Recognition device based on ultrasonic image

Abstract

The embodiments of the present invention relate to the field of information processing, and disclose a recognition device based on ultrasonic images. It includes: an acquisition module, which is used to acquire ultrasonic images of different parts of the user; a recognition module, which is used to use a preset set of N training models to identify each characteristic symptom existing in the ultrasonic image; a scoring module, which is used to obtain the user's scoring result based on the characteristic symptoms existing in the ultrasonic image, and the preset correspondence between each characteristic symptom and the score value; an output module, which is used to output the user's test result based on the scoring result and the preset score threshold. In an embodiment of the present invention, by collecting ultrasonic images of various parts, using a training model to analyze the ultrasonic images, and then scoring the user according to the analyzed characteristic symptoms, the user's health status is converted into a visual numerical value, providing a quantitative indicator for the severity of gout or asymptomatic hyperuricemia.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of information processing, and in particular to recognition based on ultrasound images. Background Art

[0002] Both asymptomatic hyperuricemia and intermittent gout have high serum uric acid indexes, such as serum uric acid (SUA)>7.0mg / dl, and neither has obvious clinical symptoms. The disease is very harmful to the human body and requires timely treatment according to the actual condition.

[0003] However, due to its unclear clinical manifestations, misdiagnosis often occurs in the traditional diagnosis and treatment process. For example, when the blood uric acid level reaches hyperuricemia, due to the lack of hallmark features, it is impossible to intuitively judge whether it is asymptomatic hyperuricemia or intermittent gout, which has an adverse effect on actual treatment. At the same time, understanding the severity of these two diseases is very important for formulating the intensity of their treatment plans, but in practice there is no quantitative indicator that can reflect the severity. Summary of the invention

[0004] The purpose of the embodiments of the present invention is to provide an ultrasonic image-based recognition device to provide a quantitative indicator for the severity of gout or asymptomatic hyperuricemia.

[0005] In order to solve the above technical problems, an embodiment of the present invention provides a recognition device based on ultrasound images, comprising:

[0006] An acquisition module, used to acquire ultrasonic images of different parts of the user, including: a knee joint, an ankle joint, and a first metatarsophalangeal joint;

[0007] An identification module is used to use N preset training model sets to identify the ultrasound images of different parts and obtain various characteristic symptoms existing in the ultrasound images; there is a one-to-one correspondence between the N training model sets and the parts, and the number of training models included in the training model set is the number of characteristic symptoms to be detected in the parts corresponding to the training model sets;

[0008] A scoring module, used for obtaining a scoring result of the ultrasound image according to the characteristic symptoms existing in the ultrasound image and a preset correspondence between each characteristic symptom and a score value;

[0009] An output module is used to output the user's test results based on the scoring results and a preset score threshold; wherein the test results include: gout or asymptomatic hyperuricemia.

[0010] Compared with the related art, the embodiments of the present invention collect ultrasound images of a preset area and use a training model to analyze the obtained ultrasound images to identify the characteristic symptoms in the ultrasound images, and then score the user according to the characteristic symptoms, and convert the user's health status into a visual numerical value. Since the numerical value is more objective, it can provide a quantitative indicator for the severity of gout or asymptomatic hyperuricemia, and at the same time reduce the interference of other uncontrollable factors. The scoring value is combined with a preset score threshold to judge the health status of the current user, and can also be used to quickly distinguish whether the current user is asymptomatic hyperuricemia or in the intermittent period of gout, so that the user can make corresponding follow-up treatment in time.

[0011] In addition, the score value corresponding to the characteristic symptom is determined by the odds ratio (OR) of the characteristic symptom. Among them, OR is the ratio of the proportion of the test factor in the positive reaction population to the proportion of the test factor in the negative reaction population. For example, if there is an experiment divided into a control group and an experimental group, the OR value is calculated by dividing the ratio of the number of patients to the number of non-patients in the experimental group by the ratio of the number of patients to the number of non-patients in the control group. That is, the OR value can objectively reflect the degree of correlation between the characteristic symptoms and the actual illness, so the score value of the characteristic symptoms is determined by OR, which makes it more effective to infer the user's actual health status based on the characteristic symptoms.

[0012] In addition, after obtaining the test result of the user according to the scoring result and the preset score threshold, the following is included: judging the problem level of the test result according to the scoring result, and the problem level of the disease test result is proportional to the scoring result. That is, the scoring result is proportional to the problem level of the user's test result. The higher the score of the scoring result, the higher the problem level of the test result, and the more serious the problem found in the test result; this feature can quantify the severity of the test result. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] One or more embodiments are exemplarily described by pictures in the corresponding drawings, and these exemplifications do not constitute limitations on the embodiments. Elements with the same reference numerals in the drawings represent similar elements, and unless otherwise stated, the figures in the drawings do not constitute proportional limitations.

[0014] Figure 1 is a schematic diagram of an ultrasonic image-based recognition device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings. However, it will be appreciated by those skilled in the art that in the embodiments of the present invention, many technical details are proposed in order to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical scheme claimed in the present application can also be implemented. The division of the following embodiments is for the convenience of description and should not constitute any limitation on the specific implementation of the present invention. The various embodiments can be combined and referenced with each other without contradiction.

[0016] The terms "first" and "second" in the embodiments of the present application are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present application, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions. For example, a system, product or device comprising a series of components or units is not limited to the listed components or units, but may optionally include components or units that are not listed, or may optionally include other components or units that are inherent to these products or devices. In the description of the present application, the meaning of "multiple" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0017] One embodiment of the present invention relates to an ultrasonic image-based recognition device. Figure 1 shown.

[0018] The acquisition module 101 is used to acquire ultrasound images of different parts of the user, including: knee joint, ankle joint and first metatarsophalangeal joint;

[0019] The recognition module 102 is used to use the preset N training model sets to recognize the ultrasound images of different parts and obtain the characteristic symptoms existing in the ultrasound images; there is a one-to-one correspondence between the N training model sets and the parts, and the number of training models included in the training model set is the number of characteristic symptoms to be detected in the parts corresponding to the training model sets;

[0020] The scoring module 103 is used to obtain a scoring result of the ultrasound image according to the characteristic symptoms existing in the ultrasound image and the preset corresponding relationship between each characteristic symptom and the score value;

[0021] The output module 104 is used to output the user's test results according to the scoring results and the preset score threshold; wherein the test results include: gout or asymptomatic hyperuricemia.

[0022] In this embodiment, ultrasound images of a preset area are collected, and the obtained ultrasound images are analyzed using a training model to identify characteristic symptoms in the ultrasound images. The user is then scored based on the characteristic symptoms, and the user's health status is converted into a visual numerical value. Since the numerical value is more objective, it can provide a quantitative indicator for the severity of gout or asymptomatic hyperuricemia, reducing the interference of other uncontrollable factors. The scoring value is combined with a preset score threshold to judge the health status of the current user, and can also be used to quickly distinguish whether the current user is in asymptomatic hyperuricemia or in the intermittent period of gout, so that the user can make corresponding follow-up treatment in time.

[0023] The implementation details of the ultrasonic image-based recognition device of this embodiment are described in detail below. The following content is only provided for easy understanding of the implementation details and is not necessary for implementing this solution.

[0024] In the acquisition module 101, ultrasound images of different parts of the user are acquired, including the knee joint, ankle joint and first metatarsophalangeal joint. The device used to acquire the ultrasound image may be a medical ultrasound instrument, such as an Applio 500 ultrasound machine with a multi-frequency linear transducer (12-14 MHz). In addition, the knee joint, ankle joint and first metatarsophalangeal joint are all parts where gout is prevalent, that is, by taking ultrasound images of these parts and surrounding structures and soft tissues, the obtained health status information is easier to identify the user's gout-related status information.

[0025] In one example, the user's blood uric acid level meets the criteria of hyperuricemia; that is, the user's blood uric acid level index has been obtained in a previous test, and the blood uric acid level index meets the criteria of hyperuricemia, such as a serum uric acid (SUA) level higher than 7.0 mg / dl. Since hyperuricemia is a well-known cause of gout, or a prerequisite for the development of gout, it is estimated that 5-18.83% of people will later develop gout, so it is necessary to test it in advance. At the same time, many hyperuricemia patients are clinically asymptomatic, and gout patients also have atypical clinical features. The two are easily confused, resulting in that only detecting this prerequisite is not enough to determine the user's actual health status; but the identification device provided in the embodiment of the present invention can solve this easily confused problem.

[0026] In addition, to improve the accuracy of the identification results, users are not recommended to take any medications that may affect uric acid metabolism before taking the test, including: losartan, benzbromarone, allopurinol, febuxostat and non-potassium sparing diuretics.

[0027] In the recognition module 102, the preset N training model sets are used to recognize the ultrasonic images of different parts obtained above, and obtain the characteristic symptoms existing in the ultrasonic images; there is a one-to-one correspondence between the N training model sets and the parts, and the number of training models contained in the training model set is the number of characteristic symptoms to be detected in the parts corresponding to the training model set. Since the parts to be detected are the knee joint, ankle joint and first metatarsophalangeal joint, that is, N is taken as 3, and a total of 3 training models are preset.

[0028] Before conducting statistical analysis, those skilled in the art will know that routine physical condition tests on sample individuals are used to obtain physiological condition information of the sample individuals to make the results of statistical analysis more accurate; wherein, during the physical condition test, the information that needs to be obtained includes: age, gender, disease duration, comorbidities, drug treatment, current and previous diseases, and frequency of attacks. The physical examination included height, weight, waist circumference, hip circumference and blood pressure. The body mass index (BMI) was calculated by dividing the weight (kg) by the square of the height (m). Venous blood samples were collected in the morning after an overnight fast to measure fasting blood glucose (FBG), alanine aminotransferase (ALT), serum creatinine (SCr), serum urate (SUA), blood urea nitrogen (BUN), cystatin C (CYC), C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), glycosylated hemoglobin (HbA1c) and lipid profile.

[0029] To select the best combination of ultrasound abnormalities, the following analyses were performed: First, the chi-square test was used to identify ultrasound features that differed significantly between the sample groups (variables were significantly different, p < 0.05); second, the ultrasound features that differed significantly between the groups were entered into a binary logistic regression analysis to identify discriminative factors, which were the focus of the identification process. Among them, the p value needs to be calculated bilaterally, and a p value < 0.05 is considered statistically significant; for skewed variables in the calculation process, the mean standard deviation, percentage, or median (interquartile range 25-75%) can be used to determine whether it is left-skewed or right-skewed; in addition, the independent sample t test was used to compare normally distributed continuous variables, the Mann-Whitney U test was used to compare non-normally distributed continuous variables, and the chi-square test was used to compare frequencies. The above calculations can be statistically analyzed using the SPSS 23.0 software program.

[0030] In one example, after statistical analysis, it is necessary to detect whether there are symptoms of synovial effusion or characteristic symptoms such as tophi at the knee joint. Among them, tophi are a collection of monosodium urate (MSU) crystals surrounded by chronic inflammatory cells. The penetration of tophi into the bone is closely related to the development of gouty bone erosion, which in turn can lead to joint damage, deformity and eventual musculoskeletal disability. Therefore, there are two training models in the training model set corresponding to the knee joint, including: a first training model for judging whether there are symptoms of synovial effusion at the knee joint; and a second training model for judging whether there are tophi at the knee joint.

