CN115879126B - Medical information sharing method based on secure cloud storage, electronic equipment and storage medium - Google Patents

Abstract

The present invention discloses a medical information sharing method, electronic device and storage medium based on secure cloud storage, the method steps include: 1. The hospital generates the patient's medical information; 2. Encrypts the patient's privacy information; 3. Embeds the patient's medical information into a medical image; 4. Uploads the encrypted information to the cloud storage system; 5. The hospital retrieves relevant medical information from the cloud storage system; 6. Medical information extraction and data recovery; 7. Patient identification and information decryption. The present invention can encrypt the patient's personal information, hide the medical records, use deep learning to retrieve medical information and match images, and share medical information without leaking the patient's privacy information.

Description

Medical information sharing method, electronic device and storage medium based on secure cloud storage

Technical Field

The present invention relates to the technical field of medical information cloud sharing and information hiding, and specifically to a medical information sharing method, electronic device and storage medium based on secure cloud storage.

Background technique

With the development of cloud computing technology, cloud storage and sharing services for data have received widespread attention. In the medical field, traditional medical images are in the form of film, which has some problems: 1. The imaging effect is affected by the quality of the film; 2. It needs to be stored in a suitable environment and occupies a large space; 3. It is not convenient to share, etc. It is bound to become a trend for modern medical care to adopt digital imaging and cloud storage technology. Storing electronic medical images in a database system greatly facilitates management and solves the problem of storing massive image information. Storing medical image information in a cloud storage system realizes the digitization of medical images and facilitates information sharing.

However, cloud storage and information sharing may have privacy leakage issues. Medical images usually contain sensitive information such as the patient's identity. If the patient's medical images and information are directly uploaded to the cloud storage system, there will be privacy security issues. If there is an untrusted third-party server that leaks the patient's privacy information for commercial interests, or the party that obtains the shared information uses the patient's privacy information to gain benefits, it will cause trouble to the patient. The traditional cloud storage security method is cryptographic encryption, which encrypts the image to generate ciphertext. Only the key owner can decrypt it and view the image information after decryption, which is not conducive to information sharing. Although the encryption method provides a strong security guarantee, the cumbersome encryption and decryption as well as the data transmission fee increase the cost.

Summary of the invention

The present invention aims to address the deficiencies of the above-mentioned prior art and proposes a medical information sharing mechanism method, electronic device and storage medium based on secure cloud storage, in order to provide privacy protection for cloud storage data in the medical field through information hiding technology, thereby ensuring the security of data in the cloud storage system while sharing medical information.

In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical scheme:

The medical information sharing method based on secure cloud storage of the present invention is characterized in that it is applied to a medical environment consisting of n hospitals H = {h ₁ ,h ₂ ,..., _hi ,..., _hn }, m patients P = {p ₁ ,p ₂ ,...,p _j ,...,p _m } and a cloud storage system, wherein _hi represents the i-th hospital, 1≤i≤n; p _j represents the j-th patient, 1≤j≤m; the medical information sharing method is performed in the following steps:

Step 1: The hospital generates the patient's medical information, including medical history information and medical records:

Step 1.1: Generate the patient's medical record information:

Assume that the jth patient p _j visits the ith hospital _hi for treatment. The ith hospital _hi obtains the personal identity information ID _j and medical image I _j of the jth patient p _j and generates an electronic medical record r _j , thereby constructing the medical record information R _j = (ID _j , r _j ) of the jth patient p _j ;

Step 1.2: Generate the patient's medical records:

Step 1.2.1, after the i-th hospital h _i treats the j-th patient p _j, a medical record of the j-th patient p _j is generated, wherein the medical record includes a treatment plan and a treatment record;

Step 1.2.2, parsing the medical record of the jth patient p _j to generate text information MR _j of the medical record;

Step 2: Encrypt patient medical records:

Step 2.1, generate the encryption key key _j and identity token token _j of the jth patient p _j , and write them into the medical card of the jth patient p _j ;

Step 2.2: Use the AES encryption algorithm to encrypt the medical record information _Rj of the jth patient _pj to obtain the encrypted medical record information ciphertext _Rsj ;

Step 3: Embed the encrypted medical record information ciphertext _Rsj as secret information into the medical image _Ij to obtain a stego-image containing secret information.

