Middle-axis spinal arthritis image processing method based on magnetic resonance image
Through the magnetic resonance image-based image processing methods, including signal standardization and the use of deep learning networks, the inconsistency and signal overlap of the judgment of bone marrow edema in patients with sacroiliac arthritis is solved, and more accurate early judgment is achieved.
Patent Information
- Application Number
- CN202411374871.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-29
- Publication Date
- 2025-02-11
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art has problems of inconsistency and indistinguishable signal overlap when interpreting sacroiliac bone marrow edema in patients with central axial spondylitis, especially when the signal changes in the early stages of the disease are weak, it is difficult to accurately judge.
Using a magnetic resonance image-based image processing method, by acquiring healthy controls and patients' magnetic resonance images, establishing areas of interest, performing signal standardization and Gaussian noise addition, training using ResNet network, combining 50% cross-validation and gradient-weighted class activation Mapping to improve the accuracy of judgment.
It improves the accuracy of judging bone marrow edema in patients with central axial spondylitis, reduces the inconsistency of judgment results, especially in the early stages of the disease, which can more accurately identify weak signal changes.
Smart Images
Figure CN119417758A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of magnetic resonance image processing, and proposes an axial spondyloarthritis image processing method based on magnetic resonance images for axial spondyloarthritis. Background Art
[0002] Axial spondyloarthritis is a global chronic health challenge that is receiving increasing attention, and its prevalence is increasing worldwide. Once patients suffer from axial spondyloarthritis, they will face long-term symptoms such as joint pain, swelling and stiffness, which not only limit patients' daily activities, but also seriously affect their quality of life. What's more, persistent pain and discomfort may cause patients' mental health problems, such as anxiety and depression, further aggravating the negative impact of the disease. In response to the global health problem of axial spondyloarthritis, we need to strengthen disease prevention, early diagnosis and effective treatment. By raising public health awareness, promoting the rational allocation of medical resources, and encouraging the innovation and development of medical technology, we can work together to provide patients with better medical services, alleviate their pain, improve their quality of life, and also reduce the burden on patients' families.
[0003] However, the interpretation of sacroiliac joint bone marrow edema requires rich professional knowledge and experience. Different doctors or technicians may have different interpretations of the same set of images, which may lead to inconsistent judgment results. In addition, the signal of bone marrow edema may overlap with the signal of surrounding normal tissue or other pathological changes (such as tumors, inflammation, etc.), increasing the difficulty of accurate distinction. In the early stages of the disease, bone marrow edema may be relatively mild, and may only appear as weak signal changes on magnetic resonance images, which is difficult to distinguish from normal tissue, thereby increasing the difficulty of judgment. With the continuous development of deep learning, intelligent judgment technology is expected to provide a new means for the judgment of sacroiliac joint bone marrow edema in patients with axial spondyloarthritis. Summary of the invention
[0004] The present invention aims to solve the problems described in the background art and provides an image processing method for axial spondyloarthritis based on magnetic resonance imaging.
[0005] In order to achieve the above object, the present invention adopts the following technical scheme:
[0006] Step 1: Obtain magnetic resonance images of sacroiliac joint bone marrow edema in a group of healthy controls (HC) and a group of patients with axial spondyloarthritis; for the data collection method of the subjects, it is required to use a 3T scanner for data collection, collect T2-weighted images, with TR of 544ms, TE of 11ms, flip angle of 140°, bandwidth of 326Hz, field of view of 250×250mm2, layer thickness of 2mm, layer spacing of 0.2mm, scanning matrix of 448×448, number of layers of 37, and fat suppression scanning method.
[0007] Step 2: Establish the region of interest and exclude the areas other than the sacrum and ilium. When establishing the region of interest, three experienced radiologists performed image segmentation on all subjects, retaining the areas where the sacrum, ilium and joints are located, and took the union of the three doctors' segmentation on the entire data set.
