Sacroiliac arthritis automatic report generation method and system based on magnetic resonance image

Abstract

The invention provides an automatic report generation method and system for sacroiliac arthritis based on a magnetic resonance image. The method comprises the following steps: accurately recording the age, the disease course duration and the quantification level of waist and back pain of a patient; the magnetic resonance imaging technology is comprehensively applied; the method comprises the following steps of: manually defining and taking an interested area confirmed by an intersection through a plurality of imaging department doctors, and constructing a comprehensive and meticulous classification feature set in combination with an image evaluation grade, an active inflammation degree and laboratory features; a logistic regression model is adopted as a classification tool to ensure the stability and accuracy of a judgment result; and the system automatically generates a structured judgment report based on a case illness probability value calculated based on a Sigmoid function. Based on the technical scheme, a structured report conforming to clinical specifications and standard description is formed, knowledge points, key images, structures and the like displayed on the report are clear at a glance, the readability and comparability of the report are improved, and information capture and data analysis are facilitated.

Description

Technical Field

[0001] The present invention relates to the field of medical imaging technology, and in particular to a method and system for automatically generating a sacroiliitis report based on magnetic resonance imaging. Background Art

[0002] Sacroiliitis is a disease characterized by inflammation of the sacroiliac joints, which is usually associated with inflammatory diseases of the spine (such as ankylosing spondylitis and arthritis associated with inflammatory bowel diseases). The main symptoms of sacroiliitis include pain in the lower back or buttocks, which may be unilateral or bilateral. The pain often worsens after sitting or walking and is slightly relieved when resting. Sacroiliitis usually occurs in young adults, with most patients onset between the ages of 20 and 40, and in general, the prevalence in men is significantly higher than in women. The exact incidence of sacroiliitis varies by region and ethnic group, but its prevalence shows a clear ethnic and genetic tendency, especially with a close association with specific HLA genes (such as HLA-B27).

[0003] At present, imaging diagnosis of sacroiliitis mainly includes X-ray examination and magnetic resonance imaging (MRI). X-ray examination can show non-specific changes such as bone hyperplasia of the sacroiliac joint, blurring of the joint surface, and narrowing of the joint space. MRI is more sensitive to early lesions or inflammatory activity and can show inflammation and edema of the sacroiliac joint soft tissue.

[0004] However, there is still obvious inconsistency in the judgment of sacroiliitis. First, the lack of unified imaging standards may lead to differences in the interpretation of MRI images by different medical institutions and imaging doctors. Secondly, the performance of different MRI equipment, magnetic field strength and sequence parameters used will affect the quality of the image and the anatomical structure displayed, thereby affecting the consistency of the imaging results. In addition, individual anatomical variations and different manifestations of sacroiliitis at different stages of the disease are also important reasons for the inconsistency of judgment standards. Although MRI has a high sensitivity for early sacroiliitis and can show characteristics such as bone marrow edema and soft tissue inflammation, in the judgment of late sacroiliitis, MRI may show more typical manifestations such as bone hyperplasia and joint fusion. Therefore, in clinical practice, there is an urgent need for a consistent and reliable imaging judgment process and standards to promote early intervention for patients.

[0005] In addition, there are certain problems with the current writing mode of imaging judgment reports, especially the imaging manifestations and judgment conclusions in the reports, which still use descriptive language and rely on the subjective judgment and experience of doctors. Although there are certain standard requirements for the writing of reports in the industry, different doctors still use different description habits and terms when analyzing the same case. From the perspective of information technology, this kind of report belongs to unstructured data. Computers can only store imaging manifestations and judgment conclusions in the form of free text, resulting in information retrieval can only be done through keyword combinations. Due to the subjectivity of this method, it leads to low recall and precision, too much irrelevant information returned by retrieval, and the inability to reflect the upper and lower relationships between word meanings and the overall and local relationships. From the perspective of imaging examination, there are multiple purposes such as judgment, follow-up, preoperative and postoperative evaluation. The focus of clinical departments on imaging reports is not limited to judgment conclusions, but more and more attention is paid to the expression of disease manifestations and their related secondary manifestations, which requires the content reflected in imaging reports to be more substantial and objective. Currently, most imaging reports of sacroiliitis are handwritten reports in plain text, but the writing process is time-consuming and labor-intensive, there is no automated judgment, lacks quantitative analysis and information integration, the image description lacks intuitive display, and there are differences in report content and terminology, which affects the comprehensiveness, professionalism and mutual recognition of the imaging report, and limits the role of imaging reports in the diagnosis and treatment pathway. Therefore, whether it is from the construction of big data or individualized diagnosis and treatment, it is necessary to improve the form and content of the current imaging reports. Therefore, the establishment of a structured imaging reporting system for sacroiliitis is imminent. The present invention proposes an intelligent judgment and automated reporting system for sacroiliitis based on magnetic resonance imaging. By comprehensively and comprehensively characterizing the imaging and clinical characteristics of sacroiliitis, a set of intelligent magnetic resonance imaging-based sacroiliitis automated report generation methods and systems are formed, which provides a new perspective for establishing a comprehensive, professional and interoperable sacroiliitis judgment standard.

