE algorithm can effectively enhance the details and contrast of the image by performing adaptive histogram equalization on each small block in the image and limiting the contrast, while avoiding over-enhancement or distortion. It can be applied to various types of image enhancement tasks, especially to images with uneven contrast distribution.

[0126] Specifically, step S103 includes:

[0127] S1031, using the marked paraspinal muscle regions, calculate the segmentation region rectangular frame of each paraspinal muscle region.

[0128] S1032, based on the rectangular frames of each segmented area, retain the image in each segmented area, set the remaining areas to 0, and use the CLAHE image enhancement algorithm to highlight the local details and image contrast of the image in the segmented area.

[0129] Specifically, in S1032, the images in the segmented area corresponding to each paraspinal muscle area are processed respectively. Specifically, for the images in one segmented area, step S1032 includes:

[0130] S10321, dividing the image into a plurality of overlapping or non-overlapping blocks.

[0131] In the embodiment of the present invention, a preset segmentation rule may be used to divide the image into a plurality of overlapping or non-overlapping tiles, for example, with a size of 8x8 or 16x16 pixels.

[0132] S10322, calculating a contrast function of each image block to adjust the pixel value of each image block.

[0133] Specifically, a histogram of the gray level of each image block may be calculated, and histogram equalization may be performed on two or more pixels in each image block to increase contrast and clarity.

[0134] In the embodiment of the present invention, the CLAHE image enhancement algorithm can enhance the contrast of the image by performing adaptive histogram equalization on the image blocks in the segmented area, so that the details in the image in the segmented area are clearer and more prominent.

[0135] In the embodiment of the present invention, the CLAHE image enhancement algorithm can effectively enhance the local contrast of the image in the segmented area, thereby improving the detail information of the image in the segmented area, making the fine structure in the image in the segmented area more prominent.

[0136] S10323: merge adjacent adjusted tiles to avoid generating boundaries.

[0137] S10324, cropping the histogram of the merged image block; when the value of the cropped histogram segment is higher than the preset contrast value, evenly distributing the pixels corresponding to the segment to other segments to limit the contrast and balance the image.

[0138] S10325, applying bilinear interpolation to the cropped and merged image to eliminate artifacts in the tile boundary.

[0139] Figure 3The paraspinal muscle regions obtained based on the image enhancement algorithm in the paraspinal muscle fat segmentation method based on the magnetic resonance T2 axial image in the embodiment of the present invention. Wherein, Raw represents the unprocessed image of each paraspinal muscle region, and HE represents the image of each paraspinal muscle region obtained based on the HE image enhancement algorithm provided by the embodiment of the present invention. CLAHE represents the image of each paraspinal muscle region obtained based on the CLAHE image enhancement algorithm provided by the embodiment of the present invention. It can be seen from the figure that the CLAHE image enhancement algorithm can effectively enhance the detail information of the image in the segmented area, making the fat distribution in the image in the segmented area more obvious.

[0140] In the embodiment of the present invention, the CLAHE image enhancement algorithm can perform image enhancement in a local area, which can effectively reduce the illumination non-uniformity effect that may exist in the image in the segmented area, making the overall image brightness more uniform. The CLAHE image enhancement algorithm can perform adaptive processing based on the local features of the image, and can better adapt to the characteristics of the image in different scenes, thereby improving the adaptability and flexibility of the image enhancement effect.

[0141] S104, performing image segmentation on the paraspinal muscle region enhanced image based on a PSO-Kapur threshold segmentation algorithm to determine a muscle fat segmentation result of each paraspinal muscle.

[0142] In the embodiment of the present invention, the optimal threshold of the image can be automatically determined to achieve adaptive segmentation of the image.

[0143] The PSO-Kapur algorithm (an image segmentation algorithm that combines particle swarm optimization (PSO) and Kapur entropy) uses the PSO algorithm to automatically search for the optimal threshold to maximize Kapur entropy (an indicator that measures the difference between image classes), thereby determining the optimal segmentation threshold and achieving adaptive image segmentation.

[0144] Specifically, step S104 includes:

[0145] S1041, the enhanced images of each paraspinal muscle region are segmented using the PSO-Kapur algorithm.

[0146] S1042, taking Kapur entropy as the objective function to maximize the objective function, and using the particle swarm optimization algorithm to optimize the running speed of Kapur multi-threshold segmentation to obtain a value close to the optimal solution.

[0147] In the embodiment of the present invention, the number of thresholds of Kapur multi-threshold segmentation is 2. The problem of how to separate different categories can be solved based on Kapur multi-threshold segmentation.

