Cloud transmission method for MRI image information

Abstract

The application discloses a cloud transmission method for MRI image information and belongs to the technical field of nuclear magnetic image transmission. The cloud transmission method for MRI image information comprises the following steps: step 1, a terminal device generates original information to be sent, and converts the original information into encrypted information; step 2, the terminal device sends the encrypted information and domain name information of the terminal device to a cloud server; and the encrypted information is downloaded from the cloud server, and then the encrypted information is decrypted into the original information by using a decryption key. According to the scheme, a safe and efficient transmission framework is constructed for sensitive MRI image data shared across institutions through encrypted transmission, source authentication and hierarchical key mechanism; and the original data is transmitted after being encrypted, so that information leakage in the transmission process is effectively prevented.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of nuclear magnetic image information communication, in particular to a cloud transmission method for MRI image information. BACKGROUND

[0002] The emergence of the MR-Linac technology brings a major breakthrough to the online adaptive radiotherapy (OART). The core of this technology platform is to highly integrate the diagnostic level magnetic resonance imaging (MRI) system and the medical linear accelerator (Linac) in the same rack to form an integrated treatment device. It is based on this unique architecture that the MR-Linac can fully exert its core advantage - strong real-time imaging and response capability.

[0003] During the treatment process, the patient needs to undergo multiple MR-Linac scans to generate a large number of magnetic resonance images. These images accurately mark the radiotherapy target area (such as tumors and organs at risk), constituting a valuable artificial intelligence (AI) training data set. It is particularly important to use actual treatment efficacy information of the patient to mark the target area, which is of key significance for training high-precision target automatic segmentation models.

[0004] However, the amount of MR image data accumulated by a single medical institution is often insufficient to support robust AI model training. Therefore, multiple hospitals often need to establish an information sharing network to exchange these labeled data through a secure transmission mechanism for their own internal model training and optimization. These magnetic resonance images contain a large amount of sensitive patient privacy information. During cross-institutional transmission, the data is at risk of being illegally stolen, which can easily lead to serious patient privacy leakage. SUMMARY

[0005] The summary part of the present application is used to introduce the concept in a brief form, which will be described in detail in the specific embodiment part. The summary part of the present application is not intended to identify the key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0006] As a first aspect of the present application, in order to solve the technical problems mentioned in the background part, some embodiments of the present application provide a cloud transmission method for MRI image information, comprising the following steps:

[0007] Step 1: The terminal device generates the original information that needs to be sent, and converts the original information into encrypted information;

[0008] Step 2: The terminal device sends the encrypted information and the domain name information of the terminal device to the cloud server.

[0009] Step 3: The cloud server receives the encrypted information and domain name information of the terminal device, verifies the source of the domain name information, saves the encrypted information that passes the verification, and delivers the decryption key of the corresponding level according to the number of encrypted information uploaded by the terminal device.

[0010] Step 4: The terminal device downloads the encrypted information from the cloud server according to its own decryption key, and then decrypts the encrypted information into original information using the decryption key.

[0011] This scheme builds a secure and efficient transmission framework for cross-institutional sharing of sensitive MRI image data through encrypted transmission, source authentication, and hierarchical key mechanism: the original data is transmitted after encryption, effectively preventing information leakage during transmission; the cloud strictly verifies the domain name information to ensure the legality and traceability of the data source; the design of dynamically allocating decryption key levels according to the amount of uploaded data not only encourages institutions to actively contribute sufficient data to build high-quality training sets, but also effectively restricts the ability of a single institution (especially small institutions) to decrypt and leak large amounts of original sensitive information, significantly reducing the risk of patient privacy leakage, laying a solid foundation for safe and reliable medical AI model collaboration training.

[0012] Further,

[0013] Step 1 includes the following steps:

[0014] Step 11: Divide the original information into picture information and N levels of annotation information, where N is a positive integer;

[0015] Step 12: Extract N levels of annotation information from the picture information, and process the annotation information as key information;

[0016] Step 13: Assign a unique key information to different levels of annotation information, encrypt the corresponding annotation information with the key information, and generate encrypted labels;

[0017] Step 14: All encrypted labels and key information are used as encrypted information.