[0031] Based on the image feature of widening of the anechoic or hypoechoic joint cavity, it is possible to conclude that synovial effusion symptoms exist. Therefore, for the training of the first training model, there is a first sample set, which includes: a normal knee joint cavity and a set of ultrasound images of the knee joint cavity with synovial effusion. During the training process, the first sample set is input into the basic model, so that the basic model makes corresponding parameter adjustments and result verification with reference to the first sample set, so as to achieve the effect of being able to identify whether synovial effusion exists, that is, to obtain the first training model. In actual use, the function of the first training model is to input the acquired ultrasound image of the user's knee joint into the first training model, and the first training model outputs the conclusion of whether synovial effusion exists or not.

[0032] Based on the image feature that the edges of the uneven high-echo aggregates are not obvious, it can be concluded that there are tophi. Therefore, for the training of the second training model, there is a second sample set, which includes: a set of ultrasound images of normal knee joint high-echo aggregates and knee joint high-echo aggregates that are determined to have tophi; during the training process, the second sample set is input into the basic model, so that the basic model refers to the second sample set to make corresponding parameter adjustments and result verification, so as to achieve the effect of being able to identify whether there are tophi, that is, to obtain the second training model. In actual use, the function of the second training model is reflected in that the acquired ultrasound image of the user's knee joint is input into the second training model, and the second training model outputs the conclusion of the presence or absence of tophi.

[0033] The high, medium and low echoes are all related to the tissue density of the detection site, and the actual judgment standard refers to the relevant standards for medical ultrasound images in the inspection discipline.

[0034] In one example, after statistical analysis, it is necessary to detect whether there are characteristic symptoms such as synovial effusion, synovial thickening, tophi and / or bone erosion at the ankle joint. Among them, bone erosion and destruction are more common in chronic patients with a long course of disease, and its formation is closely related to the deposition of MSU crystals. The deposition of MSU crystals in the musculoskeletal system is the basis of gout attacks, that is, bone erosion and gout symptoms are also closely related. Therefore, there are four training models in the training model set corresponding to the ankle joint, including: a third training model, used to determine whether the ankle joint has synovial effusion symptoms; a fourth training model, used to determine whether the ankle joint has tophi; a fifth training model, used to determine whether the ankle joint has synovial thickening symptoms; and a sixth training model, used to determine whether the ankle joint has bone erosion.

[0035] The training of the third training model is substantially the same as the training of the first training model, except that the third sample set includes: a set of ultrasound images of a normal ankle joint cavity and an ankle joint cavity with confirmed synovial effusion. The training of the fourth training model is substantially the same as the training of the second training model, except that the fourth sample set includes: a set of ultrasound images of a normal ankle joint hyperechoic aggregate and an ankle joint hyperechoic aggregate with confirmed tophi.

[0036] According to the image feature of the presence of abnormal low-echo or high-echo tissue in the joint cavity, it is possible to draw a conclusion that synovial thickening exists. Therefore, for the training of the fifth training model, there is a fifth sample set, which includes: a set of ultrasound images of echo tissue in the joint cavity of a normal ankle joint and an ankle joint with confirmed synovial thickening; during the training process, the fifth sample set is input into the basic model, so that the basic model makes corresponding parameter adjustments and result verification with reference to the fifth sample set, so as to achieve the effect of being able to identify whether the symptom of synovial thickening exists, that is, to obtain the fifth training model. In actual use, the function of the fifth training model is reflected in that the acquired ultrasound image of the user's ankle joint is input into the fifth training model, and the fifth training model outputs the conclusion of whether synovial thickening exists or not.

[0037] According to the image feature of cortical interruption accompanied by stepped contour defects in the longitudinal and transverse fields of view of the joint, it is possible to draw a conclusion that bone erosion exists. Therefore, for the training of the sixth training model, there is a sixth sample set, which includes: a collection of ultrasound images of the cortex of a normal ankle joint and the cortex of an ankle joint with confirmed bone erosion. During the training process, the sixth sample set is input into the basic model, so that the basic model makes corresponding parameter adjustments and result verification with reference to the sixth sample set, so as to achieve the effect of being able to identify whether bone erosion exists, that is, to obtain the sixth training model. In actual use, the function of the sixth training model is to input the acquired ultrasound image of the user's ankle joint into the sixth training model, and the sixth training model outputs the conclusion of whether bone erosion exists or not.