[0006] After searching, there is no sacroiliitis MRI structured report at home and abroad. The invention patent that is similar to this structured report is: "Automated Reporting System and Method for Cognitive Impairment Imaging" (patent application number: CN202010080771.1). Summary of the invention

[0007] In view of the above problems in the prior art, this application proposes a method for automatically generating sacroiliitis reports based on magnetic resonance imaging, which integrates a comprehensive clinical and imaging data collection process to improve judgment efficiency and accuracy. The specific collection steps include:

[0008] Clinical information integration: Accurately record the patient's age, duration of illness, and quantitative level of low back pain to provide basic background information for judgment.

[0009] Multimodal imaging analysis: four advanced magnetic resonance imaging techniques: coronal and transverse T1-weighted imaging, coronal and transverse fat-suppressed T2-weighted imaging.

[0010] Oblique coronal T1-weighted images were obtained with optimized scanning parameters (TR between 300 and 600 ms, TE fixed at 10 ms, echo train length of 3 to 4, and slice thickness set at 4 mm) to clearly display the tissue structure.

[0011] Cross-sectional fat-suppressed T2-weighted imaging effectively suppresses fat signals and enhances soft tissue contrast by extending TR to more than 2000ms and adjusting TE within the range of 60-80ms. The layer thickness is also set to 4mm.

[0012] For cross-sectional T1-weighted imaging, a short TR / TE combination of TR less than 600 ms and TE less than 20 ms was used, the echo train length was set to 3, and the layer thickness was kept consistent to highlight the T1 characteristics of the tissue.

[0013] Laboratory data integration: including key biomarkers such as HLA-B27 genotype, C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR), providing important basis for inflammation assessment.

[0014] Feature extraction and classification: Experienced radiologists manually define and confirm the region of interest (ROI) by taking the intersection, and extract signal values ​​from the above four imaging modalities as key classification features. At the same time, combined with the imaging assessment level (covering bone erosion, fatty degeneration, bone sclerosis, osteophyte, bone ankylosis, joint space changes, etc.), active inflammation score and laboratory characteristics, a comprehensive and detailed classification feature set is constructed.

[0015] Intelligent model application: Using the logistic regression model as the classification tool, the data set is first randomly and evenly divided into 10 subsets. The first subset is selected as the test set and the remaining 9 subsets are used as training sets. The model is trained and its performance is evaluated on the test set. The second subset is selected as the test set. The model is retrained using the remaining 9 subsets and the performance is evaluated on the new test set. This process is continued until each subset is used as a test set. In each iteration, the performance indicators of the model on the test set are obtained. Finally, the average of these performance indicators is calculated to screen out the optimal model parameters to ensure the stability and accuracy of the judgment results.

[0016] Report generation: Finally, based on the case disease probability value (between 0 and 1) calculated by the Sigmoid function, the system automatically generates a structured judgment report, intuitively presenting the patient's sacroiliitis risk assessment and providing strong support for clinical decision-making.

[0017] In one embodiment, factors affecting the imaging assessment grade include bone erosion, fatty degeneration, bone sclerosis, osteophytes, ankylosing bone, and joint space changes.

[0018] In one embodiment, factors affecting the assessment of the degree of active inflammation include bone marrow edema, osteitis, synovitis, synovial bursitis, and enthesitis.

[0019] In one embodiment, the method for evaluating bone marrow edema includes: evaluating bone marrow edema images separately from the left and right sides, selecting 6 levels, and evaluating the same level on each side; only evaluating STIR sequence images, and using T1WI images as anatomical references; on each side, on each level, evaluating 4 quadrants, and recording the number of quadrants with bone marrow edema; on each side, on each level, evaluating the depth and intensity of the bone edema area: recording the number of homogeneous and clear high signals extending ≥1 cm inward from the cartilage surface; recording the number of bright signal intensities that reach or exceed the presacral venous blood flow signals.