[0018] This scheme significantly improves efficiency while ensuring core privacy security through separation processing and hierarchical encryption mechanism: only N levels of annotation information containing sensitive patient information (such as identity and medical history) are encrypted, while the basic image information remains unencrypted, which greatly reduces the amount of data that needs to be encrypted and optimizes processing and transmission efficiency; At the same time, the unique key generated from the features extracted from the image is used to encrypt the corresponding level of annotation, ensuring that even if the unencrypted image is obtained, malicious individuals cannot associate it with a specific patient identity or case, effectively preventing malicious use of image information and patient privacy leakage.

[0019] Further, the N levels of annotation information at least include 3 levels;

[0020] The first level is age, gender, and basic medical history of the patient corresponding to the picture information;

[0021] The second level is disease history and examination information of the patient corresponding to the picture information;

[0022] The third level is radiotherapy target area annotation information and radiotherapy treatment information corresponding to the picture information.

[0023] The present scheme realizes differentiated protection of sensitive data (the higher the value and risk of the data, the higher the encryption strength) by subdividing the annotation information into at least three security levels (basic demographic and medical history, detailed disease and examination information, and core radiotherapy target area and treatment data) and performing hierarchical processing.

[0024] Further, the calibration information is the average value of the pixel values of all pixel points in the preset region in the picture information.

[0025] The present scheme significantly enhances the security and anti-collision ability of the encryption system by generating a unique key based on the image content: the calibration information is defined as the average value of all pixel points in the preset region in the picture, and by flexibly configuring the preset region (such as position, size, and number), a variable number of unique calibration values can be dynamically generated for each picture. These calibration values derived from the content of the image itself are converted into exclusive key information for each picture after processing, ensuring that the encryption keys of different pictures and different annotation levels are all different. This design fundamentally eliminates the risk of key repetition or regularity derivation, greatly improving the anti-attack ability and overall security of the encryption system.

[0026] Further, the preset region is each organ target area in the picture information from top to bottom and from left to right; when the number of organ target areas in the picture information is less than 3, the target area is divided into 3 parts along the length direction of the target area.

[0027] The present scheme anchors the key organ target area and implements intelligent region division (selects the target area in spatial order, and evenly divides along the long axis when the number is insufficient), ensures that the key generation is closely related to the clinical semantics, and guarantees the stability of the number of preset regions (≥3); this design fully utilizes the natural spatial position and unique shape of the organ target area, maximizes the difference of the calibration information between different images, and thus generates highly unique and unpredictable keys, significantly improving the security, anti-collision ability, and robustness of the encryption system to non-target area image disturbance.

[0028] Step 13 includes the following steps:

[0029] Step 131: Convert the annotation information into binary code with an encoder, and then convert the binary code into an integer M;

[0030] Step 132: Encrypt the integer M into Me with secret key information;

[0031] Step 133: Re-convert Me into binary code, and then convert the binary code into a trigonometric function curve x j (t), j represents the index of binary code, ;

[0032] where b j represents the value of the corresponding bit in the binary code, j represents the index of the binary code, f represents the carrier frequency, and t is the time variable, and the trigonometric function curve is used as an encrypted label.

[0033] The scheme significantly improves the confidentiality and cracking resistance of the annotation information through a multi-level asymmetric encryption and information form conversion mechanism: after encoding the annotation information into an integer M, Me is generated by encryption using a private key, and then Me is reconstructed into a binary sequence and mapped into a unique trigonometric function curve x j (t) as an encrypted label. This process completely converts structured sensitive data (such as patient history, target area annotation) into abstract, continuous time-varying signals, not only completely concealing the original semantic features, but also greatly improving the ability to resist reverse engineering and brute force cracking through the double barriers of asymmetric encryption algorithm and function transformation, providing end-to-end strong security protection for high-value medical annotation data.

[0034] Further, step 14 includes the following steps:

[0035] Step 141: Obtain all encrypted labels, and divide all encrypted labels into several information sets;

[0036] Step 142: Add different noise information to the encrypted labels in different information sets;

[0037] Step 143: Use the encrypted labels with added noise information and secret key information as encrypted information.

[0038] The scheme significantly enhances the anti-correlation analysis and anti-batch cracking ability of the encrypted information through a differential noise injection mechanism: after aggregating the encrypted labels, they are grouped (information sets) and injected with unique noise for each set, and finally the encrypted labels with differential noise and secret key information are output. This design artificially introduces non-uniformity and randomness of the encrypted results, ensuring that even if part of the encrypted information is intercepted, it cannot be associated or batch cracked with other encrypted labels through pattern analysis, effectively dispersing the information leakage risk and providing an additional protection layer for the overall security of multi-level annotation data.