[0148] Specifically, the grayscale range of multiple images is defined based on different thresholds, and then the grayscale frequency of each grayscale pixel in the entire image is calculated. Kapur entropy is used to measure the inter-class differences between image grayscale levels under different thresholds. Kapur entropy is a metric that can evaluate image segmentation results. The larger its value, the better the image segmentation effect. The particle swarm optimization algorithm (PSO algorithm) is used for search. The PSO algorithm updates the position and speed of particles in each iteration to find the optimal solution by simulating the process of birds searching for the best food location. In the PSO-Kapur algorithm, each particle represents a potential optimal threshold. The position of the particle is continuously updated by optimizing the Kapur entropy until the optimal solution, that is, the optimal segmentation threshold, is reached. The particles gradually approach the optimal solution in continuous iterations by updating their speed and position in the search space. In this process, the optimal threshold found is also updated. Finally, the PSO-Kapur algorithm will find the optimal segmentation threshold that can maximize the Kapur entropy, thereby achieving adaptive image segmentation.

[0149] Specifically, in the embodiment of the present invention, step S1042 includes:

[0150] S10421, use multiple thresholds to divide the image into different segments, the following is the Kapur entropy of each threshold interval:

[0151] ......

[0153] ......

[0155]

[0156] Among them, p i is the probability of event i occurring, H represents Kapur entropy, a, b, c…j…m represent thresholds, p a , p b ,…,p m The grayscale probability of different threshold intervals is represented by a multi-threshold method;

[0157] In order to select the optimal threshold, the following objective function should be maximized:

[0158]

[0159] H i (th) represents the Kapur entropy in the i-th threshold interval when the threshold is th;

[0160] To extend the multi-level threshold expression, Kapur’s objective function is defined as:

[0161] f Kapur (TH) = f Kapur (th i ),i=1,2,3,…,k;

[0162] Among them, different thresholds are represented by TH[th1, th2, th3, ..., thk-1], and i corresponds to a specific threshold interval.

[0163] In the embodiment of the present invention, the multi-threshold Kapur algorithm can be used for different paraspinal muscle region enhanced images to obtain the corresponding thresholds of the left erector spinae, left multifidus, left psoas major, right erector spinae, right multifidus, and right psoas major, which are 56-84, 45-88, 36-73, 51-85, 21-36, and 12-47, respectively.

[0164] Specifically, in the embodiment of the present invention, the step S1042 further includes:

[0165] S10422, generate initial position and initial velocity.

[0166] S10423, evaluate the objective function for each particle.

[0167] S10424, updating a single optimal solution and a global optimal solution, wherein the single optimal solution is a single optimal particle, and the global optimal solution is a global optimal particle.

[0168] S10425, update the velocity v and position x of each particle.

[0169] S10425, repeat the particle swarm optimization process until the termination condition is reached.

[0170] In the present embodiment, the termination condition may be: maximizing Kapur entropy.

[0171] In the particle swarm optimization algorithm, the particle velocity update formula is as follows:

[0172] v j (k+1)=wv j (k)+c1r1(p j (k)-x j (k))+c2r2(g(k)-x j (k)),j=1,2,…,np.

[0173] In the particle swarm optimization algorithm, the particle position update formula is as follows:

[0174] x j (k+1)=x j (k)+v j (k);

[0175] Among them, v j (k) and x j (k) are the velocity and position of the jth particle at the kth iteration, respectively, and p j g(k) and g(k) are the single best particle and the global best particle at the kth iteration, respectively; w is the initial weight parameter, c1 is the individual cognitive parameter, and c2 is the social cognitive parameter; r1 and r2 are random numbers between 0 and 1 with uniform distribution.

[0176] In the embodiment of the present invention, the PSO algorithm uses the wisdom of the group to search for the optimal solution by simulating the behavior of biological groups such as bird flocks or fish schools. It can find the global optimal solution or the approximate optimal solution in the search space, and has a strong global search capability for complex multi-dimensional optimization problems. The PSO algorithm uses an iterative method to perform optimization search, and gradually optimizes the objective function value by continuously updating the position and speed of the particles until the stop condition is reached. This iterative optimization process enables the PSO algorithm to quickly find a better solution in the search space. The PSO algorithm is adaptive and can adjust the movement strategy of the particles according to the results of the search process to explore the search space more effectively. This adaptability makes the PSO algorithm have good adaptability and robustness for different types of optimization problems. The implementation of the PSO algorithm is relatively simple, does not require complex mathematical models and a large number of parameter adjustments, and has good parallelization characteristics. It can be easily applied to a parallel computing environment to accelerate the optimization process and improve search efficiency.

[0177] In the embodiment of the present invention, the PSO optimization algorithm can be used to optimize the running speed of Kapur multi-threshold segmentation, and a value close to the optimal solution can be obtained with a number of iterations far less than that of the enumeration method.

[0178] In the embodiment of the present invention, during the operation of the PSO optimization algorithm, the number of particles is set to 50, the maximum number of iterations is 30, w=1, c1=c2=2.2. The optimal fitness of the image is 0.93, and the optimal solution is that the left erector spinae, left multifidus, left psoas, right erector spinae, right multifidus, and right psoas correspond to thresholds of 53 85, 44 88, 35 73, 52 85, 24 37, and 11 48, respectively.

[0179] In the embodiment of the present invention, two evaluation indicators, Jaccard Similarity (JS) and Segmentation Accuracy (SA), are used to evaluate the performance of image segmentation.

[0180]

[0181] S1 is the true result of each muscle area, and S2 is the predicted result of each muscle area.

[0182]

[0183] Where S i is the actual result of each muscle area, S i ′ corresponds to S i Prediction results for muscle area

[0184] Tables 1 and 2 evaluate the experimental results of each muscle under each unsupervised segmentation algorithm from the two segmentation indicators of JS and SA respectively. The algorithms involved in the comparison are from top to bottom: single threshold segmentation without image enhancement (Otsu_noCLAHE) and single threshold segmentation under CLAHE image enhancement (Otsu), K-Means, SpectraClustering (SC), Birch, Otsu multi-threshold algorithm (Otsu_k) and Kapur algorithm (Kapur_k), where multi-threshold segmentation k = 2. Compared with the two evaluation indicators of JS and SA, the PSO-Kapur algorithm performs relatively well in the multifidus erector spinae, but is inferior to the multi-threshold Otsu algorithm in the segmentation of the psoas major. From the perspective of segmentation results, doctors are often more concerned about whether the fat infiltration measurement of the multifidus and erector spinae muscles is accurate. Therefore, the PSO-Kapur algorithm performs better than the Otsu multi-threshold and other unsupervised algorithms.

[0185] Table 1. Comparison of evaluation index JS

[0186]

[0187]

[0188] Table 2 Comparison of evaluation index SA

[0189] S Otsu_noCLAHE 0.95 Otsu 0.95 K-Means 0.95 SpectralClustering 0.95 Birch 0.98 Otsu_k 0.98 Kapur_k 0.98

[0190] In order to compare the threshold segmentation results, the experiment selected 20 images to manually segment the fat area and compared the segmentation results of the PSO-Kapur algorithm with the manual segmentation results. Figure 4 The Bland-Altman result diagram of the muscle fat segmentation results of each paraspinal muscle obtained in the paraspinal muscle fat segmentation method based on the magnetic resonance T2 axial image provided in an embodiment of the present invention, from top to bottom are the Bland-Altman result diagrams of the functional muscle areas of the erector spinae muscles, multifidus muscles, and psoas major muscles on both sides, proving the consistency of manual segmentation and PSO-Kapur algorithm segmentation.

[0191] The optimization effect of the optimization algorithm is reflected in the fact that compared with traversing the values ​​of the entire solution space, only fewer solution space values ​​need to be calculated to obtain the desired results. Therefore, the experiment verifies the performance of the algorithm used in reducing segmentation time and optimization performance by comparing the running time and iteration curves. Figure 5 The iterative curves of FA-Otsu (Feature Aggregation Otsu, a feature selection method based on the Otsu algorithm), PSO-Otsu (an image threshold selection method based on particle swarm optimization (Particle Swarm Optimization) and the Otsu algorithm), FA-Kapur (Feature Aggregation Kapur, a feature selection method based on the Kapur algorithm), and PSO-Kapur are shown respectively, indicating that the PSO-Kapur algorithm has a higher optimization efficiency than the optimization algorithms such as FA-Otsu. Figure 6 The running times of the optimization algorithms of Otsu, K-Means, SpectralClustering, birch, Otsu multi-threshold segmentation (k=2), Kapur multi-threshold segmentation (k=2), FA_Kapur multi-threshold segmentation (k=2), and PSO_Kapur multi-threshold segmentation (k=2) are shown respectively, indicating that the PSO-Kapur algorithm has a lower algorithm running time than unsupervised algorithms such as Otsu.

[0192] S105, superimposing the muscle fat segmentation results of each paraspinal muscle onto the T2 axial sequence lumbar vertebrae image, and outputting a visualized segmentation result graph.

[0193] Specifically, in step S104, the muscle fat segmentation results of the paraspinal muscles can be obtained based on the PSO-Kapur segmentation results. The segmentation results are combined with the coordinates of their respective rectangular boxes and superimposed on the corresponding area of ​​a picture. The rectangular box is represented by [x, y, w, h], where x and y are the horizontal and vertical coordinates of the upper left corner of the rectangular box, w and h are the width and height of the rectangular box. The rectangular box areas of the left erector spinae, left multifidus, left psoas major, right erector spinae, right multifidus, and right psoas major are [79,348,156,100], [274,329,165,118], [198,356,47,70], [261,341,52,84], [152,238,38,89], [323,237,29,68], which are then superimposed on the image to be segmented, and the obtained image is used as a visualization diagram of the segmentation result. Figure 7 The left picture is the image enhancement result, and the right picture is the segmentation result visualization.

[0194] Based on the same inventive concept, the embodiment of the present invention also provides a paraspinal muscle fat segmentation device based on magnetic resonance T2 axial image, such as Figure 8As shown, it shows a structural block diagram of a paraspinal muscle fat segmentation device based on magnetic resonance T2 axial image according to an embodiment of the present invention, and the device comprises:

[0195] An image data acquisition module 801 is used to acquire image data, wherein the image data is a T2 axial sequence lumbar spine image of a target object acquired by a magnetic resonance imaging device in a hospital;

[0196] The labeling information acquisition module 802 is used to acquire the labeling information of the paraspinal muscle area and obtain the image data labeled with the paraspinal muscle area;

[0197] An image preprocessing module 803 is used to perform image preprocessing on the image data marked with the paraspinal muscle regions, perform image enhancement on each paraspinal muscle region, and obtain an enhanced image of the paraspinal muscle region;

[0198] A segmentation module 804 is used to perform image segmentation on the paraspinal muscle region enhanced image based on a PSO-Kapur threshold segmentation algorithm to determine a muscle fat segmentation result of each paraspinal muscle;

[0199] The output module 805 is used to superimpose the muscle fat segmentation results of each paraspinal muscle onto the T2 axial sequence lumbar spine image and output a visualized segmentation result graph.

[0200] Based on the same inventive concept, an embodiment of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the paraspinal muscle fat segmentation method based on the magnetic resonance T2 axial image described in any of the above embodiments are implemented.

[0201] Based on the same inventive concept, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image described in any of the above embodiments are implemented.

[0202] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0203] It will be appreciated by those skilled in the art that the embodiments of the present invention may be provided as methods, devices, or computer program products. Therefore, the embodiments of the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0204] The embodiments of the present invention are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (apparatus), and computer program products according to the embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0205] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0206] These computer program instructions can also be loaded onto a computer or other programmable terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0207] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0208] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0209] The above is a detailed introduction to the paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image provided by the present invention. This article uses specific examples to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A method for segmenting paraspinal muscle fat based on magnetic resonance T2 axial images, characterized in that: include: Acquire image data, wherein the image data is a T2 axial sequence lumbar spine image of the target object acquired by a magnetic resonance imaging device in a hospital; Acquire the annotation information of the paraspinal muscle region, and obtain the image data with the paraspinal muscle region annotated; Performing image preprocessing on the image data marked with the paraspinal muscle regions, performing image enhancement on each paraspinal muscle region, and obtaining an enhanced image of the paraspinal muscle region; Performing image segmentation on the paraspinal muscle region enhanced image based on the PSO-Kapur threshold segmentation algorithm to determine the muscle fat segmentation results of each paraspinal muscle; The muscle fat segmentation results of each paraspinal muscle are superimposed on the T2 axial sequence lumbar spine image, and a visual segmentation result map is output.

2. The paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image according to claim 1, characterized in that: Acquiring image data includes: Acquire an L4-L5 regional image, wherein the L4-L5 regional image is obtained based on a T2 axial sequence lumbar spine image of the target object combined with a sagittal image; Get annotation information of the paraspinal muscle area, including: Obtain annotation information for each paraspinal muscle region in the L4-L5 region image, wherein each paraspinal muscle region includes at least one of the following: left erector spinae, left multifidus, left psoas major, right erector spinae, right multifidus, right psoas major, and an intervertebral disc region.

3. The paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image according to claim 1, characterized in that: Performing image preprocessing on the image data with the paraspinal muscle regions marked thereon, and performing image enhancement on each paraspinal muscle region, including: Calculate the segmentation area rectangle of each paraspinal muscle area using the marked paraspinal muscle area; Based on the rectangular frames of each segmented area, the images in each segmented area are retained, the grayscale values ​​of pixels outside the segmented area are set to 0, and the CLAHE image enhancement algorithm is used to highlight the local details and image contrast of the image in the segmented area.