[0038] In one example, after statistical analysis, the first metatarsophalangeal joint needs to be tested for characteristic symptoms such as synovial thickening, tophi, bone erosion and / or double track syndrome. Among them, double track syndrome is also one of the early suspected symptoms of gout. Specifically, the deposition of MSU is first deposited on the surface of the hyaline cartilage of the joint. On high-frequency ultrasound sonograms, the normal hyaline cartilage in the joint is a thin layer structure that is almost echo-free on the surface of the strong echo cortical bone, and its edge echo is slightly stronger. In untreated gout patients, MSU is deposited on the surface of the hyaline cartilage, and the surface of the hyaline cartilage is thickened and the echo is enhanced. The image shows a line-like strong echo on the surface of the cartilage parallel to the joint bone cortex, which is the double track syndrome. Therefore, there are four training models in the training model set corresponding to the first metatarsophalangeal joint, including: the seventh training model, used to determine whether there is tophi in the first metatarsophalangeal joint; the eighth training model, used to determine whether there are symptoms of synovial thickening in the first metatarsophalangeal joint; the ninth training model, used to determine whether there is bone erosion in the first metatarsophalangeal joint; and the tenth training model, used to determine whether there is double track disease in the first metatarsophalangeal joint.

[0039] The training of the seventh training model is roughly the same as the training of the second training model, and the difference is that the seventh sample set includes: a normal first metatarsophalangeal joint high echo aggregate and a set of ultrasound images of the first metatarsophalangeal joint high echo aggregate with tophi. The training of the eighth training model is roughly the same as the training of the fifth training model, and the difference is that the eighth sample set includes: a set of ultrasound images of the normal echo tissue of the first metatarsophalangeal joint cavity and a set of ultrasound images of the echo tissue of the first metatarsophalangeal joint cavity with synovial thickening. The training of the ninth training model is roughly the same as the training of the sixth training model, and the difference is that the ninth sample set includes: a normal first metatarsophalangeal joint cortex and a set of ultrasound images of the first metatarsophalangeal joint cortex with bone erosion.

[0040] According to the image feature of irregular high echo enhancement on the edge of the hyaline cartilage surface in the joint, the conclusion of the existence of double track disease can be obtained. Therefore, for the training of the tenth training model, there is a tenth sample set, which includes: a normal hyaline cartilage surface of the first metatarsophalangeal joint and a set of ultrasound images of the hyaline cartilage surface of the first metatarsophalangeal joint with double track disease; during the training process, the tenth sample set is input into the basic model, so that the basic model refers to the tenth sample set for corresponding parameter adjustment and result verification, so as to achieve the effect of being able to identify whether double track disease exists, that is, to obtain the tenth training model. In actual use, the function of the tenth training model is reflected in that the ultrasound image of the first metatarsophalangeal joint of the user obtained is input into the tenth training model, and the tenth training model outputs the conclusion of the existence or non-existence of double track disease. In addition, the feature of the echo enhancement is independent of the acoustic incident angle of the ultrasound beam.

[0041] In the scoring module 103, the scoring result of the ultrasound image is obtained according to the characteristic symptoms existing in each part of the ultrasound image identified in the identification module 102, and the preset correspondence between each characteristic symptom and the score value. Among them, there is a corresponding score value for the characteristic symptom. When a characteristic symptom is identified in the above steps, that is, the part has a score value corresponding to the characteristic symptom, and the score of the part is the sum of the corresponding scores of the characteristic symptoms of the part.