[0039] The generation of the noise information comprises the following steps:

[0040] S1: using a random algorithm with a budget, generating a random sequence G from the original picture information;

[0041] S2: setting a noise intensity a, generating a Gaussian noise sequence Y;

[0042] S3: uniformly inserting the Gaussian noise sequence Y into the encrypted label.

[0043] Further, step 3 comprises the following steps:

[0044] Step 31: the cloud server receives the encrypted information and the domain name information of the terminal device, checks whether the domain name information is correct, and if correct, saves the encrypted information, otherwise deletes the encrypted information;

[0045] Step 32: the cloud server extracts the secret key information in the encrypted information and saves the secret key information to the secret key database;

[0046] Step 33: the cloud server extracts the encrypted label in the encrypted information, arranges the encrypted label to obtain a label string, then defines an initial encryption key for the first encrypted label, and generates a following key using the serial number of the subsequent encrypted label and the initial encryption key of the previous one, and encrypts the encrypted label using the following key to generate a second encrypted label;

[0047] Step 34: the cloud server issues the corresponding following key according to the number of encrypted information uploaded by the terminal device.

[0048] The scheme realizes fine-grained data access control and system load optimization through the cooperative mechanism of cloud domain name verification, secret key separation storage, chain secondary encryption, and dynamic key issuance according to upload volume: after ensuring the legal identity of the terminal, the chain encryption structure (generating a following key based on the serial number and the previous key) combined with the design of strictly binding the key issuance volume and the terminal upload data volume accurately limits the range of data that the terminal can decrypt (uploading less decrypts permission), effectively preventing data abuse; at the same time, this mechanism allows the terminal to pre-download encrypted labels and operate offline, and only obtains the key decryption as needed after submitting the encrypted information, which significantly reduces the real-time request pressure and network congestion of the cloud, and avoids the transmission of original data throughout the process, greatly improving the overall security and efficiency.

[0049] Further, the encrypted labels in the label string are sorted according to the upload time; the initial encryption key is dynamically updated, and each initial key corresponds to a preset number of label strings.

[0050] In the scheme, the initial encryption key is dynamically updated, which can reduce the risk of analyzing the initial encryption key and increase the information security.

[0051] Further, the key information corresponding to the terminal device downloaded in the key database needs the authorization information of the terminal device.

[0052] In the scheme, by setting a key for the terminal device and the cloud server, the disclosure range of the information can be controlled by both, and the patient information is avoided to be disclosed to the untrusted field.

[0053] Step 4 includes the following steps:

[0054] Step 41: the terminal device downloads all the secondary encrypted labels from the cloud server, and then uses the following key provided by the cloud server to sequentially restore the corresponding following key;

[0055] Step 42: the obtained following key is corresponding to the secondary encrypted label, and the secondary encrypted label is decrypted to obtain the encrypted label, and then the noise information in the encrypted label is removed;

[0056] Step 43: download the key information corresponding to the encrypted label from the cloud server, and use the key information to parse the encrypted label to obtain the required label information.

[0057] In the scheme provided by the application, although the original information is encrypted twice, but the encryption object is the label information, the information amount contained in the standard information itself is small, and the length of the information string is short, and the encryption difficulty and the encryption time will not be increased after two times of encryption, in practice, the picture information and the standard information can be transmitted in two channels, so that the transmission time + decryption time of the standard information is basically equal to the download time of the picture information. BRIEF DESCRIPTION OF DRAWINGS

[0058] The accompanying drawings, which form a part of this application, are included to provide a further understanding of the application and are incorporated in and constitute a part of this application. The drawings illustrate embodiments of the present application and, together with the description, serve to explain the principles of the present application.

[0059] In addition, throughout the drawings, the same or similar reference numerals refer to the same or similar elements. It should be understood that the drawings are schematic, and the elements and elements are not necessarily drawn according to the proportion.

[0060] In the drawings:

[0061] Figure 1 The flow chart for the cloud transmission method of MRI image information.

[0062] Figure 2 The structure diagram of the cloud server and the terminal device.