Step 3.1: Decompose the bit plane of medical image _Ij into HSB bit planes and LSB bit plane make Indicates the HSB bit plane The pixel value of the uth row and vth column in , Indicates the LSB bit plane The pixel value at the uth row and vth column in ;

Step 3.2: HSB bit plane The pixels that may overflow are marked to establish the position map LM _j :

Set the maximum threshold T _max and the minimum threshold T _min ;

like or It represents the HSB bit plane The pixel value in the uth row and vth column overflows, and the HSB bit plane The position at row u and column v in is marked as "1";

like or It represents the HSB bit plane The pixel value in the uth row and vth column overflows, and the HSB bit plane The position at row u and column v in is marked as "2";

Otherwise, it indicates the HSB bit plane. The pixel value in the uth row and vth column in the image does not overflow, and the HSB bit plane The position of the uth row and the vth column in is marked as "0", thus obtaining the HSB bit plane The corresponding position map LM _j ;

Step 3.3, modify the HSB bit plane The pixel value that may overflow, thus obtaining the processed HSB bit plane

like The HSB bit plane The pixel value at row u and column v in Subtract "2";

like The HSB bit plane The pixel value at row u and column v in Add "2";

like The HSB bit plane The pixel value at row u and column v in Subtract "1";

like The HSB bit plane The pixel value at row u and column v in plus 1";

Step 3.4: Use arithmetic coding to perform lossless compression on the position map LM _j to obtain the compressed position map

The processed HSB bit plane Except for the first row of pixels, the first column of pixels, and the last row and the last column, the remaining pixels are divided into checkerboards;

Step 3.5: Replace the pixel value of the u′th row and v′th column in the chessboard The adjacent 8 pixel values are arranged in ascending order, and the average of the first six pixel values among the sorted pixel values is calculated and then rounded up to be recorded as the predicted value q ₁ (u′, v′) of the u′th row and v′th column; the average of the last six pixel values among the sorted pixel values is calculated and then rounded up to be recorded as the predicted value q ₂ (u′, v′) of the u′th row and v′th column;

Step 3.6: Use the prediction error expansion method to embed each pixel in the chessboard twice to obtain a new HSB bit plane.

Step 3.6.1: Compress the location map Add to the end of the encrypted medical record information ciphertext Rs _j to obtain the new secret data Rs′ _j , and define the secret information as b and b′;

Step 3.6.2. Calculate the prediction error of the first embedding

Step 3.6.3: When e ₁ = 1, the pixel value of the u′th row and v′th column in the chessboard plus part of the secret information b;

When e ₁ = 0, the pixel value of the u′th row and v′th column in the chessboard Subtract part of the secret information b;

When e ₁ > 1, the pixel value of the u′th row and v′th column in the chessboard Increment by 1;

When e ₁ ＜0, the pixel value of the u′th row and v′th column in the chessboard Decrement by 1;

Thus, the pixel value of the u′th row and v′th column after the first embedding is obtained.

Step 3.6.4. Calculate the prediction error of the second embedding

When e ₂ = 1, the pixel value of the u′th row and v′th column Add part of the secret information b′;

When e ₂ = 0, the pixel value of the u′th row and v′th column Subtract part of the secret information b′;

When e ₂ > 1, the pixel value of the u′th row and v′th column Increment by 1;

When e ₂ ＜0, the pixel value of the u′th row and v′th column Decrement by 1;

Thus, the pixel value of the u′th row and v′th column after the second embedding is obtained

Step 3.7: Set the new HSB bit plane and LSB bit plane The combination obtains the stego-image containing the new secret data Rs′ _j

Step 4: Merge medical information and upload to cloud storage system:

Step 4.1: Merge the text information MR _j , the identification token token _j and the steganagraph image of the medical record of the jth patient p _j Form the complete encrypted medical information of the jth patient p _j

Step 4.2, uploading the complete encrypted medical information M _j to the cloud storage system;

Step 5: Retrieve medical information of similar cases from the cloud storage system:

Step 5.1: Use deep learning models for text matching and information retrieval:

Step 5.1.1, according to the condition of the jth patient _pj , input the symptom information of the condition to be queried;

Step 5.1.2: Analyze the disease symptom information using a deep learning model and extract semantic feature vectors;

Step 5.1.3, calculate the feature similarity between the semantic feature vector and the medical record information in the cloud storage system, and generate a text similarity descending sorting result;

Step 5.1.4, sort the results according to the descending similarity of the texts, so as to obtain _the text information set MR _j = {MR _1,j ,MR _2,j ,...,MR _k,j ,...,MR _r,j } of the cases similar to the j-th patient p j, where MR _k,j represents the k-th text information of the case similar to the j-th patient p _j , 1≤k≤r; r represents the total number of similar cases;

Step 5.2: Use the deep learning model to perform feature extraction and feature retrieval on medical image _Ij :

Step 5.2.1, input the medical image _Ij of the jth patient _pj into the deep learning model for feature extraction, obtain the features of the lesion area in the medical image _Ij and save it as a lesion feature vector;

Step 5.2.2, calculate the image feature similarity between the lesion feature vector of the medical image _Ij of the jth patient _pj and the lesion feature vector of the medical image in the cloud storage system, and generate a descending sorting result of the image similarity;