[0020] In one embodiment, each layer has a corresponding drop-down box, and bone marrow edema assessment adopts the mode of navigation diagram. When a layer is selected, the user can click once in the quadrant of the corresponding lesion in the navigation diagram on the right, and the layer will be recorded once, and the recorded value will be automatically filled in the scoring text box; the doctor manually selects the depth and enhancement assessment; after completing the assessment of 6 layers in turn, the system automatically calculates the total recorded values ​​for the left side, right side, and both sides.

[0021] The present invention also relates to a sacroiliitis automated report generation system based on magnetic resonance imaging, which adopts any of the above-mentioned sacroiliitis automated report generation methods based on magnetic resonance imaging.

[0022] The above technical features can be combined in various suitable ways or replaced by equivalent technical features as long as the purpose of the present invention can be achieved.

[0023] The method and system for automatically generating a sacroiliitis report based on magnetic resonance imaging provided by the present invention have at least the following beneficial effects compared with the prior art:

[0024] 1. For the first time, a structured imaging report was constructed for sacroiliitis MRI examination.

[0025] 2. Integrate the MRI active inflammation assessment of sacroiliitis into the structured report to vividly display the location and degree of bone marrow edema and realize automatic assessment.

[0026] 3. Create a structured report that complies with clinical norms and standard descriptions. The knowledge points, key images, structures, etc. displayed on the report are clear at a glance, which improves the readability and comparability of the report and facilitates information capture and data analysis. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The present invention will be described in more detail below based on embodiments and with reference to the accompanying drawings, wherein:

[0028] Figure 1 shows the automated report generation flow chart of the present invention;

[0029] Figure 2 The structured report interface diagram of sacroiliitis MRI of the present invention is shown;

[0030] Figure 3 A partial schematic diagram of the sacroiliitis MRI structured report of the present invention is shown;

[0031] Figure 4 A schematic diagram of an imaging report of the present invention is shown. DETAILED DESCRIPTION

[0032] The present invention will be further described below in conjunction with the accompanying drawings.

[0033] The present invention provides a method for generating an automated report of sacroiliitis based on magnetic resonance imaging, which integrates a comprehensive clinical and imaging data collection process to improve judgment efficiency and accuracy. The specific collection steps include:

[0034] Clinical information integration: Accurately record the patient's age, duration of illness, and quantitative level of low back pain to provide basic background information for judgment.

[0035] Multimodal imaging analysis: four advanced magnetic resonance imaging techniques: coronal and transverse T1-weighted imaging, coronal and transverse fat-suppressed T2-weighted imaging.

[0036] Oblique coronal T1-weighted images were obtained with optimized scanning parameters (TR between 300 and 600 ms, TE fixed at 10 ms, echo train length of 3 to 4, and slice thickness set at 4 mm) to clearly display the tissue structure.

[0037] Cross-sectional fat-suppressed T2-weighted imaging effectively suppresses fat signals and enhances soft tissue contrast by extending TR to more than 2000ms and adjusting TE within the range of 60-80ms. The layer thickness is also set to 4mm.

[0038] For cross-sectional T1-weighted imaging, a short TR / TE combination of TR less than 600 ms and TE less than 20 ms was used, the echo train length was set to 3, and the layer thickness was kept consistent to highlight the T1 characteristics of the tissue.

[0039] Laboratory data integration: including key biomarkers such as HLA-B27 genotype, C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR), providing important basis for inflammation assessment.

[0040] Feature extraction and classification: The region of interest (ROI) was manually defined and confirmed by multiple experienced radiologists, and the signal values ​​were extracted from the above four imaging modalities as key classification features. At the same time, a comprehensive and detailed classification feature set was constructed by combining the imaging assessment level (covering bone erosion, fatty degeneration, osteosclerosis, osteophytes, ankylosing, joint space changes, etc.), active inflammation score and laboratory characteristics.