[0008] Step 3: Standardize the region of interest. Standardization includes setting the region of interest in each layer of each subject and extracting the signal of the region of interest. After the signal of each pixel in the region of interest is extracted, the logarithm is taken and multiplied by 10. 4 To improve the effectiveness of calculation. In addition, in order to remove the influence of signal acquisition between different subjects, the invention proposes a new magnetic resonance signal standardization method, the specific standardization method is as follows:
[0009]
[0010] Where M represents the signal of voxel point i of the magnetic resonance image, min represents the minimum value, α% represents the first percent of the signal, which is 50% here, and N represents the number of all voxel points. is the signal mean representing the fat area, is the mean signal value of the spine area.
[0011] Step 4: Superimpose Gaussian random noise on the region of interest. The noise follows a Gaussian distribution with a mean of 0 and a variance of 0.3.
[0012] Step 5: Mark the images of patients with bone marrow edema as 1 and the images of healthy subjects as 0;
[0013] Step 6: Divide the data sets of all subjects into training set and test set according to the ratio of 7:3, and use 5-fold cross validation to evaluate the performance of the model;
[0014] Step 7: Use the ResNet network to train the model. The model loss function uses the cross entropy loss function. ResNet contains five convolution stages. The first stage contains 64 7×7 convolution kernels, and the second to fifth convolution stages contain 3×3 convolution blocks. The model structure is shown in the attached figure.Figure 5 As shown. The fully connected layer consists of a 512-dimensional vector, and the output layer is determined by the Sigmoid function. Specifically, the Sigmoid function is determined by the following probabilities,
[0015]
[0016] Among them, x is the 512-dimensional feature of each subject input, and y is the label to be predicted. When the probability is greater than 0.5, the predicted label is 1, and when the probability is less than 0.5, the predicted label is 0. w and b are the parameters for linearizing the features, which can be obtained using the maximum likelihood probability.
[0017]
[0018] That is, the greater the probability that each sample belongs to its true label, the better. According to convex optimization theory, using the gradient descent method, w and b can be solved.
[0019] Let γ=(w;b), Then T x+b can be abbreviated as
[0020] Re-order
[0021]
[0022] but
[0023]
[0024] is equivalent to,
[0025]
[0026] In order to obtain γ,
[0027]
[0028] The accuracy, precision, recall, specificity and score A were used to evaluate the model classification effect; the specific calculation method is as follows:
[0029]
[0030]
[0031] Step 8: Use gradient-weighted class activation mapping to explore the areas that contribute most to the classification. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which:
[0033] Figure 1 This is a flow chart of the intelligent method for determining bone marrow edema of sacroiliac joints in patients with axial spondyloarthritis based on magnetic resonance imaging of the present invention;
[0034] Figure 2 Schematic diagram of the sacroiliac joint used in the present invention; (A) is a schematic diagram of the entire joint, (B) is a schematic diagram of the sacrum (green) and ilium (red) and bone marrow edema (yellow); (C) is a schematic diagram of the region of interest, which extracts the image signals of the blue areas on both sides of the subject; (D) is a schematic diagram of the region of interest after standardization and with random noise added.
[0035] Figure 3 Schematic diagram of the intelligent model proposed in the present invention; (A) is a schematic diagram of image standardization and adding Gaussian noise; (B) is a schematic diagram of the deep learning judgment model for bone marrow edema.
[0036] Figure 4 This is a schematic diagram of the ResNet network module;
[0037] Figure 5 is the receiver operating curve using 5-fold cross validation on the example data;
[0038] Figure 6 The contribution of gradient-weighted class activation mapping to classification. DETAILED DESCRIPTION
[0039] In order to make the purpose and technical implementation of the present invention clearer, embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present invention.
[0040] In the description of the present invention, unless otherwise specified and limited, it should be noted that terms such as "signal", "image", "framework", "weight", "connection", "model", "training set", "validation set" and "test set" should be understood in a broad sense, and those of ordinary skill in the art can understand the specific meanings of the above terms according to specific circumstances.