4. The paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image according to claim 3, characterized in that: The CLAHE image enhancement algorithm is used to highlight the local details and image contrast of the image in the segmented area, including: For an image within a segmented region: dividing the image into a plurality of overlapping or non-overlapping tiles; Calculate the contrast function of each tile to adjust the pixel value of each tile; Merge adjacent adjusted tiles to avoid borders; The histogram of the merged image block is cropped; when the value of the cropped histogram segment is higher than the preset contrast value, the pixels corresponding to the segment are evenly distributed to other segments to limit the contrast and balance the image; Bilinear interpolation is applied to the cropped and merged images to remove artifacts at tile boundaries.

5. The paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image according to claim 1, characterized in that: Performing image segmentation on the paraspinal muscle region enhanced image based on the PSO-Kapur threshold segmentation algorithm includes: The enhanced images of each paraspinal muscle region were segmented using the PSO-Kapur algorithm; Kapur entropy is used as the objective function to maximize the objective function, and the particle swarm optimization algorithm is used to optimize the running speed of Kapur multi-threshold segmentation to obtain a value close to the optimal solution; Among them, the threshold number of Kapur multi-threshold segmentation is 2.

6. The paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image according to claim 5, characterized in that: Taking Kapur entropy as the objective function to maximize the objective function includes: Use multiple thresholds to divide the image into different segments. The following is the Kapur entropy for each threshold interval: Among them, p i is the probability of event i occurring, H represents Kapur entropy, a, b, c…j…m represent thresholds, p a , p b ,…,p m The grayscale probability of different threshold intervals is represented by a multi-threshold method; In order to select the optimal threshold, the following objective function should be maximized: H i (th) represents the Kapur entropy in the i-th threshold interval when the threshold is th; To extend the multi-level threshold expression, Kapur’s objective function is defined as: f Kapur (TH)=f Kapur (th i ),i=1,2,3,…,k; Among them, different thresholds are represented by TH[th1, th2, th3, ..., thk-1], and i corresponds to a specific threshold interval.

7. The paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image according to claim 6, characterized in that: The particle swarm optimization algorithm is used to optimize the running speed of Kapur multi-threshold segmentation and obtain values ​​close to the optimal solution, including: Generate initial position and initial velocity; Evaluate the objective function for each particle; Update the single optimal solution and the global optimal solution, the single optimal solution is the single optimal particle, and the global optimal solution is the global optimal particle Update the velocity v and position x of each particle; Repeat the particle swarm optimization process until the termination condition is reached; In the particle swarm optimization algorithm, the particle velocity update formula is as follows: v j (k+1)=wv j (k)+c1r1(p j (k)-x j (k))+c2r2(g(k)-x j (k)),j=1,2,…,np. In the particle swarm optimization algorithm, the particle position update formula is as follows: x j (k+1)=x j (k)+v j (k); Among them, v j (k) and x j (k) are the velocity and position of the jth particle at the kth iteration, respectively, and p j g(k) and g(k) are the single best particle and the global best particle at the kth iteration, respectively; w is the initial weight parameter, c1 is the individual cognitive parameter, and c2 is the social cognitive parameter; r1 and r2 are random numbers between 0 and 1 with uniform distribution.

8. A paraspinal muscle fat segmentation device based on magnetic resonance T2 axial image, characterized in that: include: An image data acquisition module, used to acquire image data, wherein the image data is a T2 axial sequence lumbar spine image of the target object acquired by a magnetic resonance imaging device in a hospital; The labeling information acquisition module is used to acquire the labeling information of the paraspinal muscle area and obtain the image data with the paraspinal muscle area labeled; An image preprocessing module is used to perform image preprocessing on the image data marked with the paraspinal muscle area, perform image enhancement on each paraspinal muscle area, and obtain an enhanced image of the paraspinal muscle area; A segmentation module, used for performing image segmentation on the paraspinal muscle region enhanced image based on a PSO-Kapur threshold segmentation algorithm to determine the muscle fat segmentation result of each paraspinal muscle; The output module is used to superimpose the muscle fat segmentation results of each paraspinal muscle on the T2 axial sequence lumbar spine image and output a visual segmentation result map.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the paraspinal muscle fat segmentation method based on magnetic resonance T2 axial image according to any one of claims 1 to 7 are implemented.

Images