[0041] Intelligent model application: Using the logistic regression model as the classification tool, the data set is first randomly and evenly divided into 10 subsets. The first subset is selected as the test set and the remaining 9 subsets are used as training sets. The model is trained and its performance is evaluated on the test set. The second subset is selected as the test set. The model is retrained using the remaining 9 subsets and the performance is evaluated on the new test set. This process is continued until each subset is used as a test set. In each iteration, the performance indicators of the model on the test set are obtained. Finally, the average of these performance indicators is calculated to screen out the optimal model parameters to ensure the stability and accuracy of the judgment results.

[0042] Report generation: Finally, based on the case disease probability value (between 0 and 1) calculated by the Sigmoid function, the system automatically generates a structured judgment report, intuitively presenting the patient's sacroiliitis risk assessment and providing strong support for clinical decision-making.

[0043] In one embodiment, Figure 1 As shown, the method for generating an automated report of sacroiliitis based on magnetic resonance imaging includes the following steps:

[0044] Step 1: Recruitment of subjects: A group of participants will undergo MRI scans related to sacroiliitis. All scans are performed at designated medical centers, and the specific schedule will be determined based on the availability and coordination of the participants. The personal information and privacy of all subjects will be strictly protected. In order to make the system stable, reliable, and generalizable, more than 1,000 patients with sacroiliitis and healthy controls need to be recruited.

[0045] Step 2: Clinical data collection. All recruited subjects were interviewed to collect their age, course of disease, and back pain severity. The back pain severity was divided into 10 levels, with 1 indicating the mildest symptom and 10 indicating severe pain.

[0046] Step 3: Imaging examination. In order to fully obtain the characteristic information of patients with sacroiliitis, the invention uses four modal imaging data to characterize it, including oblique coronal and transverse T1-weighted imaging, oblique coronal and transverse fat-suppressed T2-weighted imaging; the scanning parameters of oblique coronal T1-weighted imaging are TR<300-600ms, TE=10ms, echo chain=3-4, and layer thickness is 4mm; the scanning parameters of transverse fat-suppressed T2-weighted imaging are TR>2000ms, TE=60-80ms, and layer thickness is 4mm; the scanning parameters of transverse T1-weighted imaging are TR<600ms, TE<20ms, echo chain=3, and layer thickness is 4mm.

[0047] Step 4: Laboratory data collection. In order to obtain more comprehensive biochemical information of patients with sacroiliitis, the invention collects three laboratory indicators. 1) HLA-B27: This is a human leukocyte antigen that is often used to detect genetic factors associated with specific types of arthritis, especially sacroiliitis and other related rheumatoid diseases. People with HLA-B27 positive are at a higher risk of developing sacroiliitis than those with HLA-B27 negative. 2) CRP: C-reactive protein, a protein produced by the liver, usually increases when the body suffers from inflammation or tissue damage. CRP measurement is often used to assess the degree of inflammation and its changes in various diseases (such as infection, rheumatic diseases, etc.). 3) ESR: Erythrocyte sedimentation rate. ESR measurement is a method of evaluating nonspecific inflammatory responses by detecting the speed at which red blood cells settle to the bottom of a blood tube within a certain period of time. It can be used as an indicator of the degree of inflammation and infection, but cannot be directly used to judge specific diseases. The three indicators of HLA-B27, CRP, and ESR are all measured using blood samples collected in the vein. HLA-B27 uses polymerase chain reaction (PCR) technology to amplify the HLA-B27 gene region, and the PCR amplification products are sequenced or tested for specific genotypes to determine whether the HLA-B27 gene exists. CRP is measured using an immunoassay, and the CRP concentration in the test blood is analyzed and recorded.