[0063] Figure 3 The schematic diagram of the secondary encrypted label generation. Detailed Implementation

[0064] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While some embodiments of this application are shown in the drawings, it should be understood that this application can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this application. It should be understood that the drawings and embodiments of this application are for illustrative purposes only and are not intended to limit the scope of protection of this application.

[0065] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0066] The present application will now be described in detail with reference to the accompanying drawings and embodiments.

[0067] refer to Figure 1 and Figure 2 Example 1: A cloud transmission method for MRI image information includes the following steps:

[0068] Step 1: The terminal device generates the original information to be sent and converts the original information into encrypted information.

[0069] Step 11: Divide the original information into image information and N levels of annotation information, where N is a positive integer.

[0070] The labeling information for N levels must include at least the following three levels:

[0071] The first level is the patient's age, gender, and basic medical history corresponding to the image information;

[0072] The second level consists of the patient's medical history and examination information corresponding to the image information;

[0073] The third level consists of radiotherapy target area annotation information and radiotherapy treatment information corresponding to the image information.

[0074] MRI images are generally used for radiotherapy, and therefore contain a lot of additional textual information in addition to the image data. This includes the patient's age, gender, basic medical history (hypertension, hyperglycemia, hyperlipidemia, and other underlying diseases), disease history (tumor stage, size, and diffusion status), examination information (biopsy information, information from various examinations), standard information on the radiotherapy target area (the extent of the radiotherapy target area), and radiotherapy treatment information (treatment effect, degree of tumor reduction), etc.

[0075] The above information has little reference value for individual patients. However, it plays a great role in the training of an AI model. For example, training an AI model with the target range of radiotherapy, the AI model can automatically generate a target. It can be predicted that the more data input into the AI model, the more comprehensive the data, and the more the AI model can learn the delineation method of the radiotherapy target of the good radiotherapy plan.

[0076] Based on this, the original information is divided into picture information and standard information. The picture information is actually the MRI image information obtained by the patient during radiotherapy. These individual image information has little value in the medical field and can at most be used to study the common burst position of different tumors, but this is not enough in clinical application, and it is difficult to determine who the patient is by leaking individual image information. Therefore, the picture information does not participate in the encryption process, but only labels the picture information and the label information to bind the picture information and the standard information together.

[0077] Different levels of label information represent different information importance, and relatively more important information will have more complex encryption algorithm.

[0078] Step 12: Extract N calibration information from the picture information, and process the calibration information as secret key information.

[0079] The most difficult part of encrypting information is how to obtain the random factor of the secret key. Based on this, the calibration information is extracted from the picture information, and the calibration information is used as the random factor. The calibration information is the average value of the pixel values of all pixel points in the preset region in the picture information. The preset region is each organ target area in the picture information from top to bottom and from left to right; when the number of organ target areas in the picture information is less than 3, the target area is divided into 3 parts along the length direction of the target area. For example, there are 3 organs in an MRI picture information, and the organ in the upper left corner is used as the preset region. Then calculate the average value of the pixel values of all pixel points in the preset region, and use the average value to generate secret key information.

[0080] Step 13: Assign a unique secret key information to different levels of label information, encrypt the corresponding label information with the secret key information, and generate an encrypted label. In step 12, N secret key information is generated, so each different level of label information can correspond to a unique secret key information, and the specific encryption method is as follows.

[0081] Step 13 includes the following steps:

[0082] Step 131: Convert the label information into binary code using an encoder, and then convert the binary code into an integer M.

[0083] The annotation information is text information, and therefore needs to be converted into binary code by an encoder. The encoder can be understood as a pre-trained dictionary, and according to the index directory of the text information in the dictionary, binary code is generated. After obtaining the binary code, the encoder corresponding decoder can be translated into the annotation information.

[0084] Step 132: encrypt the integer M into Me using the key information;

[0085] Step 133: convert Me into binary code again, and then convert the binary code into a trigonometric function curve x j (t), j represents the index of the binary code,

[0086] where b j represents the value of the corresponding bit in the binary code, j represents the index of the binary code, f represents the carrier frequency, and t is a time variable. The trigonometric function curve is used as an encryption label.

[0087] Step 14: all encryption labels and key information are used as encryption information.

[0088] Step 14 includes the following steps:

[0089] Step 141: obtain all encryption labels, and divide all encryption labels into a plurality of information sets;

[0090] Step 142: for encryption labels in different information sets, different noise information is added;

[0091] The generation method of the noise information includes the following steps:

[0092] S1: using a pre-budget random algorithm, a random number sequence G is generated using original picture information;

[0093] S2: set the noise intensity a, and generate a Gaussian noise sequence Y;

[0094] S3: uniformly insert the Gaussian noise sequence Y into the encryption label.

[0095] The noise information is mainly used to increase the randomness of the information. Relatively speaking, the homogeneity of the annotation information is very serious (a large number of same Chinese characters and same descriptions), which can cause the similarity of the encrypted information to be high, and the key information to be easily cracked. By increasing the noise information, the homogeneity problem of the encryption label can be reduced.

[0096] Step 143: the encryption label to which the noise information is added and the key information are used as encryption information.

[0097] Step 2: the terminal device sends the encryption information and the domain name information of the terminal device to the cloud server.​

[0098] The domain name information of the terminal device is essentially the IP address of the terminal device, and checking the IP address can determine whether the terminal device is a device that joins the alliance, thereby avoiding malicious upload of false information by illegal persons and polluting the database.

[0099] Step 3: The cloud server receives the encrypted information and the domain name information of the terminal device, checks the source of the domain name information, saves the encrypted information that passes the check, and delivers the decryption key of the corresponding level according to the number of encrypted information uploaded by the terminal device.

[0100] Step 3 includes the following steps:

[0101] Step 31: The cloud server receives the encrypted information and the domain name information of the terminal device, checks whether the domain name information is correct, and saves the encrypted information if it is correct, or deletes the encrypted information if it is not.

[0102] Step 32: The cloud server extracts the key information in the encrypted information and saves the key information to the key database.

[0103] The key database saves the key information, and therefore after obtaining the encrypted information, the corresponding key information needs to be downloaded from the key database to parse the encrypted information. In practice, the key information is a private key generated by asymmetric encryption, and the public key of the terminal device is saved in the key database. In this way, using the public key information, the source of the encrypted information can be determined, so that the terminal device that downloads the encrypted information can accurately determine the source of the information, and then train the AI model accordingly.

[0104] Step 33: The cloud server extracts the encrypted label in the encrypted information, arranges the encrypted labels to obtain a label string, then defines an initial encryption key for the first encrypted label, and uses the serial number of the subsequent encrypted label and the initial encryption key of the previous one to generate a following key, and uses the following key to encrypt the encrypted label to generate a secondary encrypted label.

[0105] The key of step 33 is that the cloud server also encrypts the annotation information, and the reason why the cloud server encrypts is to let as many downloaded terminal devices as possible upload annotation information and increase the diversity of their own databases. Therefore, the cloud server uses a chain encryption method.

[0106] For example, the cloud server saves 10 encryption labels, D1~D10. The cloud server generates an initial encryption key K1 for D1, and then generates a follow-up key by compounding the initial encryption key and the serial number 2. That is, the serial number of each encryption label is compounded with the follow-up key of the previous encryption label to generate a follow-up key, and then the encryption label is encrypted with the follow-up key to obtain a secondary encryption label.

[0107] Thus, when decrypting these secondary encryption labels, the degree of decryption can be controlled by controlling the given encryption key. For example, if the follow-up key K5 of the 5th encryption label is given, K5~K10 can be restored using K5, and then D5~D10 can be decrypted. Furthermore, the scheme provided in the present application allows the cloud server to control the degree of information opening at a lower encryption complexity.

[0108] Step 34: The cloud server issues the corresponding follow-up key according to the number of encryption information uploaded by the terminal device.

[0109] For example, if a terminal device only uploads 10 encryption labels, the cloud server can only give it the follow-up key with serial number K9, and then the terminal device can only restore two secondary encryption labels.

[0110] Step 4: The terminal device downloads the encryption information from the cloud server according to its decryption key, and then decrypts the encryption information into original information using the decryption key.

[0111] Step 4 includes the following steps:

[0112] Step 41: The terminal device downloads all secondary encryption labels from the cloud server, and then uses the follow-up key provided by the cloud server to restore the corresponding follow-up key in sequence;

[0113] Step 42: The obtained follow-up key is matched with the secondary encryption label, and the secondary encryption label is decrypted to obtain the encryption label, and then the noise information in the encryption label is removed.