[0041] like Figure 1-4As shown, the present application provides an axial spondyloarthritis image processing method based on magnetic resonance imaging, and is tested on a 815-layer dataset consisting of sacroiliac joints from 79 subjects (60 of whom are patients with axial spondyloarthritis and accompanied by sacroiliac joint bone marrow edema, and 19 of whom are healthy controls);
[0042] The test on the data includes the following steps:
[0043] Step 1: Magnetic resonance images of sacroiliac joint bone marrow edema were obtained from 19 healthy controls (HC) and a group of patients with axial spondyloarthritis.
[0044] Step 2: Establish the region of interest and exclude the area other than the sacrum and ilium;
[0045] Step 3: Standardize the region of interest. In this test, the number of image layers including the sacrum and ilium of the subjects ranges from 18 to 25, so 23 layers are selected as the final input image for each subject in this test.
[0046] Step 4: Superimpose Gaussian random noise on the region of interest; the noise follows a Gaussian distribution with a mean of 0 and a variance of 0.3.
[0047] Step 5: Mark the images of patients with bone marrow edema as 1 and the images of healthy subjects as 0;
[0048] Step 6: Divide the data sets of all subjects into training set and test set according to the ratio of 7:3, and use 5-fold cross validation to evaluate the performance of the model;
[0049] Step 7: Use the ResNet network to train the model; use accuracy, precision, recall, specificity, F1 score and ROC curve to evaluate the performance of the model. The classification results for each layer are as follows:
[0050] The results of five-fold cross validation are as follows Figure 5 shown.
[0051] The classification results for each subject are as follows:
[0052]
[0053] Step 8: Use gradient weighted class activation mapping to explore the area that contributes most to classification. The result is shown in the figure below. Figure 6 As shown, the redder the color, the greater the contribution to the classification.
Claims
1. A method for processing axial spondyloarthritis images based on magnetic resonance imaging, characterized in that: The steps include: Step 1: Obtain magnetic resonance images of sacroiliac joint bone marrow edema in a group of healthy controls and a group of patients with axial spondyloarthritis; Step 2: Establish the region of interest and exclude the area other than the sacrum and ilium; Step 3: Standardize the region of interest; Step 4: Superimpose Gaussian random noise on the region of interest; Step 5: Mark the images of patients with bone marrow edema as 1 and the images of healthy subjects as 0; Step 6: Divide the data sets of all subjects into training set and test set according to the ratio of 7:3, and use 5-fold cross validation to evaluate the performance of the model; Step 7: Use the deep learning classification model ResNet network to train the model; Step 8: Use gradient-weighted class activation mapping to explore the areas that contribute most to the classification.
2. The method for processing axial spondyloarthritis images based on magnetic resonance imaging according to claim 1, characterized in that: In step 1, the data collection method of the subjects required the use of a 3T scanner to collect T2-weighted images, with a TR of 544ms, a TE of 11ms, a flip angle of 140°, a bandwidth of 326Hz, and a field of view of 250×250mm. 2 The layer thickness is 2 mm, the layer spacing is 0.2 mm, the scanning matrix is 448×448, the number of layers is 37, and the fat suppression scanning method is used.
3. The method for processing axial spondyloarthritis images based on magnetic resonance imaging according to claim 2, characterized in that: In step 3, standardization includes setting the region of interest in each layer of each subject and extracting the signal of the region of interest; after the signal of each pixel in the region of interest is extracted, the logarithm is taken and multiplied by 10 4 .
4. The method for processing axial spondyloarthritis images based on magnetic resonance imaging according to claim 3, characterized in that: The construction of the deep learning classification model involves using a median filter after inputting each layer of images of each subject into ResNet.
5. The method for processing axial spondyloarthritis images based on magnetic resonance imaging according to claim 4, characterized in that: In step 7, the evaluation of the model classification effect uses accuracy, precision, recall, specificity and score A; the specific calculation method is as follows: