Medical image multi-level evaluation platform capable of protecting privacy

Abstract

The invention relates to a privacy-protecting medical image multi-level evaluation platform, which can solve the problems of abnormal doctor seeing behaviors and patient privacy exposure in the doctor seeing process at the same time. The platform comprises a medical information mobile terminal used for acquiring treatment information associated with a patient, a first user terminal used for acquiring medical image information provided by the patient, and a server in signal connection with the medical information mobile terminal and the first user terminal, the server is configured to judge whether an abnormal doctor seeing behavior exists or not according to medical image information provided by a patient and acquired by the first user side and doctor seeing information associated with the patient and acquired by the medical information mobile side; when it is determined that the abnormal doctor-seeing behavior exists, the patient associated with the doctor-seeing information is marked, and the doctor-seeing information comprises at least one or more of the following items: the identity of the patient, the doctor-seeing time, the examination time, the diagnosis and treatment item and the disease diagnosis.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and in particular to a medical image evaluation platform, and in particular to a privacy-protecting medical image multi-level evaluation platform. Background Art

[0002] Among the abnormal behaviors in medical treatment, the problem of using someone else's name to seek medical treatment has existed for a long time. After the establishment of the medical insurance system, due to the contradiction between the "existence and non-existence" of the medical insurance system among different groups and the "high and low" level of protection, the problem of using someone else's social security card still exists. In addition, due to the indifference of some groups to the rule of law and the influence of their family's economic conditions, they choose to use someone else's social security card to save medical expenses. This behavior disrupts the order of medical security management and harms the vital interests of the majority of insured people. Although the real-name system is currently implemented for medical treatment, in some special cases, such as two people of similar age, similar appearance, and no difference in size, who look like twins, there will still be cases of using medical insurance cards. With the development of the Internet, online medical treatment has gradually become popular, and patients can transmit medical records, medical images and other information to doctors online. However, to a certain extent, doctors cannot judge whether medical images belong to the person himself, especially medical images that hide the patient's personal information to protect privacy. It may happen that a patient provides medical images to a doctor, who uses the images to diagnose and give a diagnosis. Although the patient is the real owner of the medical insurance card, the medical images provided by the patient actually belong to another individual. The current Internet diagnosis and treatment management solutions still cannot solve this problem.

[0003] In today's digital age, the widespread use of medical imaging databases has brought great convenience and opportunities to the medical field. A medical imaging database refers to a system that collects and stores a large amount of medical imaging data, which can provide support for medical research and clinical diagnosis. By analyzing and evaluating these data, doctors and researchers can make more accurate disease diagnoses, further understand the development of diseases and treatment effects, and thus provide more accurate treatment plans. At present, the common means of diagnosing and treating patients is to use a certain medium (X-rays, electromagnetic fields, ultrasound, etc.) to interact with the human body, and to show the structure and density of internal tissues and organs in the form of images. Doctors can judge the health status of the human body based on the information provided by the images. For example, cerebrovascular diseases require imaging examinations such as CT, magnetic resonance imaging, cerebral angiography, and transcranial Doppler ultrasound to assist in determining whether cerebral blood vessels are bleeding, the location, range, and severity of infarction.

[0004] However, with the accumulation and sharing of medical imaging data, the issue of privacy protection has become increasingly prominent. Medical imaging data contains patients' personal physical information, such as medical records, disease diagnosis, etc. The leakage of this information may lead to many problems such as patient identity theft, affecting medical treatment and insurance, and adversely affecting personal freedom and dignity.

[0005] The images obtained by the scanning device include the patient's bones and soft tissues together. However, in some cases, such as when a patient's cerebral blood vessels are found to have lesions, it is actually only necessary to obtain targeted image layers to achieve the purpose of diagnosis. The existing images used for diagnosis at least contain layer information of the patient's bone morphology, which is not obviously helpful for disease diagnosis and may cause the patient's privacy to be leaked, causing trouble to the patient.

[0006] In the prior art, there is an operation of encrypting the privacy data in the medical images uploaded to the medical image database, but the protection of the patient's privacy data in the doctor's diagnosis stage is ignored. For example, the invention application with the publication number CN113853658A provides a method and system for anonymizing the original surgical procedure video, the method comprising: receiving a set of original surgical videos corresponding to the surgical procedure performed in the OR; merging the set of original surgical videos to generate a surgical procedure video corresponding to the surgical procedure; detecting image-based personal identifiable information embedded in a set of original video images of the surgical procedure video; and when the image-based personal identifiable information is detected, the detected image-based personal identifiable information in the surgical procedure video is de-identified. However, the technical solution of the invention can only protect the personal information (age, name, gender, examination number, etc.) on the video material or image material, while the patient's facial image, bone morphology and other information displayed on the video or image are still exposed to the diagnosing doctor. Medical image databases are non-human behaviors based on processor operations for privacy data, and doctors will inevitably base their medical diagnosis on medical image data when making medical diagnosis on patients. At least all of the patient's medical images will be presented to the doctor without reservation, which is obviously a loophole in the comprehensive protection of patient privacy. However, if the treatment method of completely removing the privacy data in the medical images is adopted, the doctor will not be able to give the patient an accurate diagnosis result due to the lack of diagnostic basis. Therefore, to what extent the retention and removal of privacy data can not affect the doctor's diagnosis results, but also to the greatest extent to protect the privacy rights and interests of patients, this is a problem worthy of study and solution. On the other hand, when conducting remote consultation or secondary consultation, the patient provides imaging materials to the hospital for the doctor's remote diagnosis and treatment, involving the technical problem of pretending to be someone else's medical consultation. CN109524098A discloses a method for processing medical information based on data analysis. Although it provides a processing method for fake medical consultation, it is limited to marking the target medical information when it is determined that there is an abnormal use of the target medical consumables, so this processing method is post-supervision, and only for offline medical treatment behaviors that require prescriptions or consumables for medical treatment. However, for remote diagnosis using imaging data, which often only requires a doctor to give treatment recommendations or determine the specific cause and condition of a disease, the technical solution of CN 109524098A is not applicable.

[0007] When seeing a doctor in an outpatient clinic, diagnosis with the help of imaging data requires high-level experts with many years of experience, and often requires appointments a long time in advance, which cannot effectively utilize the experts' abilities to benefit more patients. For this reason, Internet medical care has proposed a business model of Internet + AI medical imaging in this precise diagnosis. However, due to the rich formats of medical images provided by various hospitals, different imaging methods, and different resolutions, AI medical imaging has focused on overcoming these existing technical problems in recent years and has been able to complete corresponding recognition for various types of medical images. However, in order to protect patient privacy or to highlight the location of the lesion, many hospitals provide medical images that are processed by subtraction and only contain images of key parts such as blood vessels (such as DSA imaging), and delete the patient's privacy data. At this time, Internet hospitals are powerless in judging fake visits.

[0008] In addition, on the one hand, there are differences in understanding among those skilled in the art; on the other hand, the applicant studied a large number of documents and patents when making the present invention, but due to space limitations, not all details and contents are listed in detail. However, this does not mean that the present invention does not have the characteristics of these prior arts. On the contrary, the present invention already has all the characteristics of the prior art, and the applicant reserves the right to add relevant prior art to the background technology. Summary of the invention

[0009] With the development of the interaction between the medical system and the Internet, the protection of data information containing the patient's personal privacy is challenged. In particular, medical image-related data (CT, MRI, X-rays and other related images and diagnostic reports), these images usually come with the patient's personal information (age, name, gender, lesion site, etc.), so these medical information needs to be strictly protected, otherwise it may be illegally used. However, if the privacy data in the medical image is completely removed, the subsequent diagnostic doctor will not be able to give the patient an accurate diagnosis due to the lack of diagnostic basis. The existing technology is currently unable to retain the appropriate degree of medical information for subsequent diagnostic doctors to judge the condition, that is, it is impossible to obtain the standard for the degree of retention or removal of privacy data in medical images based on the patient's condition. Too little information on the medical image retained will reduce the accuracy of the diagnostic doctor's judgment of the patient's condition; too much information on the medical image retained will cause the leakage of the patient's personal privacy. In addition, in order to protect patient privacy or highlight the location of the lesion, many hospitals provide medical images that have been subtracted and only contain images of key parts such as blood vessels (such as DSA imaging), and the patient's privacy data has been deleted. At this time, Internet hospitals or second visits cannot determine whether the visit is fake.

[0010] In view of the deficiencies in the prior art, the present invention provides a privacy-protected medical imaging multi-level evaluation platform, the platform comprising: a medical information mobile terminal for acquiring medical information associated with a patient, a first user terminal for acquiring medical imaging information provided by the patient, and a server signal-connected to the medical information system mobile terminal and the first user terminal, the server being configured to: determine whether there is abnormal medical treatment behavior based on the medical imaging information provided by the patient acquired by the first user terminal and the medical information associated with the patient acquired by the medical information mobile terminal; when it is determined that there is abnormal medical treatment behavior, mark the patient associated with the medical information, wherein the medical information includes at least any one or more of the following: patient identity, appointment time, examination time, diagnosis and treatment items, and disease name.

[0011] Beneficial effects of this technical solution: Since the current level of medical and health services cannot meet the actual needs of the general public, uneven allocation of medical resources is a common problem, especially in the medical imaging industry. According to relevant reports, the annual growth rate of medical imaging data is about 30%, while the growth rate of the number of imaging doctors is only 4.1%. Based on this, Internet imaging medicine is proposed. Patients send their historical imaging data to the doctor's end and use online medical treatment to see a doctor, which can promote the rational allocation of medical resources and simplify the medical treatment process and cost. However, in the actual medical treatment process, it cannot be guaranteed whether the medical imaging data provided by the patient is true, and there are abnormal medical treatment behaviors. For patients who visit the hospital for the first time, if the medical imaging data taken in the hospital does not match the associated medical information, it may be that there is a problem with the upload process within the hospital, for example, mixing up the medical imaging information of different patients, and this solution can detect errors in time and avoid misdiagnosis. For the second visit or online medical treatment, the patient will send the historical medical imaging information to the doctor's end. In this case, it cannot be guaranteed whether the medical imaging information provided by the patient is true. In particular, for the purpose of protecting privacy or highlighting the lesion, the image information has been subtracted, so the doctor cannot determine whether the image provided by the patient is his or her own medical image information. This technical solution can determine whether there is abnormal medical treatment behavior by checking the patient's associated medical information with the medical image information provided by the patient.

[0012] Preferably, the medical imaging information provided by the patient and obtained by the first user terminal includes any one or more of the following: original medical imaging information containing the patient's personal privacy information, de-privatized medical imaging information, and medical imaging information after subtraction processing, wherein the de-privatized medical imaging information refers to medical imaging information obtained by removing the patient's basic personal information from the original medical imaging information; and the medical imaging information after subtraction processing refers to medical imaging information obtained by removing image layer information irrelevant to the patient's medical condition.

[0013] Preferably, the server is also used to perform subtraction processing on the original medical image information containing the patient's personal privacy information, wherein the server is configured to: when performing the subtraction processing, pre-set at least one image area in the original medical image information as a verification area, and use the patient's associated medical information as a key to encrypt the image information of the area to form encrypted verification data for verification.

[0014] Beneficial effects of the technical solution: The medical imaging information provided by the patient to the doctor by the first user terminal includes different types, wherein the original medical imaging information contains the patient's personal identity information (name, consultation time, treatment items, etc.), and the server can check the personal identity information on the original medical imaging information and the consultation information associated with the patient and determine whether there is abnormal consultation information. For the de-privatized medical imaging information, the server can obtain the patient's personal identity information corresponding to the de-privatized medical imaging information by AI restoration, and then check it with the consultation information associated with the patient to determine whether there is abnormal consultation information. For the medical imaging information after subtraction processing, the server can pre-set at least one image area in the medical image as a verification area, and use the consultation information associated with the patient as a key to encrypt the image information of the area to form encrypted verification data for verification. When conducting a second visit, the Internet hospital uses the consultation information associated with the patient to decrypt the encrypted verification data for verification, and then obtains the verification area, and determines whether there is abnormal medical behavior by analyzing the similarities and differences between the verification area and at least one image area in the medical imaging information provided by the patient as the verification area.

[0015] Preferably, the privacy-protected medical imaging multi-level evaluation platform also includes a second user terminal for evaluating the patient's medical condition, wherein the server is configured to: split the de-privacyized medical imaging information into layers based on the patient's medical condition evaluated by the second user terminal, screen the lesion image layers and associated image layers that can be used to evaluate the patient's condition to form primary image information, the second user terminal forms subtraction-processed medical imaging information based on the correlation between the associated image layers and the lesion image layers in the primary image information, and the subtraction-processed medical imaging information has at least one of the verification areas.

[0016] The beneficial effects of this technical solution: through preliminary de-privacy, the patient's personal basic information on the original medical image information is first removed and renumbered to obtain the initialized image, which can not only ensure that the patient's personal information is protected, but also ensure that the medical image information of each patient will not be confused, especially for patients with the same name and age. This setting can effectively avoid the occurrence of confusion in medical image information. Furthermore, the present application performs layer splitting on the obtained initialization image information to obtain separate image layers. The server compares the separate image layers with the standard image information, obtains the lesion image layers and the associated image layers, and generates primary image information. This step is secondary de-privacy. The steps of preliminary de-privacy and secondary de-privacy are both completed by the equipment, which reduces the workload of doctors, and extracts the layers related to the patient's condition from the original medical image information and provides them to doctors who do subsequent work. The image layers that are not related to the patient's condition or diagnosis are hidden. This process effectively protects the patient's personal information. The evaluation platform uses the server to reduce the complete image information so that the reduced image information only retains the image layers that are useful for medical diagnosis and hides the image layers that are not related to medical diagnosis. When the doctor makes a diagnosis, only part of the layer information is presented to the doctor, so that the doctor has sufficient basis for judgment while protecting the user's privacy rights to the greatest extent.

[0017] Secondly, the medical imaging information after subtraction processing has at least one verification area. When the patient provides the medical imaging information after subtraction processing to doctors in other hospitals for diagnosis, the authenticity of the medical imaging information provided by the patient can be judged through the verification area and the patient's associated medical information, thereby avoiding abnormal medical behavior.

[0018] Third, the correlation between the lesion image layer and the associated image layer in the evaluation platform of the present application is also evaluated by the second user. This is because the medical differences have significant characteristics, such as different diseases with the same symptoms, and the same disease with different symptoms. For different individuals with different conditions, their clinical symptoms and image manifestations may be consistent; and when different individuals have the same condition, their clinical symptoms and image manifestations may be inconsistent. Therefore, the correlation between the associated image layer and the lesion image layer in the medical image information after subtraction processing can be evaluated by the second user. For example, a doctor with rich work experience diagnoses and identifies to accurately judge the correlation between the lesion image layer and the associated image layer, thereby greatly avoiding the occurrence of delays in the patient's diagnosis due to machine misjudgment. Compared with the prior art, the present application hides information that is not related to the diagnosis of the condition through two deprivalizations to avoid unnecessary privacy leaks. After obtaining the medical image information after subtraction processing, the correlation between the lesion image layer and the associated image layer is manually reviewed to see if it is accurate. This step of the present application can overcome the errors caused by machine interpretation. This step in the prior art is completed by a machine. Although it can reduce the burden on doctors to a great extent, the error value of the machine's judgment of correlation is relatively large. For example, the human anatomical structure is quite complex, and it may have normal changes or variations that are not caused by diseases. When interpreted by a machine, this part of the layer will be included in the medical image information after subtraction processing, affecting the subsequent doctor's judgment, which is very unfavorable for the patient's condition assessment.

[0019] Preferably, the initial de-privacy refers to removing or concealing the basic personal information of the patient such as name and gender in the original medical image information.

[0020] Preferably, the privacy-protected medical imaging multi-level evaluation platform also includes a medical imaging database, wherein an imaging model of the lesion site and related sites is pre-established in the medical imaging database to obtain the lesion image layer and the associated image layer. The server is connected to the medical imaging database by signal to retrieve the image data model established in the medical imaging database, wherein the image data model includes standard imaging information as a basis for judging the patient's medical condition, and the server compares each independent image layer with the standard imaging information stored in the medical imaging database, and then selects the lesion layer and the lesion-related layer to form the primary imaging information.

[0021] Taking brain CT as an example, CT results can be used for skull lesions (skull structure, morphological changes, such as skull fractures, dislocations, skull defects, atlanto-occipital malformations caused by trauma), vascular lesions (brain CT can be used to detect high-density or low-density lesions in the cranial area, and intracranial hematomas caused by trauma), brain tissue lesions (whether the brain tissue has atrophied, and the location and degree of atrophy), ventricular lesions (whether the ventricles have uniform watery low-density areas, and then check whether there are diseases such as hydrocephalus, holoprosencephaly, and subependymoma), etc. According to the patient's examination items and specific conditions, the initialized image information is split into layers, and the lesion image layers and associated image layers are output according to the image model. This setting can not only retain the effective layers for patient condition assessment or diagnosis, but also reduce the difficulty of doctor's diagnosis and reduce the possibility of leakage of patients' personal privacy. In addition, in the past, directly using raw medical images would take up a lot of storage space. The large size of medical images caused delays or freezes in the data processing process. The raw medical image information took up more computer resources, reducing the computer's recognition efficiency of target information in medical images. Medical images also contain a lot of irrelevant information, which not only interferes with the prediction of target information but also reduces the accuracy of target information prediction.

[0022] Preferably, the server is connected to the medical imaging database signal to retrieve the image model established in the medical imaging database, wherein the image model contains standard imaging information as a basis for judging the patient's medical condition. The server compares each independent image layer with the standard imaging information stored in the medical imaging database, and then screens the lesion layer and the lesion-related layer to form primary imaging information.

[0023] Preferably, the second user terminal adopts an evaluation system including at least three experienced evaluators performing independent evaluation and review evaluation to evaluate the medical image information after subtraction processing.

[0024] Preferably, the second user terminal is configured to: determine the correlation between the associated image layer and the lesion image layer in the primary image information, display or hide the associated image layer based on the correlation to form medical image information after subtraction processing, evaluate the secondary image and generate an evaluation report.

[0025] The beneficial effects of this technical solution are as follows: the first-level doctor can judge the associated image layers in the first-level image information and evaluate the importance of the associated image layers in evaluating the patient's condition. For example, the first-level doctor makes an initial assessment of whether the "brain tissue layer" is a layer that is truly needed for medical diagnosis. If it is judged to be truly needed, the layer is set as a development layer; if it is judged not to be needed, the layer is set as a subtraction layer. In this way, subtraction-processed medical image information is generated for evaluation by doctors other than the first-level doctor. This setting can filter out image information that can truly and accurately reflect the patient's condition and provide it to other doctors. It not only protects the patient's privacy information to the greatest extent, but also eliminates interference information when the doctor makes a medical diagnosis, reduces the possibility of interference from irrelevant layers affecting the doctor's judgment, and improves the efficiency of the inspection.

[0026] Specifically, two evaluators among the second users perform independent evaluations, and one evaluator performs a review evaluation.

[0027] Preferably, the three experienced evaluators of the second user end include at least a first-level doctor who is an imaging diagnosis doctor, a second-level doctor who is an attending physician, and a third-level doctor who is an in-hospital and out-of-hospital expert.

[0028] Preferably, the correlation between the associated image layer and the lesion image layer in the medical image information after subtraction processing can be judged according to the work experience of the second user. The advantage of this setting is that the server first splits the initialization image to obtain independent image layers, and assigns weights to each independent layer by comparing with the standard image, that is, obtains the importance of each independent image layer in the patient's condition assessment. In this step, the server pre-screens the independent image layers that are different from the standard image to form the first-level image information. This step reduces the workload of the image diagnosis doctor. Further, the image diagnosis doctor evaluates or confirms the correlation between the associated image layer and the lesion image layer in the first-level image information. This is because medicine has distinct characteristics: different diseases have the same symptoms, and the same disease has different symptoms. For example, in the field of tumors, due to individual differences, the laws they embody vary greatly. Each specific clinical symptom and image manifestation are completely inconsistent, and doctors with rich work experience are required to diagnose and identify; in addition, the human body structure may have normal variations that are not caused by the disease. If this correlation judgment work is completed by a machine, it will increase the possibility of misjudgment. Manual review can reduce the error value, ensure that the screened layers can accurately reflect the patient's condition, and minimize the possibility of leakage of the patient's personal privacy.

[0029] Preferably, the privacy-protected medical imaging multi-level evaluation platform also includes a management module capable of obtaining original medical imaging information from a first user terminal and obtaining subtraction-processed medical imaging information from a second user terminal, wherein the management module is provided with special access rights for the original medical imaging information. The benefit of this setting is that the platform stores the original medical imaging information in an independent management module, encrypts it additionally, and establishes unique access rights, so that the privacy information of all users can be managed in a unified manner, thereby reducing the possibility of privacy information leakage. At the same time, the original images containing all personal information are also important for filling and enriching the medical imaging database, as well as for further research on related pathologies.

[0030] Preferably, the image model established in the medical image database is a correlation model for the lesion sites and sites related to the lesions of various diseases, wherein the correlation model can be continuously updated based on the image information fed back by the management module. The benefit of this setting is that since the image information fed back by the management module is medical image information after subtraction processing judged by a first-level doctor, the medical image database is updated based on actual manual judgment. The updated data has the advantages of accuracy and reliability, thereby ensuring that the judgment of the correlation between the lesion sites and related sites of the same disease in the medical image database is more accurate.

[0031] Preferably, the server obtains the privacy protection percentage of the first user to measure the degree of privacy protection of the first user.

[0032] Preferably, the first user terminal is operated by a patient, and the evaluation result of the medical image information after the subtraction processing generated by the second user terminal can be displayed on the first user terminal so that the patient can view the evaluation result.

[0033] Preferably, the server can be an embedded server. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a simplified module connection relationship diagram of a privacy-protecting medical image multi-level evaluation platform according to a preferred embodiment of the present invention;

[0035] Figure 2 This is a simplified module connection relationship diagram of a medical image multi-level evaluation platform for protecting privacy according to another preferred embodiment of the present invention;

[0036] Figure 3 is a schematic diagram of original medical imaging information of a preferred embodiment provided by the present invention;

[0037] Figure 4 is a schematic diagram of initializing image information according to a preferred embodiment of the present invention;

[0038] Figure 5 It is a schematic diagram of subtracted image information according to a preferred embodiment of the present invention.

[0039] Reference numerals list

[0040] 100: first user terminal; 200: server; 300: second user terminal; 400: medical image database; 500: management module; 600: medical information mobile terminal. DETAILED DESCRIPTION

[0041] The following is a detailed description with reference to the accompanying drawings.

[0042] Original medical imaging information refers to the initial images reflecting the physical information of the people being examined, obtained by scanning with specific examination equipment (such as computer tomography equipment, magnetic resonance imaging, etc.).

[0043] Image information irrelevant to the medical condition of the first user refers to information that is not necessary for medical diagnosis and will not affect the medical diagnosis, such as cerebrovascular lesions and skull layer information, which are irrelevant image information.

[0044] Medical conditions refer to the types of conditions or illnesses in the population being examined.

[0045] Special access rights refer to a security access mechanism that provides a control access restriction that only allows specific users or programs to access specific parts of the internal network or specific resources.

[0046] The medical information mobile terminal refers to the port where patients can make appointments, follow-up appointments, online consultations, etc. For example, the medical information mobile terminal can be a self-service registration machine in a hospital, or a reservation registration software in a mobile device.

[0047] Abnormal medical treatment behavior in this application refers to the behavior where the medical treatment information associated with the patient is inconsistent with the medical imaging information provided by the patient.

[0048] Example 1

[0049] This embodiment provides a privacy-protected medical imaging multi-level evaluation platform. Figure 1As shown, the platform includes: a medical information mobile terminal 600 for obtaining medical information associated with a patient, a first user terminal 100 for obtaining medical imaging information provided by a patient, and a server 200 connected to the medical information mobile terminal and the first user terminal 100 by signal, and the server 200 is configured to: determine whether there is abnormal medical behavior according to the medical imaging information provided by the patient obtained by the first user terminal 100 and the medical information associated with the patient obtained by the medical information mobile terminal 600; when it is determined that there is abnormal medical behavior, the patient associated with the medical information is marked. According to a preferred embodiment, the medical information includes at least one or more of the following: patient identity, medical time, examination time, medical treatment items, and disease name. Preferably, when it is determined that there is abnormal medical behavior, for example, the patient's identity, medical time, examination time, medical treatment items, and disease name do not match the medical imaging information provided by the patient, the server 200 marks the patient. Preferably, the privacy-protecting medical imaging multi-level evaluation platform also includes a marking module connected to the server 200 by signal to mark patients with abnormal medical treatment. Preferably, the server 200 generates a labeling task after identifying abnormal medical treatment behavior, and then sends a labeling request for abnormal patients to the labeling module, and the labeling module labels the corresponding patients after receiving the request. Preferably, the labeling module can be an independent application or software component of the labeling client.

[0050] According to a preferred embodiment, the medical imaging information provided by the patient and obtained by the first user terminal 100 includes any one or more of the following: original medical imaging information containing the patient's personal privacy information, de-privatized medical imaging information, and medical imaging information after subtraction processing, wherein the de-privatized medical imaging information refers to medical imaging information in which the basic personal information of the patient is removed from the original medical imaging information; and the medical imaging information after subtraction processing refers to medical imaging information in which the image layer information irrelevant to the patient's medical condition is removed.

[0051] According to a preferred embodiment, the server 200 is also used to perform subtraction processing on the original medical image information containing the patient's personal privacy information, wherein the server is configured to: when performing the subtraction processing, pre-set at least one image area in the original medical image information as a verification area, and use the patient's associated medical information as a key to encrypt the image information of the area to form encrypted verification data for verification.

[0052] Preferably, the verification area is set by using digital watermark information to perform a preset encryption process in the original medical image information, wherein the digital watermark information is encrypted using the patient's associated medical information as a key.

[0053] Preferably, when the medical consultation information associated with the patient and the verification area do not match, the patient is marked.

[0054] According to a preferred embodiment, the privacy-protected medical imaging multi-level evaluation platform further includes a second user terminal 300 for evaluating the medical condition of the patient, such as Figure 2 As shown, the server 200 is configured to: split the de-privacy medical image information into layers based on the medical condition of the patient evaluated by the second user terminal 300, and screen the lesion image layer and the associated image layer that can be used to evaluate the patient's condition to form the primary image information. The second user terminal 300 forms the subtracted medical image information based on the correlation between the associated image layer and the lesion image layer in the primary image information, and the subtracted medical image information has at least one of the verification areas.

[0055] This embodiment is described in detail using the head diagnosis of a patient as an example. Preferably, the first user terminal 100 refers to a device or port where the patient can upload and submit historical medical imaging information or query reports by himself. For example, the first user terminal 100 is a mobile device, a hospital test report self-service machine, an ultrasound report self-printer, a hospital automatic film pickup machine, etc. Due to the current problem of uneven distribution of medical resources between regions, the online telemedicine model has been promoted, and patients can send their historical medical imaging information to the doctor's end through mobile devices to see a doctor for treatment. Accordingly, patients can query their medical results, medical recommendations, etc. through mobile devices.

[0056] The patient undergoes a head examination, and the original medical image information is obtained after the scan. The original medical image information includes all private information such as the patient's name, age, examination time, and images of the scanned body parts. Preferably, the first user terminal 100 is connected to the server 200 by signal. Specifically, the connection method between the first user terminal 100 and the server 200 is a wireless connection or a wired connection. Specifically, the connection method between the first user terminal 100 and the server 200 can be a Bluetooth connection, WCDMA, WLAN, FSO, etc. The first user terminal 100 sends the original medical image information obtained by the scan to the server 200. The medical image information here refers to the patient's head CT image. The server 200 preliminarily de-privatizes the patient's head CT image to obtain the initialized image information, such as Figure 3 Preferably, the initial de-privacy refers to hiding the basic personal information marked on the image, such as the name, gender, age, etc. of the first user in the original medical image information. Preferably, the de-privacy includes at least the initial de-privacy and the secondary de-privacy.

[0057] Preferably, the method for preliminary de-privatization includes: receiving an original medical image; detecting image-based identifiable basic personal information embedded in the original medical image; when identification text of the identifiable basic personal information is detected, blurring the identification text or concealing the identification text by other means, so as to de-identify / de-privatize the image-based basic personal information.

[0058] After the server 200 performs preliminary de-privacy processing on the original image information, it generates initialization image information, such as Figure 4 Preferably, the initialization image information replaces the basic personal information of the patient in the original medical image information with the new number information. Preferably, the replacement of the basic personal information with the new number information is automatically completed by the smart device without the participation of the doctor, and this step can reduce the leakage of the patient's personal basic information. Preferably, the removal of the basic personal information can be achieved by removing or blacking out the corresponding part in the original medical image.

[0059] In the past, in order to distinguish the imaging data of different patients, staff or equipment usually automatically attached the patient's specific identity information to the medical image: such as name, gender, age (date of birth), examination date / time, etc., but this operation method easily leads to the leakage of the patient's personal privacy. Therefore, it is necessary to de-privatize the acquired original images and add appropriate labels. This method can not only protect the privacy of patients, but also improve the identifiability of patients' imaging data. In particular, in the past, the patient's name, gender, examination time, etc. were used as labels to distinguish the imaging data of different patients, but it is inevitable that there will be patients with the same name and gender on the same day. When there are many patients, the doctor himself is busy with work and cannot remember whether there are patients with the same name and the same gender, which may cause the patient's imaging data to be indistinguishable.

[0060] Preferably, the numbering method of medical images includes: setting a numbering cycle, determining whether the current numbering cycle is the beginning; when it is determined that the current numbering cycle is the beginning, setting the examination label number as the starting value within the numbering cycle. The advantage of setting a cycle is that the image information generated in a certain cycle is divided into a management data set to avoid generating an examination sequence number with too much data due to too many examinations, thereby reducing the number of examination label numbers to be managed, facilitating management, and reducing the workload of doctors. In addition, label management is performed within a certain cycle to avoid duplication of patient examination label numbers, which brings additional difficulties to the distinction of image data.

[0061] When a patient undergoes an examination, the starting value of the examination tag number is assigned to the patient currently undergoing a medical imaging examination and the examination tag number is marked on the patient's initialization image; the examination tag number is incremented according to a preset increment interval to generate a new examination tag number.

[0062] Preferably, the numbering method of medical images can be random numbering, specifically: set a numbering cycle, in each numbering cycle, compare the currently generated examination tag number with the examination tag number that has been generated in the current numbering cycle, if there is an identical examination tag number, then regenerate an examination tag number until the generated examination tag number is different from the examination tag number that has been generated in the current numbering cycle. The advantage of randomly generating examination tag numbers in each numbering cycle is that it reduces the number of examination tag numbers and the number of comparisons, thereby reducing the number of management of working doctors and reducing the operating pressure of the system. Preferably, the examination tag number is unique and definite.

[0063] Preferably, the numbering cycle can be pre-set according to actual conditions. For example, the numbering cycle can be set to "day", that is, the inspection tag number is reset once a day, and the inspection tag number returns to the starting value every day. The numbering cycle can be set to "week", that is, the inspection tag number is reset once a week, and the inspection tag number returns to the starting value every week. The numbering cycle can be set to "month", that is, the inspection tag number is reset once a month, and the inspection tag number returns to the starting value every month. Specifically, when the numbering cycle is "day", if the date of the day is February 1, 2023, the starting value of the inspection tag number of the day can be set to 20230201001, 202302010001, 2023020100001, etc. The inspection tag numbers of the next patient are 20230201002, 202302010002, 2023020100002, that is, the increment interval of the inspection tag number is +1. Preferably, the increment interval of the inspection tag number can also be other values, such as +2, etc. Preferably, when the numbering cycle is "week" or "month", the setting method of the inspection tag number is similar to that of the numbering cycle is "day", which will not be repeated here.

[0064] Preferably, the initialization image is processed to obtain an independent image layer. Preferably, the obtained independent image layer constitutes primary image information. The independent image layer here refers to the separation of the initialization image that can reflect the patient's body structure or physiological health information to obtain a separate image layer reflecting a specific direction or angle. For example, for cerebrovascular disease imaging evaluation, only vascular information is required here. Therefore, after processing the initialization image, independent layers containing only vascular information, only skull information, and only brain tissue information are obtained. The initialization image is subtracted according to the actual examination purpose or medical condition of the patient.

[0065] Preferably, the method for processing the initialization image is: selecting a target initialization image stored in a medical image data set; and performing image segmentation.

[0066] Preferably, image segmentation includes manual segmentation, threshold segmentation and segmentation algorithm.

[0067] Preferably, manual segmentation requires the processing personnel to perform a simple splitting of the initialization image, such as extracting layers of images of different parts such as soft tissues and bones, so that the layers are simply layered, and then judging each layer to retain the lesion or the layer associated with the lesion. For example, an independent image layer only shows the bones, soft tissues / cartilages or only blood vessels in the inspection area.

[0068] Region-based image segmentation technology - threshold segmentation principle: divide image pixels into several categories (not only can greatly compress the amount of data, but also greatly simplify the analysis or processing steps), which is a necessary image preprocessing process before image analysis, feature extraction and pattern recognition. Through threshold segmentation, the head frame CT value of the patient's brain image can be analyzed while the skin and subcutaneous tissue are segmented to form several layers. (Threshold segmentation is real-time. For different times, the image thresholds generated by the same part of the human body are different, that is, the threshold standards of different batches of images to be processed change).

[0069] The segmentation algorithms include region growing algorithm, module matching, fuzzy C-means and level set method.

[0070] Here, the segmentation of the vascular image is taken as an example, which specifically includes: extracting the vascular from the initialization image; obtaining the corresponding vascular distribution map. Preferably, the vascular extraction can be performed on the initialization image by a threshold segmentation method, or the initialization image can be processed by at least one filtering algorithm. Specifically, the filtering algorithm can be at least one of a linear filtering algorithm, a Gaussian filtering algorithm, a threshold filtering algorithm, and a Wiener filtering algorithm. Specifically, the steps of using at least one filtering algorithm to process the initialization image to obtain the vascular distribution map are:

[0071] Calculate the eigenvalues ​​of the Hessian matrix for each pixel in the initialization image;

[0072] The above eigenvalues ​​are processed by a tubular filter to obtain the response value;

[0073] The response values ​​corresponding to each pixel are combined to obtain a blood vessel distribution map.

[0074] The advantage of this method is that the eigenvalue of the Hessian matrix of the pixel point can be associated with the shape feature corresponding to the pixel point, so the initialization image is associated and mapped by calculating the eigenvalue of the Hessian matrix of each pixel point, and each eigenvalue is enhanced by a tubular filter to enhance the image of the vascular tissue with tubular morphology and suppress the image of other non-vascular tissues. After the eigenvalue is enhanced by the tubular filter, the response value corresponding to each pixel point is obtained, and each response value is combined according to the arrangement order of each pixel point to obtain a blood vessel distribution map.

[0075] The initialization image and the obtained blood vessel distribution map are input into the blood vessel segmentation model to obtain the blood vessel segmentation result. The blood vessel segmentation model can be obtained by collecting a large amount of learning data to train the existing model. Specifically, the existing model is a convolutional neural network model, such as a U-Net convolutional neural network, a V-Net convolutional neural network, etc.

[0076] Preferably, the above method for obtaining a blood vessel layer can also be used to extract layers of other tissue or structural features, for example, to extract a bone layer, a tissue layer, etc.

[0077] Preferably, secondary de-privacy refers to removing independent layers that are not related to the patient's medical condition or illness, and only retaining layers that can directly reflect the patient's lesions or are related to the lesions. In this embodiment, the layers related to the lesions are also represented by associated image layers, and the two have the same meaning. The lesion image layer refers to the layer that directly reflects the patient's medical condition or illness.

[0078] Preferably, the second user terminal 300 is operated by a second user, including doctors or medical personnel of different grades.

[0079] Preferably, the second user is divided into three levels of evaluators according to different experiences, including at least a first-level doctor as an imaging diagnosis doctor, a second-level doctor as an attending doctor, and a third-level doctor as an expert inside and outside the hospital, so as to ensure the accuracy and reliability of the evaluation results of the second user end. The imaging diagnosis doctor can evaluate the quality of the imaging technology, judge the nature of the lesion according to the imaging results and issue a diagnosis report. The imaging diagnosis doctor can screen out the image layers that are not related to the patient's condition, reducing the possibility of privacy leakage of the patient. The reserved image layers are then judged by the attending doctor, who is more familiar with the patient's condition and is good at comprehensive and systematic analysis. On the basis of understanding the patient's condition, he can integrate the connection between each image layer and the overall image layer, so as to judge the correlation between the associated image layer and the lesion image layer. The advantage of the independent evaluation of the image layer by the imaging diagnosis doctor and the attending doctor is that it combines the diagnostic thinking, theoretical knowledge and work experience of different doctors, the evaluation process does not affect each other, and the independence and objectivity of the evaluation process are guaranteed. In addition, the compound evaluation of the medical imaging information after subtraction processing by experts inside and outside the hospital further ensures the accuracy and reliability of the evaluation results of the second user end. This setting can not only hide unnecessary patient privacy information, but also ensure the accuracy and reliability of the screened medical imaging information after subtraction processing.

[0080] Preferably, the server 200 splits the initialization image information and obtains a plurality of independent image layers. Preferably, the number of independent image layers is affected by the means of obtaining the image information. Different acquisition means will result in different numbers of image layers finally split out. For example, head image information can be obtained by means of X-ray, CT, magnetic resonance imaging, cerebral angiography, and electroencephalogram, and the processor generates independent image layers including skull, cerebral blood vessels, brain tissue, brain vessel wall, cerebral hemodynamics, etc.

[0081] Preferably, in this embodiment, images related to cerebrovascular diseases are mainly collected, such as brain tissue examination images - brain CT and MRI; cerebrovascular examination images - cerebral angiography (DSA), cerebrovascular CTA, cerebrovascular MRA; brain vessel wall evaluation images - high-resolution nuclear magnetic resonance imaging and intravascular optical coherence imaging (OCT). Cerebral hemodynamic examination images - CTP, PWI and ASL, etc.

[0082] Preferably, the server 200 is connected to the medical image database 400 by signal to retrieve the image model established in the medical image database 400. The server 200 selects the lesion image layer and its associated image layer that include the disease representation in each layer to form the primary image information. The medical image database 400 is a data storage center in the medical image information system, which provides a large amount of medical image data based on the second user terminal label. The data set can be used to train or verify the artificial intelligence medical image evaluation model to serve medicine.

[0083] Preferably, the medical image database 400 is pre-established with image models of lesion sites and lesion-related sites of various diseases. Preferably, the image model contains standard image information as a basis for judging the patient's medical condition and illness. Preferably, the image comparison method of this embodiment includes the following steps: acquiring a target image and dividing the target image into light and dark, and completing the image comparison by comparing the divided areas with the standard images stored in the medical image database 400. Specifically, the area division can be performed in the target image according to the light and dark changes. Accordingly, the area division method of the standard image and the target image is the same. The divided areas in the target image are matched with the divided areas of the standard image to obtain the light and dark differences between the corresponding areas in the target image and the standard image, and all the divided areas are matched to complete the image comparison.

[0084] Preferably, the server 200 compares each independent image layer with the standard image information stored in the medical image database 400 and then selects the lesion layer and the lesion-related layer to form primary image information.

[0085] Preferably, a correlation model between the lesion site and the related site is pre-established in the medical image database 400. Preferably, the correlation model is established based on the statistical results of the medical record data of the lesion site and the related site of each disease collected in the past. The correlation model outputs the lesion image layer and the related image layer related to the lesion after comparing each independent image layer with the image information stored in the medical image database 400. For example, through comparison, if there is a lesion in the patient's cerebral blood vessels, the cerebral blood vessel layer is marked and defined as the lesion image layer; and the result of the correlation model shows that there is a correlation between the brain tissue and the cerebral blood vessels, that is, the brain tissue may also reflect the pathological condition of the cerebral blood vessels to a certain extent, so the brain tissue layer is also marked and defined as the associated image layer, and the associated image layer dependent on the lesion image layer can be obtained through the correlation model. There may be an invisible pathological relationship between the lesion image layer and the associated image layer. Due to the shortage of professional doctor resources, when facing the imaging data of many patients, the workload of professional doctors is huge. Therefore, the association model between the lesion site and the related sites set in the present application can not only reduce the burden on doctors, but also provide a comprehensive and accurate treatment basis for subsequent disease assessment on the basis of screening out the subtraction layer.

[0086] Preferably, an image marking module is provided in the server 200. The image marking module is used to mark and / or assign weights to independent image layers. For example, if the server 200 detects that a lesion exists in the patient's cerebral blood vessels, the image marking module marks the cerebral blood vessel layer and defines it as a lesion image layer, and assigns the image dimension type to the lesion image layer, indicating that the layer is a preliminary diagnosis of any type and disease in dimensions such as histology, morphology, structure and function, and includes it in the first-level image layer. In the above case, the remaining layers are subtraction layers, which are non-essential layers for medical diagnosis and will not affect the medical diagnosis. Therefore, the patient's privacy information can be hidden by hiding the subtraction layers, that is, the information such as the skull layer is removed to obtain the subtracted image information, thereby avoiding the leakage of layers that are not necessary for diagnosis, such as Figure 5 shown.

[0087] According to the image dimension type and the preliminary diagnosis of the disease marked by the server 200, the second user terminal 300 will select different evaluation index templates. After three-level evaluation, the secondary image layer will be given medical experience evaluation and marking, such as 87% vascular stenosis, cerebral artery-arterial infarction mechanism, hemodynamic disorder, plaque instability and other professional evaluation results. Finally, a standard medical image marking database is formed.

[0088] Preferably, the second user terminal 300 is connected to the server 200 by signal. Specifically, the connection mode between the second user terminal 300 and the server 200 is wireless connection or wired connection. Specifically, the connection mode between the second user terminal 300 and the server 200 can be Bluetooth connection, WCDMA, WLAN, FSO, etc. Preferably, the server 200 sends the primary image information to the second user terminal 300 for the second user to evaluate.

[0089] Preferably, the second user terminal 300 is configured to: determine the correlation between the associated image layer and the lesion image layer in the primary image information; display or hide the associated image layer based on the correlation to form the medical image information after subtraction processing; evaluate the secondary image and generate an evaluation report. Preferably, the first user terminal 100 allows the first user to view the evaluation report by obtaining the evaluation report from the second user terminal 300.

[0090] Preferably, the first-level doctor makes a judgment on the associated image layers in the first-level image information and evaluates the importance of the associated image layers in evaluating the patient's condition. For example, the first-level doctor makes an initial evaluation on whether the "brain tissue layer" is a layer that is truly required for medical diagnosis. If it is determined to be truly required, the layer is set as a development layer; if it is determined not to be required, the layer is set as a subtraction layer. In this way, medical image information after subtraction processing is generated for evaluation by other doctors other than the first-level doctor. The advantage of this setting is that it screens out image information that can truly and accurately reflect the patient's condition and provides it to other doctors, which not only protects the patient's privacy information to the greatest extent, but also eliminates interference information when the doctor makes a medical diagnosis, reduces the possibility of interference from irrelevant layers affecting the doctor's judgment, and improves the efficiency of the examination.

[0091] Preferably, the correlation between the associated image layer and the lesion image layer in the medical image information after subtraction processing is determined based on the work experience of the second user.

[0092] Preferably, the second user uses an evaluation system including at least back-to-back evaluation and review evaluation to evaluate the medical image information after subtraction processing. The first-level doctor can distribute the medical image information after subtraction processing, and the second-level and above doctors can evaluate the medical image information after subtraction processing in the evaluation system. The evaluation system can be a back-to-back evaluation and a review evaluation. These two evaluation systems are widely used in the field of medical imaging, especially in the diagnosis and treatment of important diseases. Through the independent evaluation and review of multiple experts, the accuracy and reliability of medical image diagnosis can be improved, and better medical services can be provided to patients.

[0093] Preferably, the first user terminal 100 can be controlled by the first user. The evaluation result of the medical image information after the subtraction processing generated by the second user terminal 300 can be displayed on the first user terminal 100, so that the patient can view the evaluation result, and then the patient can understand his own condition. Figure 2 shown.

[0094] Preferably, the image model established in the medical image database 400 is an association model for the lesion site and related parts of various diseases. The association model can be continuously updated based on the image information fed back by the management module 500. Since the image information fed back by the management module 500 is the medical image information after the subtraction process judged by the first-level doctor, the medical image database 400 is updated based on the actual manual judgment, and the updated data has the advantages of accuracy and reliability. Therefore, when the medical image database 400 is used for the subsequent association judgment of the lesion site and related parts of the same disease, it is guaranteed that the obtained results are more accurate. Preferably, the management module 500 can encrypt the channel for medical information. Preferably, the management module 500 can use HTTPS to encrypt the patient's medical image information.

[0095] Preferably, when the image models of the second user terminal 300 and the medical image database 400 have different judgment results on the correlation between the associated image layer and the lesion image layer, the management module 500 feeds back the lesion image layer and the associated image layer output by the second user terminal 300 to the medical image database 400, and updates the association model of the medical image database 400.

[0096] Preferably, the management module 500 can obtain the original medical image information from the first user terminal 100 and the medical image information after subtraction processing from the second user terminal 300. The management module 500 sets a special access right for the original medical image information. The establishment of this special right restricts the access to the original medical image information containing the most patient privacy information, which plays a positive role in protecting the privacy of patients.

[0097] Example 2

[0098] This embodiment is a further improvement on Embodiment 1, and the repeated contents will not be repeated here.

[0099] In this embodiment, the privacy protection percentage of the first user can be obtained according to the server 200 to measure the degree of privacy protection of a patient. Preferably, the basic calculation method of the privacy protection percentage is the ratio of the number of subtraction layers to the number of all image layers. This ratio has a certain reference significance for measuring the degree of privacy protection and health level of the patient. It can be seen from the calculation method that the lower the privacy protection percentage, the fewer subtraction layers and more development layers of the patient, which means that the patient has more lesions and the worse the health level. Correspondingly, the higher the privacy protection percentage, the more subtraction layers and fewer development layers of the patient, which means that the patient has fewer lesions and the health level may be better. It should be understood that the fewer lesions in the patient does not mean that the health level is higher. When evaluating the patient's condition, it is necessary to combine the doctor's diagnosis results to obtain a comprehensive judgment. In addition, this method is also affected by the type and number of image acquisition means on the percentage "denominator". For the comparison of health conditions between different patients who take the same image acquisition means, the privacy protection percentage can play an auxiliary role in judgment. In addition, for the privacy protection percentage of the same patient at different time stages, the change in the percentage can also reflect the patient's recovery. A higher privacy percentage means that some of the patient's diseased parts have been treated and the recovery is better; a lower privacy percentage means that the patient may have more diseased parts and the recovery is not optimistic.

[0100] Example 3

[0101] This embodiment is a further improvement on Embodiment 1, and the repeated contents will not be repeated here.

[0102] Preferably, the platform provided in this embodiment is applicable to various medical modalities, such as CT, MRT, PET, SOECT, MEG (magnetoencephalogram), and EEG (electroencephalogram).

[0103] Preferably, a medical image data set acquired using one modality can be displayed superimposed with another medical image data set acquired using another modality, and at least two medical image data sets of the same examination area, especially the same patient, can be displayed. Preferably, the same examination area can be switched between different medical image data sets.

[0104] It should be noted that the above-mentioned specific embodiments are exemplary, and those skilled in the art can come up with various solutions inspired by the disclosure of the present invention, and these solutions also belong to the disclosure scope of the present invention and fall within the protection scope of the present invention. Those skilled in the art should understand that the present invention specification and its drawings are illustrative and do not constitute a limitation of the claims. The scope of protection of the present invention is defined by the claims and their equivalents. The present invention specification contains multiple inventive concepts, such as "preferably", "according to a preferred embodiment" or "optionally", all of which indicate that the corresponding paragraph discloses an independent concept, and the applicant reserves the right to file a divisional application based on each inventive concept. Throughout the text, the features guided by "preferably" are only an optional method and should not be understood as being required. Therefore, the applicant reserves the right to abandon or delete the relevant preferred features at any time.

Claims

1. A privacy-preserving multi-level evaluation platform for medical images, comprising: A medical information mobile terminal (600) for acquiring medical consultation information associated with a patient, A first user terminal (100) for acquiring medical imaging information provided by the patient, and a server (200) connected to the medical information mobile terminal (600) and the first user terminal (100) by signal, wherein the server is configured as follows: Judging whether there is abnormal medical treatment behavior based on the medical imaging information provided by the patient acquired by the first user terminal (100) and the medical treatment information associated with the patient acquired by the medical information mobile terminal (600); When it is determined that there is abnormal medical treatment behavior, the patient associated with the medical treatment information is marked, wherein: The medical information includes at least one or more of the following: patient identity, medical consultation time, examination time, medical treatment items and disease name.

2. The privacy-preserving medical imaging multi-level evaluation platform according to claim 1, characterized in that: The medical imaging information provided by the patient and obtained by the first user terminal (100) includes any one or more of the following: original medical imaging information containing the patient's personal privacy information, de-privacy medical imaging information, and medical imaging information after subtraction processing, wherein: The de-privacy medical imaging information refers to the medical imaging information obtained by removing the basic personal information of the patient from the original medical imaging information; The medical image information after subtraction processing refers to the medical image information from which the image layer information irrelevant to the medical condition of the patient is removed.

3. The privacy-preserving medical imaging multi-level evaluation platform according to claim 1 or 2, characterized in that: The server (200) is also used to perform subtraction processing on original medical image information containing patient personal privacy information, wherein the server is configured as follows: When performing subtraction processing, at least one image area in the original medical image information is pre-set as a verification area, and the image information of the area is encrypted using the medical information associated with the patient as a key to form encrypted verification data for verification.

4. The privacy-preserving medical imaging multi-level evaluation platform according to any one of claims 1 to 3, characterized in that: The privacy-protected medical imaging multi-level evaluation platform also includes a second user terminal (300) for evaluating the medical condition of the patient, wherein: The server (200) is configured as follows: Based on the medical condition of the patient evaluated by the second user terminal (300), the de-privacy medical image information is split into layers, and the lesion image layer and the associated image layer that can be used to evaluate the condition of the patient are screened to form primary image information. The second user terminal (300) forms subtracted medical image information based on the correlation between the associated image layer and the lesion image layer in the primary image information, and the subtracted medical image information has at least one verification area.

5. The privacy-preserving medical image multi-level evaluation platform according to any one of claims 1 to 4, characterized in that: The privacy-protecting medical image multi-level evaluation platform also includes a medical image database (400), wherein an image model of a lesion site and a related site is pre-established in the medical image database (400) to obtain a lesion image layer and a related image layer. The server (200) is connected to the medical image database (400) by signal so as to retrieve an image data model established in the medical image database (400), wherein the image data model contains standard image information as a basis for determining the medical condition of a patient. The server (200) compares each independent image layer with the standard image information stored in the medical image database (400), and then selects the lesion layer and the lesion-related layer to form primary image information.

6. The privacy-preserving medical image multi-level evaluation platform according to any one of claims 1 to 5, characterized in that: The second user terminal (300) is configured as follows: Determining the correlation between the associated image layer and the lesion image layer in the primary image information; Displaying or hiding the associated image layer based on the correlation to form the medical image information after the subtraction process; The medical image information after the subtraction process is evaluated and an evaluation report is generated.

7. The privacy-protecting medical image multi-level evaluation platform according to any one of claims 1 to 6, characterized in that: The second user terminal (300) uses an evaluation system including at least three experienced evaluators to perform independent evaluation and review evaluation to evaluate the medical image information after subtraction processing.

8. The privacy-preserving medical image multi-level evaluation platform according to any one of claims 1 to 7, characterized in that: The privacy-protecting medical imaging multi-level evaluation platform also includes a management module (500) capable of obtaining the original medical imaging information from the first user terminal (100) and obtaining the subtraction-processed medical imaging information from the second user terminal (300), wherein the management module (500) is provided with special access rights for the original medical imaging information.

9. The privacy-preserving medical image multi-level evaluation platform according to any one of claims 1 to 8, characterized in that: The image data model established in the medical image database (400) is a correlation model for the lesion sites of various diseases and sites related to the lesions, wherein the correlation model can be continuously updated based on the medical image information fed back by the management module (500).

10. The privacy-preserving medical image multi-level evaluation platform according to any one of claims 1 to 9, characterized in that: The first user terminal (100) is operated by the patient, and the evaluation result of the medical image information after the subtraction process generated by the second user terminal (300) can be displayed on the first user terminal (100) so that the patient can view the evaluation result.

Images