[0114] Step 43: Download the key information corresponding to the encryption label from the cloud server, and parse the encryption label using the key information to obtain the required label information.

[0115] Embodiment 2: Embodiment 2 provides a specific encryption method of key information for label information on the basis of Embodiment 1.

[0116] The encryption information includes an invertible residual network model and model parameters.

[0117] The invertible residual network model includes:

[0118] reversible conversion module, converting the annotation information into first features and second features ;

[0119] The reversible conversion module mainly converts one-dimensional encoding information into two-dimensional features.

[0120] reversible residual block, reversibly transforming the first features and second features ;

[0121] ;

[0122] wherein, , , represents a residual function, is an activation function, is an element-wise correspondence.

[0123] output module, merging y1 and y2 according to the channel to obtain an encrypted label.

[0124] When information decryption is needed, only the encrypted label needs to be input in reverse to the reversible residual network model to restore the corresponding information.

[0125] The key of the scheme is to use a reversible neural network for encryption. In fact, the network structure is not further limited, as long as the network structure has a reversible function.

[0126] Therefore, when saving the secret key information, the cloud server only needs to save the model parameters, without saving additional model information. This encryption method is beneficial to the conventional scheme, and it is difficult to leak information.

[0127] Further, the secret key information corresponding to the terminal device downloaded from the secret key database needs the authorization information of the terminal device.

[0128] In practice, the terminal device can save its own model structure, and other users who need to obtain their own secret key information need to obtain the model structure from themselves, and then download the model parameters from the cloud server to further decrypt the information.

[0129] The specific decryption process is as follows:

[0130] T1: Obtain the trigonometric function curve, extract the amplitude of the corresponding position according to the carrier frequency of the trigonometric function curve, and generate binary encoding;

[0131] T2: convert the binary encoding into Me, then input Me into the reversible residual network model to generate new binary encoding, and then return the binary encoding to the original Wie annotation information.

[0132] refer to Figure 3 Example 3: Example 3 provides a method for generating the follow key based on Example 1.

[0133] Specifically, the encryption process is as follows:

[0134] S1: Pre-set the HMAC-SHA256 algorithm and initial encryption key, the initial encryption key K1, which is a 128-bit random number.

[0135] S2: Combine the encryption key K1 and the integer 2, then process them using the HMAC-SHA256 algorithm to generate the following key K2; combine the encryption key K2 and the integer 3, then process them using the HMAC-SHA256 algorithm to generate the following key K3; and so on, to generate the following key K. N N represents the total number of keys that follow.

[0136] S3: Obtain all encrypted tags, sort the encrypted tags according to the upload time to obtain the tag string, then match the sequence number of the tag string with the sequence number of the accompanying key, and then encrypt the encrypted tag with the accompanying key to obtain the secondary encrypted tag.

[0137] The decryption process is as follows:

[0138] Z1: Obtain the encryption key K provided by the cloud server G If G < N, then all encryption keys K are gradually reconstructed using the HMAC-SHA256 algorithm. G ~K N ;

[0139] Z2: Will be used with encryption key K G ~K N The corresponding secondary encryption tags are extracted and then decrypted to generate encrypted tags.

[0140] The above scheme allows the latest encryption key to be obtained costlessly using historical encryption keys. Based on this:

[0141] The initial encryption key is dynamically updated, and each initial key corresponds to a preset number of tag strings.

[0142] For example, the initial encryption key K1 can generate nine follow-up keys, namely K1~K. 10 All encryption tags are generated using the initial encryption key K1 in a chain, which can then encrypt 10 encryption tags. However, when encrypting the 11th encryption tag, a new encryption tag KK1 is generated, which will only generate 9 follow-up keys. Thus, if the latest information needs to be continuously obtained, the encrypted information needs to be continuously uploaded to the cloud server.

[0143] The above description is merely exemplary of the application and the application principles of the technology used. It is to be understood that the application scope of the embodiments of the application is not limited to the technical solutions formed by the specific combinations of the above technical features, and should also cover other technical solutions formed by the combinations of the above technical features or equivalent features thereof without departing from the above inventive concept. For example, the technical solutions formed by the mutual replacement of the above features and the technical features disclosed in the embodiments of the application (but not limited to) having similar functions.

Claims

1. A cloud-based transmission method for MRI image information, characterized in that, The method comprises the following steps: Step 1: the terminal device generates original information to be sent, and converts the original information into encrypted information; Step 2: the terminal device sends the encrypted information and domain name information of the terminal device to a cloud server; Step 3: the cloud server receives the encrypted information and the domain name information of the terminal device, checks the source of the domain name information, saves the encrypted information that passes the check, and delivers a decryption key of a corresponding level according to the number of encrypted information uploaded by the terminal device; Step 4: the terminal device downloads the encrypted information from the cloud server according to the decryption key of the terminal device, and then decrypts the encrypted information into original information by using the decryption key; Step 1 comprises the following steps: Step 11: dividing the original information into picture information and N levels of labeling information, wherein N is a positive integer; Step 12: extracting N pieces of labeling information from the picture information, and processing the labeling information into key information; Step 13: assigning a unique key information to labeling information of different levels, encrypting the corresponding labeling information by using the key information, and generating encrypted labels; Step 14: taking all the encrypted labels and the key information as encrypted information; Step 3 comprises the following steps: Step 31: the cloud server receives the encrypted information and the domain name information of the terminal device, checks whether the domain name information is correct, and saves the encrypted information if the domain name information is correct, or deletes the encrypted information if the domain name information is incorrect; Step 32: the cloud server extracts the key information in the encrypted information, and saves the key information into a key database; Step 33: the cloud server extracts the encrypted labels in the encrypted information, arranges the encrypted labels to obtain a label string, defines an initial encryption key for the first encrypted label, generates a following key by using the serial number of a subsequent encrypted label and the initial encryption key of the previous encrypted label, encrypts the encrypted label by using the following key to generate a second encrypted label; Step 34: the cloud server delivers a corresponding following key according to the number of encrypted information uploaded by the terminal device.

2. The cloud transfer method for MRI image information according to claim 1, characterized in that: The N levels of labeling information at least comprise the following three levels: The first level is the age, gender and basic medical history of a patient corresponding to the picture information; The second level is the disease history and examination information of the patient corresponding to the picture information; The third level is the radiotherapy target area labeling information and radiotherapy treatment information corresponding to the picture information.

3. The cloud transfer method for MRI image information according to claim 1, characterized in that: The labeling information is the average value of pixel values of all pixel points in a preset region in the picture information.

4. The cloud transmission method for MRI image information according to claim 3, wherein: The preset region is each organ target area in a direction from top to bottom and from left to right in the picture information; when the number of organ target areas in the picture information is less than 3, the target area is divided into three parts along the length direction of the target area.

5. The cloud transfer method for MRI image information as claimed in claim 1 wherein: Step 13 comprises the following steps: Step 13 comprises the following steps: Step 131: converting the labeling information into binary code by using an encoder, and then converting the binary code into an integer M; Step 132: encrypting the integer M into Me by using the key information; Step 133: reconvert Me into a binary code and then convert the binary code into a trigonometric curve , j represents an index of the binary code, ; where b j represents the value of the corresponding bit in the binary encoding, j represents the index of the binary encoding, f represents the carrier frequency, t is the time variable, and the trigonometric function curve is used as the encryption tag.

6. The cloud transfer method for MRI image information according to claim 1, characterized in that: Step 14 comprises the following steps: Step 141: obtaining all the encrypted labels, and dividing all the encrypted labels into a plurality of information sets; Step 142: adding different noise information to the encrypted labels in different information sets; Step 143: add the encrypted label with noise information and the key information as encrypted information.

7. The cloud transfer method for MRI image information according to claim 6, characterized in that: The generation method of the noise information includes the following steps: S1: using a budgeted random algorithm, generating a random number sequence G using the original picture information; S2: setting the noise intensity a, generating a Gaussian noise sequence Y; S3: uniformly inserting the Gaussian noise sequence Y into the encrypted label.

8. The cloud transmission method for MRI image information according to claim 1, characterized in that: Step 4 includes the following steps: Step 41: the terminal device downloads all the secondary encrypted labels from the cloud server, and then uses the following key provided by the cloud server to sequentially restore the corresponding following key; Step 42: correspond the obtained following key with the secondary encrypted label, and decrypt the secondary encrypted label to obtain the encrypted label, and then remove the noise information in the encrypted label; Step 43: download the key information corresponding to the encrypted label from the cloud server, and use the key information to parse the encrypted label to obtain the required label information.

Images