Step 5.2.3, according to the results of descending sorting of image similarity, obtain the medical image set I _j = {I _1,j ,I _2,j ,...,I _d,j ,...,I _{r,j }} of the case similar to the j-th patient p j, where I _d,j represents the d-th medical image of the case similar to the j-th patient p _j , 1≤d≤r;

Step 5.3: According to the personal identity information ID _j of the jth patient p _j , obtain the previous medical information of the jth patient p _j from the cloud storage system:

Use the medical card of the j-th patient p _j to authenticate at the terminal of the i-th hospital h _i , and obtain the encrypted medical information M _j of the j-th patient p _j from the cloud storage system according to the identity token token _j in the medical card;

Step 5.4: Based on the encrypted medical information M _j of the j-th patient p _j , obtain the encrypted medical information set M _j = {M _1,j ,M _2,j ,...,M _k,j ,...,M _r,j } of other similar cases, where M _k,j represents the encrypted medical information of the k-th patient of a case similar to the j-th patient p _j , 1≤k≤r;

Let the text information of the medical record in the encrypted medical information M _k,j be MR _k,j and the steganographic image be

Step 6: Steganographic Image Perform extraction and image recovery:

Step 6.1: Put the stego image Decompose into HSB bit planes and LSB bit plane

Step 6.2: HSB bit plane Each pixel in is extracted twice to obtain the HSB bit plane value after the second extraction And new secret data Rs _k ′ is composed of all the extracted secret information;

Step 6.2.1. The pixel value at row u and column v in Arrange the adjacent 8 pixels in ascending order, calculate the average of the first six pixel values in the sorted pixel values, round it up, and record it as the predicted value q ₁ ′(u,v) of the uth row and vth column; calculate the average of the last six pixel values in the sorted pixel values, round it up, and record it as the predicted value q ₂ ′(u,v) of the uth row and vth column;

Step 6.2.2: Calculate the prediction error of the first extraction When e ₂ ′＝1, 2, 0 or -1, Extract secret information from And get the HSB bit plane value after one extraction

Step 6.2.3. Calculate the prediction error of the second extraction When e ₁ ′＝1, 2, 0 or -1, Extract secret information from And get the HSB bit plane value after secondary extraction

Step 6.3: Separate the compressed location map from the new secret data Rs _k ′ and the medical record information ciphertext Rs _k ;

Step 6.4: Compress the compressed position map Decompress it and get the corresponding position map LM _{k .} Restore and then After the combination, the medical image I _k,j of the kth patient with a case similar to the jth patient p _j is obtained;

Step 7: Patient identification and information decryption:

Step 7.1, searching the cloud storage system for encrypted medical information with _the same identity token as the identity token held by the jth patient p _j ;

Step 7.2: Use the secret key key _j of the jth patient p _j and the AES algorithm to decrypt the ciphertext Rs _j of the medical record information of the jth patient p _j to obtain the medical record information R _j . Then decompose the medical record information R _j of the jth patient p _j to obtain the personal identity information ID _j and electronic medical record r _j of the jth patient p _j .

An electronic device of the present invention includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the medical information sharing method, and the processor is configured to execute the program stored in the memory.

The present invention provides a computer-readable storage medium, and the computer-readable storage medium stores a computer program, which is characterized in that when the computer program is run by a processor, the steps of the medical information sharing method are executed.

Compared with the prior art, the present invention has the following beneficial effects:

1. The present invention combines encryption technology with steganography technology to encrypt the patient's personal identity information to generate ciphertext, and then write the ciphertext and medical information into the medical image. By encrypting and steganographic information, double protection is provided for private information, thus avoiding the problem of privacy leakage caused by security risks in the cloud storage system.

2. The present invention performs dual authentication on the patient's identity. Only when the identity token matches can the patient's medical information be obtained. After obtaining the medical information, the secret key generated during encryption is required to decrypt the case information in the medical information. Through dual authentication, the privacy protection is strengthened and the problem of unauthorized parties maliciously obtaining private information is avoided.

3. The present invention uses deep learning to retrieve treatment record information and medical image feature retrieval. A well-trained deep learning model can extract the text semantic features of medical information and the lesion features of medical images. By retrieving text and images, similar medical information and images in the cloud storage system can be obtained, which is conducive to improving the work efficiency of doctors.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is an overall flow chart of the method of the present invention.