[0042] Among them, the score value corresponding to the characteristic symptom is determined by the OR of the characteristic symptom. OR is the ratio of the proportion of the test factor in the positive reaction population to the proportion of the test factor in the negative reaction population. For example, if an experiment is divided into a control group and an experimental group, the OR value is calculated by dividing the ratio of the number of sick people to the number of non-sick people in the experimental group by the ratio of the number of sick people to the number of non-sick people in the control group. That is, the OR value can objectively reflect the degree of correlation between the characteristic symptoms and the actual health problems, so the score value of the characteristic symptoms is determined by the OR value, which makes it more effective to infer the user's actual health status based on the characteristic symptoms.

[0043] After calculating the OR value, information feedback can also be collected based on actual conditions to verify whether the OR value calculation is effective, for example, using the scores of individual users to calculate c-statistics, Hosmer-Lemeshow tests, accuracy assessments, and calculate the area under the receiver operating characteristic curve (AUC) to evaluate the utility of the ultrasound-based scoring system. The above process can also be calculated using the SPSS software program.

[0044] In one example, the calculated ORs for each characteristic symptom were as follows: synovial effusion (OR, 1.5; P < 0.05), tophi (OR, 5.4; P < 0.05) in the knee; synovial effusion (OR, 2.1; P < 0.05), synovial thickening (OR, 4.9; P < 0.05), tophi (OR, 2.8; P < 0.05), and bone erosion (OR, 7.3; P < 0.05) in the ankle; and double track sign (OR, 1.8; P < 0.05), synovial thickening (OR, 3.0; P < 0.05), tophi (OR, 8.7; P < 0.05), and bone erosion (OR, 4.0; P < 0.05) in the first metatarsophalangeal joint. Therefore, the scores for each characteristic symptom are: knee joint: synovial effusion, 2 points, tophi, 5 points; ankle joint: synovial effusion, 2 points, synovial thickening, 5 points, tophi, 3 points, bone erosion, 7 points; first metatarsophalangeal joint: double track sign, 2 points, synovial thickening, 3 points, tophi, 9 points, bone erosion, 4 points. As can be seen from the above, the highest possible total score is 42 points.

[0045] The scores obtained in the ultrasound image are added together to obtain the final scoring result of the ultrasound image.

[0046] In the output module 104, the test result of the user is output according to the above-obtained scoring result and the preset score threshold; wherein the test result includes: gout or asymptomatic hyperuricemia. The scoring result obtained by the above steps has quantified the health status of the user. In this step, the test result of the user is obtained by comparing the scoring result with the score threshold.

[0047] In one example, if the score result is higher than the preset score threshold, it is determined that the user has gout and is in an intermittent period; if the score result is lower than the preset score threshold, it is determined that the user has asymptomatic hyperuricemia. For example, if the score threshold is 6.5, a score result higher than 6.5 indicates intermittent gout, and a score result lower than 6.5 indicates asymptomatic hyperuricemia; the threshold is determined by the Youden Index (YI), which indicates the ability of a diagnostic test to correctly judge patients and non-patients. The index ranges from 0 to 1. The closer the Youden Index is to 1, the better the authenticity of the diagnostic test, and vice versa.

[0048] In another example, after obtaining the user's test results according to the scoring results and the preset score threshold, it includes: judging the problem level of the test results according to the scoring results, and the problem level of the test results is proportional to the scoring results. That is, after judging that the user has intermittent gout or asymptomatic hyperuricemia, the severity of the related symptom problems can also be judged by the scoring results, and the higher the score, the more serious it is. The severity obtained according to the scoring results is helpful to grasp the corresponding degree of relevant treatment of the user in the later stage.

[0049] In addition, during the verification feedback process, patients were stratified into 6.5, 13, and 26, and the prevalence of intermittent gout was 30.0%, 78.4%, 96.6%, and 100%, respectively. That is, when the score threshold was 26, the accuracy rate of identifying those who exceeded the score threshold as intermittent gout was 100%. According to relevant verification results, the prevalence of bone erosion and tophi symptoms also increased with the increase of the scoring results.