[0048] Step 5: Feature extraction and analysis. In the invention, the features used for classification include the values ​​of two modal signals of the image data in the sacroiliac joint area, the grade characteristics of the image evaluation (bone erosion, fatty change, osteosclerosis, osteophytes, ankylosing, joint space), active inflammation score (bone edema, osteitis, synovitis, synovial bursitis) and laboratory characteristics (HLA-B27, CRP, ESR). Among them, the sacroiliac joint area is outlined by three experienced radiologists, and their intersection is taken. The six types of bone erosion, fatty change, osteosclerosis, osteophytes, ankylosing, and joint space are divided into 0-3 levels by image evaluation, where 0 means that this phenomenon does not occur on both sides; 1 means that it occurs on the left side; 2 means that it occurs on the right side; 3 means that it occurs on both sides. The scoring rules for bone marrow edema are: a) the left and right sides are evaluated separately; b) each side evaluates the same 6 levels, usually the 4th to 9th levels; only STIR sequence images are evaluated, and T1WI images are used as anatomical references. c) On each side, at each level, evaluate 4 quadrants. If there is bone marrow edema in one quadrant, 1 point is given. d) On each side, at each level, evaluate the depth and intensity of the bone marrow edema area: as long as there is a homogeneous and clear high signal extending ≥1cm from the cartilage surface, 1 point is given; if the intensity of the bright signal reaches or exceeds the presacral venous blood flow signal, 1 point is given; e) On each side of the sacroiliac joint, a maximum of 6 points are given at the same level, and the total score of 6 levels is 36 points. The total score for both sides is 72 points. The score for active inflammation (osteitis, synovitis, synovial bursitis) is 0-3, where no mark is 0, left mark is 1, right mark is 2, and bilateral mark is 3. The score of HLA-B27 is binary, 0 represents HLA-B27 negative, and 1 represents HLA-B27 positive. CRP is a continuous value in mg / L. ESR is also a continuous value in mm / h.

[0049] Step six: model training and optimization. The present invention incorporates 16 dimensions of features for model training (the values ​​of sacroiliac joint area signals in two modal images, bone erosion, fatty degeneration, bone sclerosis, osteophytes, ankylosis, joint space, bone edema, osteitis, synovitis, synovial bursitis, HLA-B27, CRP, ESR), and the feature is recorded as X (where the dimension of X is 16). During the training process, sacroiliitis patients and healthy control subjects are divided into training set, validation set, and test set in a ratio of 7:2:1. The classification model uses Logistic regression. Specifically, the samples of sacroiliitis patients are marked as 1, and the samples of healthy controls are marked as 0. Specifically, in the logistic regression model,

[0050]

[0051]

[0052] Among them, x is the input feature of each subject, y is the label to be predicted, w and b are the parameters for linearizing the features, which can be obtained using the maximum likelihood probability.

[0053]

[0054] 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.

[0055] make

[0056] z=w T x+b

[0057] Then, you can

[0058]

[0059] To calculate the probability that the subject is a sacroiliitis patient or a healthy control. If y>0.5, the system determines that the subject is a sacroiliitis patient; if y<0.5, the system determines that the subject is a healthy person.

[0060] Step 7: Judgment support and decision recommendation. According to the y value obtained in step 6, determine the probability of whether the patient has sacroiliitis. For patients with y>0.5, follow up their disease course in depth and formulate individualized treatment strategies according to the severity of different indicators. For patients with y close to 0.5, long-term follow-up is carried out, and targeted strategies for improving low back pain symptoms are proposed, especially for patients with positive HLA-B27.

[0061] Step 8: Output a structured report, which presents clinical results, imaging results, and recommendations.

[0062] In one embodiment, Figure 2-4 As shown in Figure 1, the structured judgment report includes clinical assessment, technical assessment, and imaging assessment. The imaging assessment includes structural assessment of bone erosion, fat, bone sclerosis, osteophytes, bone ankylosis, and joint space, and assessment of active inflammation (bone marrow edema, osteitis, synovitis, synovial bursitis, and enthesitis). The assessment method for bone marrow edema includes: evaluating bone marrow edema images separately from the left and right sides, selecting 6 levels, and evaluating the same level on each side; only evaluating STIR sequence images, and T1WI images as anatomical reference; evaluating 4 quadrants on each side and each level, and recording the number of quadrants with bone marrow edema; evaluating the depth and intensity of bone edema areas on each side and each level: recording the number of homogeneous and clear high signals extending ≥1 cm from the cartilage surface; recording the number of bright signal intensity reaching or exceeding the presacral venous blood flow signal.

[0063] In one embodiment, each layer has a corresponding drop-down box. The bone marrow edema assessment adopts the mode of navigation diagram. When a layer is selected, click once in the quadrant of the corresponding lesion in the navigation diagram on the right, and the layer is recorded once, and the recorded value is automatically filled in the scoring text box; the doctor manually selects the depth and enhancement assessment; the six layers are evaluated in sequence, and the system automatically calculates the total recorded values ​​of the left side, right side and both sides. The blank box of "active inflammation" on the right is used to implant key image screenshots in the active inflammation assessment.

[0064] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in other described embodiments.