Detailed ways

In this embodiment, a medical information sharing method based on secure cloud storage is applied to a medical environment consisting of n hospitals H = {h ₁ ,h ₂ ,..., _hi ,..., _hn }, m patients P = {p ₁ ,p ₂ ,...,p _j ,...,p _m } and a cloud storage system, wherein _hi represents the i-th hospital, 1≤i≤n; p _j represents the j-th patient, 1≤j≤m;

The medical information sharing method based on secure cloud storage provides data security for medical information cloud sharing in the medical field. First, the patient's identity and private information in the medical record are encrypted, and a patient identity token is generated for patient identification. The encrypted information is then embedded in the medical image as secret data, and the treatment record and the steganographic image are uploaded to the cloud storage system. Finally, the doctor retrieves the medical information of similar cases from the cloud storage system, extracts the secret data and restores the medical image, identifies the patient's identity token, and decrypts the private information according to the patient's secret key. Specifically, as shown in Figure 1, the medical information sharing method is performed in the following steps:

Step 1: The hospital generates the patient's medical information, including medical history information and medical records:

Step 1.1: Generate the patient's medical record information:

Assume that the jth patient p _j visits the ith hospital _hi for treatment. The ith hospital _hi obtains the personal identity information ID _j and medical image I _j of the jth patient p _j and generates an electronic medical record r _j , thereby constructing the medical record information R _j = (ID _j , r _j ) of the jth patient p _j ;

Step 1.2: Generate the patient's medical records:

Step 1.2.1, after the i-th hospital h _i treats the j-th patient p _j, it generates the medical record of the j-th patient p _j , and the medical record includes the treatment plan and treatment record;

Step 1.2.2, parsing the medical record of the jth patient p _j to generate text information MR _j of the medical record;

Step 2: Encrypt patient medical records:

Step 2.1, generate the encryption key key _j and identity token token _j of the jth patient p _j , and write them into the medical card of the jth patient p _j ;

Step 2.2: Use the AES encryption algorithm to encrypt the medical record information _Rj of the jth patient _pj to obtain the encrypted medical record information ciphertext _Rsj ;

Step 3: Embed the encrypted medical record information ciphertext _Rsj as secret information into the medical image _Ij to obtain a steganographic image containing secret information.

Step 3.1: Decompose the bit plane of medical image _Ij into HSB bit planes and LSB bit plane make Indicates the HSB bit plane The pixel value of the uth row and vth column in , Indicates the LSB bit plane The pixel value at the uth row and vth column in ;

Step 3.2: HSB bit plane The pixels that may overflow are marked to build the position map LM _j :

Set the maximum threshold T _max and the minimum threshold T _min ;

like or It means HSB bit plane The pixel value in the uth row and vth column overflows, and the HSB bit plane The position at row u and column v in is marked as "1";

like or It represents the HSB bit plane The pixel value in the uth row and vth column overflows, and the HSB bit plane The position at row u and column v in is marked as "2";

Otherwise, it indicates the HSB bit plane. The pixel value in the uth row and vth column in the image does not overflow, and the HSB bit plane The position of the uth row and the vth column in is marked as "0", thus obtaining the HSB bit plane The corresponding position map LM _j ;

Because the HSB bit plane There are three cases of overflow and non-overflow of the pixel value in the uth row and vth column. The mark bit needs to be set to 0, 1, or 2. If binary compression is used, the mark needs to be represented by two bits, which results in too much auxiliary information. Therefore, using ternary representation can effectively reduce the length of the compressed position map.

Step 3.3, modify the HSB bit plane The pixel value that may overflow, thus obtaining the processed HSB bit plane

like The HSB bit plane The pixel value at row u and column v in Subtract "2";

like The HSB bit plane The pixel value at row u and column v in Add "2";

like The HSB bit plane The pixel value at row u and column v in Subtract "1";

like The HSB bit plane The pixel value at row u and column v in plus 1";

In other cases, keep the HSB bit plane The pixel value at row u and column v in constant;

Step 3.4: Use arithmetic coding to perform lossless compression on the position map LM _j to obtain the compressed position map

The processed HSB bit plane Except for the first row of pixels, the first column of pixels, and the last row and the last column, the remaining pixels are divided into checkerboards;

Step 3.5: Replace the pixel value of the u′th row and v′th column in the chessboard The adjacent 8 pixel values are arranged in ascending order, and the average of the first six pixel values among the sorted pixel values is calculated and then rounded up to be recorded as the predicted value q ₁ (u′, v′) of the u′th row and v′th column; the average of the last six pixel values among the sorted pixel values is calculated and then rounded up to be recorded as the predicted value q ₂ (u′, v′) of the u′th row and v′th column;

Step 3.6: Use the prediction error expansion method to embed each pixel in the chessboard twice to obtain a new HSB bit plane.

Step 3.6.1: Compress the location map Add to the end of the encrypted medical record information ciphertext Rs _j to obtain the new secret data Rs′ _j , and define the secret information as b and b′;

Step 3.6.2. Calculate the prediction error of the first embedding

Step 3.6.3: When e ₁ = 1, the pixel value of the u′th row and v′th column in the chessboard plus part of the secret information b;

When e ₁ = 0, the pixel value of the u′th row and v′th column in the chessboard Subtract part of the secret information b;

When e ₁ > 1, the pixel value of the u′th row and v′th column in the chessboard Increment by 1;

When e ₁ ＜0, the pixel value of the u′th row and v′th column in the chessboard Decrement by 1;

Thus, the pixel value of the u′th row and v′th column after the first embedding is obtained.

Step 3.6.4. Calculate the prediction error of the second embedding

When e ₂ = 1, the pixel value of the u′th row and v′th column Add part of the secret information b′;

When e ₂ = 0, the pixel value of the u′th row and v′th column Subtract part of the secret information b′;

When e ₂ > 1, the pixel value of the u′th row and v′th column Increment by 1;

When e ₂ ＜0, the pixel value of the u′th row and v′th column Decrement by 1;

Thus, the pixel value of the u′th row and v′th column after the second embedding is obtained

Step 3.7: Set the new HSB bit plane and LSB bit plane The combination obtains the stego-image containing the new secret data Rs′ _j

Step 4: Merge medical information and upload to cloud storage system:

Step 4.1: Merge the text information MR _j , the identification token token _j and the steganagraph image of the medical record of the jth patient p _j Form the complete encrypted medical information of the jth patient p _j

Step 4.2, upload the complete encrypted medical information _Mj to the cloud storage system;

The encrypted medical information _Mj is stored in the cloud storage system, which contains three elements: the text information of the medical record _MRj , the identity token _tokenj and the stego-image The patient's private data will not be directly exposed. The patient's private data only exists in the stego image. In the data, the patient's identity token can only be used to decrypt the data using the patient's secret key. Therefore, these three elements are stored in plain text, and doctors can retrieve them.

Step 5: Retrieve medical information of similar cases from the cloud storage system:

Step 5.1: Use deep learning models for text matching and information retrieval:

Step 5.1.1, according to the condition of the jth patient _pj , input the symptom information of the condition to be queried;

Step 5.1.2: Analyze the disease symptom information using a deep learning model and extract semantic feature vectors;

Use existing medical information data to pre-train the deep learning-based text matching and information retrieval model so that the model can correctly extract semantic feature vectors;

Step 5.1.3: Calculate the feature similarity between the semantic feature vector and the medical record information in the cloud storage system, and generate the text similarity descending sorting results:

In order to improve the retrieval efficiency, a pre-trained deep learning model is used to extract semantic features from the medical information data already in the cloud storage system, and a label index file for each piece of medical data is created in the cloud storage system. When performing a search, a candidate set is selected from the index file based on the semantic feature vector of the search text;

Step 5.1.4, sort the results according to the descending similarity of the texts, so as to obtain _the text information set MR _j = {MR _1,j ,MR _2,j ,...,MR _k,j ,...,MR _r,j } of the cases similar to the j-th patient p j, where MR _k,j represents the k-th text information of the case similar to the j-th patient p _j , 1≤k≤r; r represents the total number of similar cases;

Step 5.2: Use the deep learning model to perform feature extraction and feature retrieval on medical image _Ij :

Step 5.2.1, input the medical image _Ij of the jth patient _pj into the deep learning model for feature extraction, obtain the features of the lesion area in the medical image _Ij and save it as a lesion feature vector;

Step 5.2.2, calculate the image feature similarity between the lesion feature vector of the medical image _Ij of the jth patient _pj and the lesion feature vector of the medical image in the cloud storage system, and generate a descending sorting result of the image similarity;

Step 5.2.3, according to the results of descending sorting of image similarity, obtain the medical image set I _j = {I _1,j ,I _2,j ,...,I _d,j ,...,I _{r,j }} of the case similar to the j-th patient p j, where I _d,j represents the d-th medical image of the case similar to the j-th patient p _j , 1≤d≤r;

Step 5.3: According to the personal identity information ID _j of the jth patient p _j , obtain the previous medical information of the jth patient p _j from the cloud storage system:

Use the medical card of the jth patient p _j to authenticate at the terminal of the i-th hospital h _i , and obtain the encrypted medical information M _j of the j-th patient p _j from the cloud storage system based on the identity token token _j in the medical card;

Step 5.4: Based on the encrypted medical information M _j of the j-th patient p _j , obtain the encrypted medical information set M _j = {M _1,j ,M _2,j ,...,M _k,j ,...,M _r,j } of other similar cases, where M _k,j represents the encrypted medical information of the k-th patient of a case similar to the j-th patient p _j , 1≤k≤r;

Let the text information of the medical record in the encrypted medical information M _k,j be MR _k,j and the steganographic image be