[0050] Compared with the related art, the embodiment of the present invention collects ultrasound images of the knee joint, ankle joint and first metatarsophalangeal joint, and uses the corresponding training model to analyze the obtained ultrasound images to identify the characteristic symptoms present in the ultrasound images. The characteristic symptoms have their corresponding score values ​​determined by the OR value, and then the user is scored according to the characteristic symptoms, and the user's health status is converted into a visual numerical value. Since the numerical value is more objective, it can provide a quantitative indicator for the severity of gout or asymptomatic hyperuricemia. The score value is combined with the preset score threshold to judge the health status of the current user, and can also be used to quickly distinguish whether the current user is asymptomatic hyperuricemia or in the intermittent period of gout, so that the user can make corresponding follow-up treatment in time.

[0051] The division of various device modules above is only for clear description. When implemented, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are all within the protection scope of this patent; adding insignificant modifications to the algorithm or process or introducing insignificant designs without changing the core design of the algorithm and process are all within the protection scope of this patent.

[0052] Those skilled in the art will appreciate that the above-mentioned embodiments are specific embodiments for implementing the present invention, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present invention.

Claims

1. A recognition device based on ultrasonic images, characterized in that: include: an acquisition module, for acquiring ultrasound images of different parts of a user, the parts including: a knee joint, an ankle joint, and a first metatarsophalangeal joint, and the blood uric acid level of the user is consistent with hyperuricemia; An identification module is used to use N preset training model sets to identify the ultrasound images of different parts and obtain various characteristic symptoms existing in the ultrasound images; there is a one-to-one correspondence between the N training model sets and the parts, and the number of training models included in the training model set is the number of characteristic symptoms to be detected in the parts corresponding to the training model sets; A scoring module, used for obtaining a scoring result of the ultrasound image according to the characteristic symptoms existing in the ultrasound image and the preset correspondence between each characteristic symptom and a score value; wherein, when the recognition module recognizes that a certain characteristic symptom of the part exists, the part has the score value corresponding to the characteristic symptom, and the score of the part is the sum of the score values ​​corresponding to the characteristic symptoms of the part; An output module, used to output the test result of the user according to the scoring result and a preset score threshold; wherein the test result includes: gout or asymptomatic hyperuricemia; Wherein, the training model set corresponding to the knee joint includes: A first training model is used to determine whether the knee joint has synovial effusion symptoms, the ultrasound image of the knee joint of the user is input into the first training model, and the first training model outputs whether the knee joint has the synovial effusion symptoms; The second training model is used to determine whether there is tophi in the knee joint. The ultrasound image of the knee joint of the user is input into the second training model, and the second training model outputs whether there is tophi in the knee joint.

2. The ultrasonic image-based recognition device according to claim 1, characterized in that: The score value corresponding to the characteristic symptom is determined by the odds ratio OR of the characteristic symptom; the output module is specifically used to output the detection result of gout when the scoring result is higher than the preset score threshold; when the scoring result is lower than the preset score threshold, output the detection result of asymptomatic hyperuricemia.

3. The ultrasonic image-based recognition device according to claim 2, characterized in that: The output module is further used to output the problem level of the detection result according to the scoring result, and the problem level of the detection result is proportional to the scoring result.

4. The ultrasonic image-based recognition device according to claim 1, characterized in that: The training model set corresponding to the ankle joint includes: A third training model is used to determine whether there is a symptom of synovial effusion in the ankle joint; A fourth training model is used to determine whether there is tophi in the ankle joint; A fifth training model is used to determine whether the ankle joint has synovial thickening symptoms; The sixth training model is used to determine whether bone erosion exists in the ankle joint.

5. The ultrasonic image-based recognition device according to claim 1, characterized in that: The training model set corresponding to the first metatarsophalangeal joint includes: a seventh training model, for determining whether tophi exist in the first metatarsophalangeal joint; an eighth training model, for determining whether the first metatarsophalangeal joint has synovial thickening symptoms; a ninth training model, for determining whether bone erosion exists in the first metatarsophalangeal joint; The tenth training model is used to determine whether the first metatarsophalangeal joint has double track syndrome.

Images