Claims

1. A method for generating an automated report of sacroiliitis based on magnetic resonance imaging, characterized in that: The following steps are involved: Accurately record the patient's age, duration of illness, and quantitative level of back pain; Magnetic resonance imaging of the sacroiliac joint was performed using a combination of coronal oblique T1-weighted imaging, transverse fat-suppressed T2-weighted imaging, and transverse T1-weighted imaging. HLA-B27 genotype, C-reactive protein, and erythrocyte sedimentation rate were used as markers; Through the regions of interest that were manually defined and confirmed by multiple radiologists, signal values ​​were extracted from four imaging modalities, including oblique coronal and transverse T1-weighted imaging, and oblique coronal and transverse fat-suppressed T2-weighted imaging, as key classification features. At the same time, a comprehensive and detailed classification feature set was constructed by combining the imaging assessment grade, degree of active inflammation, and laboratory characteristics. Using the logistic regression model as a classification tool, the data set is first randomly and evenly divided into 10 subsets. The first subset is selected as the test set, and the remaining 9 subsets are used as training sets. The model is trained and its performance is evaluated on the test set. The second subset is selected as the test set, and the remaining 9 subsets are used to retrain the model and evaluate the performance on the new test set. This process is continued until each subset is used as a test set. In each iteration, the performance indicators of the model on the test set are obtained. Finally, the average of these performance indicators is calculated to screen out the optimal model parameters to ensure the stability and accuracy of the judgment results. Finally, based on the case disease probability value calculated by the Sigmoid function, the system automatically generates a structured judgment report.

2. The method for generating an automated report of sacroiliitis based on magnetic resonance imaging according to claim 1, characterized in that: For oblique coronal T1-weighted images, the scanning parameters TR was between 300-600ms, TE was fixed at 10ms, echo train length was 3-4, and layer thickness was set to 4mm.

3. The method for generating an automated report of sacroiliitis based on magnetic resonance imaging according to claim 1, characterized in that: Cross-sectional fat-suppressed T2-weighted imaging effectively suppresses fat signals and enhances soft tissue contrast by extending TR to more than 2000ms and adjusting TE within the range of 60-80ms. The layer thickness is also set to 4mm.

4. The method for generating an automated report of sacroiliitis based on magnetic resonance imaging according to claim 1, characterized in that: For cross-sectional T1-weighted imaging, a short TR / TE combination of TR less than 600 ms and TE less than 20 ms was used, the echo train length was set to 3, and the layer thickness was kept consistent to highlight the T1 characteristics of the tissue.

5. The method and system for generating an automated report of sacroiliitis based on magnetic resonance imaging according to claim 1, characterized in that: Factors that affect the grade of imaging assessment include bone erosion, fatty degeneration, bone sclerosis, osteophytes, ankylosis, and joint space changes.

6. The method for generating an automated report of sacroiliitis based on magnetic resonance imaging according to claim 1, characterized in that: Factors that affect the assessment of the degree of active inflammation include bone marrow edema, osteitis, synovitis, bursitis, and enthesitis.

7. The method for generating an automated report of sacroiliitis based on magnetic resonance imaging according to claim 6, characterized in that: The evaluation method for bone marrow edema includes: evaluating bone marrow edema images separately from the left and right sides, selecting 6 levels, and evaluating the same level on each side; only evaluating the oblique coronal T2 fat-suppressed images, and using T1WI images as anatomical references; evaluating 4 quadrants on each side and at each level, and recording the number of quadrants with bone marrow edema; evaluating the depth and intensity of the bone edema area on each side and at each level: recording the number of homogeneous and clear high signals extending ≥1 cm from the cartilage surface inward; recording the number of bright signal intensities that reach or exceed the presacral venous blood flow signals.

8. The method for generating an automated report of sacroiliitis based on magnetic resonance imaging according to claim 7, characterized in that: Each level has a corresponding drop-down box. The bone marrow edema assessment adopts the mode of navigation diagram. Select a level and click once in the quadrant of the corresponding lesion in the navigation diagram on the right. The level will be recorded once and the recorded value will be automatically filled in the scoring text box. The doctor manually selects the depth and enhancement assessment. After completing the assessment of 6 levels in turn, the system automatically calculates the total recorded values ​​for the left, right and bilateral sides.

9. An automated report generation system for sacroiliitis based on magnetic resonance imaging, characterized in that: The method for generating an automated report of sacroiliitis based on magnetic resonance imaging as described in any one of claims 1 to 8 is adopted.

Images