Step 6: Steganographic Image Perform extraction and image recovery:

Step 6.1: Put the stego image Decompose into HSB bit planes and LSB bit plane

Step 6.2: HSB bit plane Each pixel in is extracted twice to obtain the HSB bit plane value after the second extraction And the new secret data Rs′ _k is composed of all the extracted secret information;

Since the prediction error expansion method is used to embed each pixel in the chessboard twice during the embedding process and the pixel value is changed, the reverse order must be followed during the extraction process to extract the correct prediction value and restore each pixel in the chessboard;

Step 6.2.1. The pixel value at row u and column v in Arrange the adjacent 8 pixels in ascending order, calculate the average of the first six pixel values in the sorted pixel values, round it up, and record it as the predicted value q ₁ ′(u,v) of the uth row and vth column; calculate the average of the last six pixel values in the sorted pixel values, round it up, and record it as the predicted value q ₂ ′(u,v) of the uth row and vth column;

Step 6.2.2: Calculate the prediction error of the first extraction When e ₂ ′＝1, 2, 0 or -1, Extract secret information from And get the HSB bit plane value after one extraction

Step 6.2.3. Calculate the prediction error of the second extraction When e ₁ ′＝1, 2, 0 or -1, Extract secret information from And get the HSB bit plane value after secondary extraction

Step 6.3: Separate the compressed position map from the new secret data Rs′ _k and the medical record information ciphertext Rs _k ;

Step 6.4: Compress the position map Decompress it and get the corresponding position map LM _{k .} Restore and then After the combination, the medical image I _k,j of the kth patient with a case similar to the jth patient p _j is obtained;

Step 7: Patient identification and information decryption:

Step 7.1, according to the identity token token _j held by the jth patient p _j, search the encrypted medical information with the same identity token from the cloud storage system;

Step 7.2: Use the secret key key _j of the jth patient p _j and the AES algorithm to decrypt the ciphertext Rs _j of the medical record information of the jth patient p _j to obtain the medical record information R _j . Then, decompose the medical record information R _j of the jth patient p _j to obtain the personal identity information ID _j and electronic medical record r _j of the jth patient p _j .

In this embodiment, an electronic device includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the above-mentioned medical information sharing method, and the processor is configured to execute the program stored in the memory.

In this embodiment, a computer-readable storage medium stores a computer program on the computer-readable storage medium, and the computer program executes the steps of the above-mentioned medical information sharing method when executed by a processor.

Claims (3)

1. The medical information sharing method based on the secure cloud storage is characterized by being applied to a medical environment consisting of n hospitals H= { H ₁,h₂,...,h_i,...,h_n }, m patients P= { P ₁,p₂,...,p_j,...,p_m } and a cloud storage system, wherein H _i represents an ith hospital, and i is more than or equal to 1 and less than or equal to n; p _j represents the j patient, j is more than or equal to 1 and less than or equal to m; the medical information sharing method comprises the following steps:

Step 1, generating medical information of a patient by a hospital, wherein the medical information comprises medical record information and medical records:

step 1.1, generating medical record information of a patient:

Assuming that the j-th patient p _j is to be in a visit to the I-th hospital h _i, the I-th hospital h _i acquires the personal identification information ID _j of the j-th patient p _j and the medical image I _j and generates an electronic medical record R _j so as to construct medical record information R _j＝(ID_j,r_j of the j-th patient p _j;

step 1.2, generating a medical record of the patient:

Generating a medical record of the j patient p _j after the j patient p _j is treated by the h _i of the ith hospital in step 1.2.1, wherein the medical record comprises a treatment scheme and a treatment record;

step 1.2.2, analyzing the medical record of the j patient p _j to generate text information MR _j of the medical record of the j patient p _j;

Step 2, encrypting patient medical record information:

Step 2.1, generating an encryption key _j and an identification token _j of a j patient p _j, and writing a diagnosis card of the j patient p _j;

Step 2.2, encrypting the medical record information R _j of the jth patient p _j by adopting an AES encryption algorithm to obtain a medical record information ciphertext Rs _j encrypted by the jth patient p _j;

Step 3, embedding the encrypted medical record information ciphertext Rs _j as secret information into the medical image I _j to obtain a steganographic image containing the secret information

Step 3.1, decomposing the bit plane of the medical image I _j into an HSB bit plane of the medical image I _j And LSB bit plane of medical image I _j Order theRepresentation ofThe pixel value of the ith row and the jth column,Representation ofPixel values of the u th row and v th column of the matrix;

step 3.2 HSB bit plane for medical image I _j Pixels that may spill are marked to create an HSB bit plane for medical image I _j Corresponding position map LM _j:

setting a maximum threshold T _max and a minimum threshold T _min;

If it is Or (b)HSB bit plane representing medical image I _j Pixel values of the ith row and the ith column overflow and willThe position of the u-th row and v-th column in the list is marked as '1';

If it is Or (b)HSB bit plane representing medical image I _j Pixel values of the ith row and the ith column of the medical image I _j overflow and the HSB bit plane of the medical image I _j is processedThe position of the u-th row and v-th column in the list is marked as '2';

Otherwise, HSB bit plane representing medical image I _j Pixel values of the ith row and the ith column of the medical image I _j are not overflowed and the HSB bit plane of the medical image I _j isThe position of the u-th row and v-th column in the list is marked as 0; thereby obtaining the HSB bit plane of the medical image I _j A corresponding position map LM _j;

step 3.3, modifying the HSB bit plane of the medical image I _j Pixel values that may overflow, resulting in a processed HSB bit-plane of the medical image I _j

If it isThe HSB bit plane of the medical image I _j Pixel value of the ith row and the ith column of the middleMinus "2";

If it is The HSB bit plane of the medical image I _j Pixel value of the ith row and the ith column of the middleAdd "2";

If it is The HSB bit plane of the medical image I _j Pixel value of the ith row and the ith column of the middleMinus "1";

If it is The HSB bit plane of the medical image I _j Pixel value of the ith row and the ith column of the middleAdd "1";

step 3.4, performing lossless compression on the position map LM _j by using arithmetic coding to obtain a position map after compression of the medical image I _j

HSB bit plane after processing of medical image I _j The other pixel points except the first row of pixel points, the first column of pixel points, the last row of pixel points and the last column of pixel points are all divided into checkerboard patterns;

step 3.5, pixel values of the ith row and the ith column in the checkerboard are calculated The adjacent 8 pixel values are arranged in ascending order, and the average value of the first six pixel values in the ordered pixel values is calculated and then rounded and recorded as a predicted value q ₁ (u ', v') of the (u 'row and the (v') column; calculating the average value of the last six pixel values in the ordered pixel values, rounding the average value, and marking the average value as a predicted value q ₂ (u ', v') of the (v 'th) column of the (u' th row;

Step 3.6, embedding each pixel point in the checkerboard twice by using a prediction error expansion method, thereby obtaining a new HSB bit plane

Step 3.6.1, the compressed position diagramThe new secret data Rs ' _j is obtained by adding the new secret data Rs ' to the tail end of the encrypted medical record information ciphertext Rs _j, and the first secret information is defined as b and the second secret information b ';

step 3.6.2, calculating the first embedded prediction error

Step 3.6.3, when e ₁ =1, for the pixel value of the ith row and the ith column in the checkerboardAdding part of the first secret information b;

when e ₁ =0, for the pixel value of the ith row and the ith column in the checkerboard Subtracting part of the first secret information b;

When e ₁ > 1, the pixel value of the ith row and the ith column in the checkerboard is calculated Self-increasing 1;

when e ₁ < 0, the pixel value of the ith row and the ith column in the checkerboard is calculated Self-subtracting 1;

Thereby obtaining the pixel value of the (v 'th) column of the (u' th) row after the first embedding

Step 3.6.4, calculating the prediction error of the second embedding

When e ₂ =1, the pixel value for the ith row and the ith columnAdding part of the second secret information b';

When e ₂ =0, the pixel value for the ith row and the ith column Subtracting part of the second secret information b';

when e ₂ > 1, the pixel value for the (v 'th) column of the (u' th) row Self-increasing 1;

when e ₂ < 0, the pixel value of the v 'column of the u' th row Self-subtracting 1;

thereby obtaining the pixel value of the (v 'th) column of the (u' th) row after the second embedding

Step 3.7, new HSB bit planeAndCombining to obtain a steganographic image I _j ^em containing secret information;

Step 4, merging the medical information and uploading the medical information to a cloud storage system:

Step 4.1, combining the text information MR _j of the medical record of the j patient p _j, the identification token _j and the steganographic image I _j ^em containing secret information to form complete encrypted medical information M _j＝(MR_j,I_j ^em,token_j of the j patient p _j;

Step 4.2, uploading the complete encrypted medical information M _j of the jth patient p _j to the cloud storage system;

Step 5, retrieving medical information of similar cases from the cloud storage system:

step 5.1, performing text matching and information retrieval by using a deep learning model:

step 5.1.1, inputting queried disease symptom information according to the disease condition of the jth patient p _j;

Step 5.1.2, analyzing the disease symptom information by using a deep learning model, and extracting semantic feature vectors;

step 5.1.3, calculating the feature similarity between the semantic feature vector and the medical record information in the cloud storage system, and generating a text similarity descending order ordering result;

Step 5.1.4, according to the text descending order similarity sorting result, obtaining a text information set MR '_j＝{MR'_1,j,MR'_2,j,...,MR'_k,j,...,MR'_r,j of a case similar to the j patient p _j, wherein MR' _k,j represents the kth text information of the case similar to the j patient p _j, and k is more than or equal to 1 and less than or equal to r; r represents the total number of similar cases;

step 5.2, performing feature extraction and feature retrieval on the medical image I _j by using a deep learning model:

Step 5.2.1, inputting the medical image I _j of the jth patient p _j into a deep learning model for feature extraction, obtaining features of a focus region in the medical image I _j and storing the features as focus feature vectors;

Step 5.2.2, performing image feature similarity calculation on the focus feature vector of the medical image I _j of the jth patient p _j and the focus feature vector of the medical image in the cloud storage system, and generating an image similarity descending order sequencing result;

Step 5.2.3, according to the image similarity descending order sequencing result, obtaining a medical image set I' _j＝{I_1,j,I_2,j,...,I_d,j,...,I_r,j of a case similar to the p _j th patient, wherein I _d,j represents a d-th medical image of the case similar to the p _j th patient, and d is more than or equal to 1 and less than or equal to r;

Step 5.3, according to the personal identity information ID _j of the jth patient p _j, acquiring the previous medical information of the jth patient p _j from the cloud storage system:

The method comprises the steps that a treatment card of a j patient p _j is used for carrying out identity verification on a terminal of an h _i of an i-th hospital, and complete encrypted medical information M _j of the j patient p _j is obtained from the cloud storage system according to an identity identification token _j in the treatment card;

step 5.4, obtaining an encrypted medical information set MR "_j＝{MR″_1,j,MR″_2,j,...,MR″_k,j,...,MR″_r,j" of other similar cases according to the complete encrypted medical information M _j of the jth patient p _j, wherein MR "_k,j represents the encrypted medical information of the kth patient of the similar cases as the jth patient p _j, and k is more than or equal to 1 and less than or equal to r;

Let the text information of the medical record in the encrypted medical information MR' _k,j of the kth patient similar to the jth patient p _j be MR _k,j and the steganographic image be

Step 6, for steganographic imageExtraction and image recovery are performed:

step 6.1, steganographic image Decomposition into steganographic imagesHSB bit plane of (v)Steganographic imageLSB bit plane of (b)

Step 6.2, for steganographic imageHSB bit plane of (v)Extracting each pixel point twice to obtain HSB bit plane value after the second extractionAnd composing new secret data Rs' _k from all the extracted secret information;

Step 6.2.1 for steganographic image LSB bit plane of (b)Pixel value of the ith row and the ith column of the middleThe 8 adjacent pixels are arranged in ascending order, and the average value of the first six pixel values in the ordered pixel values is calculated and then is rounded and then is recorded as a predicted value q' ₁ (u, v) of a ith row and a ith column; calculating the average value of the last six pixel values in the sorted pixel values, rounding the average value, and marking the average value as a predicted value q' ₂ (u, v) of a ith row and a ith column;

Step 6.2.2, calculating the prediction error of the first extraction When e' ₂ =1, 2,0 or-1, fromExtracting secret information extracted onceAnd obtaining HSB bit plane value after one extraction

Step 6.2.3, calculating the prediction error of the second extractionWhen e' ₁ = 1,2,0 or-1, fromExtracting the secret information extracted twiceAnd obtaining HSB bit plane value after secondary extraction

Step 6.3, separating the steganographic image from the new secret data Rs' _k Compressed position diagram of (2)And medical record information ciphertext Rs _k;

Step 6.4, for the steganographic image Compressed position diagram of (2)Decompression is performed to obtain a position map LM _k corresponding to the new secret data Rs '_k, and the pair of position maps LM _k corresponding to the new secret data Rs' _k Reducing and then combining withAfter combination, a medical image I _k,j of the kth patient of a case similar to the jth patient p _j is obtained;

step 7, patient identification and information decryption:

Step 7.1, searching encrypted medical information with the same identification token from the cloud storage system according to the identification token _j held by the j-th patient p _j;

And 7.2, decrypting the medical record information ciphertext Rs _j encrypted by the j patient p _j by using a key _j of the j patient p _j and adopting an AES algorithm to obtain medical record information R _j, and decomposing the medical record information R _j of the j patient p _j to obtain personal identity information ID _j and electronic medical record R _j of the j patient p _j.

2. An electronic device comprising a memory and a processor, wherein the memory is configured to store a program that supports the processor to perform the medical information sharing method of claim 1, the processor being configured to execute the program stored in the memory.

3. A computer readable storage medium having a computer program stored thereon, characterized in that the computer program when executed by a processor performs the steps of the medical information sharing method of claim 1.