WO2023092342A1 - 一种基于密文图像修复的医学图像隐私保护方法 - Google Patents
一种基于密文图像修复的医学图像隐私保护方法 Download PDFInfo
- Publication number
- WO2023092342A1 WO2023092342A1 PCT/CN2021/132861 CN2021132861W WO2023092342A1 WO 2023092342 A1 WO2023092342 A1 WO 2023092342A1 CN 2021132861 W CN2021132861 W CN 2021132861W WO 2023092342 A1 WO2023092342 A1 WO 2023092342A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image
- texture
- block
- medical
- patch set
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 20
- 230000009466 transformation Effects 0.000 claims abstract description 6
- 230000003902 lesion Effects 0.000 claims description 11
- 238000003709 image segmentation Methods 0.000 claims description 4
- 238000003064 k means clustering Methods 0.000 claims description 4
- 238000003062 neural network model Methods 0.000 claims description 3
- 230000000903 blocking effect Effects 0.000 claims 2
- 230000000694 effects Effects 0.000 abstract description 3
- 239000011159 matrix material Substances 0.000 description 2
- 208000003174 Brain Neoplasms Diseases 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H30/00—ICT specially adapted for the handling or processing of medical images
- G16H30/20—ICT specially adapted for the handling or processing of medical images for handling medical images, e.g. DICOM, HL7 or PACS
Definitions
- the invention relates to the technical field of medical information security protection, in particular to a medical image privacy protection method based on ciphertext image restoration.
- image encryption algorithm and image restoration technology based on plaintext The main methods to protect the privacy of medical images are: image encryption algorithm and image restoration technology based on plaintext.
- image encryption algorithms cannot effectively resist chosen plaintext attacks, and hackers may crack encrypted images, so they cannot achieve good results in terms of security
- image restoration techniques based on plaintext require additional databases to store cut lesions image, and then directly transmit the cut-out image to the image processor.
- the cut-out image does not contain lesion information, the image processor can still infer some diseases from it, and there is a risk of leaking the patient’s privacy. Therefore, This method is not very secure.
- the current ciphertext image restoration technology cannot be applied to protect the privacy of medical images.
- the purpose of the present invention is to provide a medical image privacy protection method based on ciphertext image restoration.
- the present invention provides the following scheme:
- a medical image privacy protection method based on ciphertext image restoration comprising:
- the copy image is a copy image of the medical image
- the local image block is composed of each pixel in the image to be repaired and s-1 pixels adjacent to each pixel;
- the block to be repaired is composed of all pixels in the image block whose corresponding positions in the mask image have a value of 1;
- an encrypted image with embedded information is generated.
- the invention discloses the following technical effects:
- the invention adopts the technical means of ciphertext image repair, thereby achieving the technical effect that the image processor completes the damaged image repair work under the premise of completely ignoring the image content information, and improves the safety of the method.
- Fig. 1 is a flowchart of a medical image privacy protection method based on ciphertext image restoration.
- a kind of medical image privacy protection method based on ciphertext image restoration comprises the following steps:
- Step 101 segment the lesion area in the medical image through the trained image segmentation neural network model, and generate a mask image.
- the database manager takes the medical image I taken by the patient for diagnosis, such as a brain tumor slice image with a size of 240 ⁇ 240.
- the trained image segmentation neural network automatically segments the lesion area in the image to generate a mask image I M (All pixels belonging to the lesion area in the medical image have a value of 1 at the corresponding position in the mask image, and the remaining pixels in the medical image have a value of 0 at the corresponding position in the mask image).
- Step 102 Cut out the lesion area in the copy image based on the mask image to generate an image to be repaired; the copy image is a copy image of the medical image.
- Step 103 Perform convolution and clustering on the image to be repaired by linear spatial filter and K-means clustering algorithm to obtain a texture image.
- Step 104 Perform JL transform encryption on each local image block in the image to be repaired to obtain a JL transform encrypted result image; the local image block is composed of each pixel in the image to be repaired and s-1 pixels adjacent to each pixel.
- the database manager generates a random matrix of size 7*25 according to the encryption key ⁇ J of the JL transformation algorithm and a randomly generated 7-dimensional noise vector ⁇ , which is used to encrypt each pixel in the image I R to be repaired in step 102 according to the formula (1) and the local image block G composed of s-1 pixels adjacent to it, For pixels on the image border, missing pixels within the image block can be filled with surrounding pixels.
- G(i, j) represents the matrix composed of all pixel values in the local image block composed of the i-th row, j-th column pixel and its adjacent s-1 pixels in the image to be repaired.
- Step 105 Encrypt the medical image with an image encryption algorithm to obtain an encrypted medical image.
- the database manager uses an image encryption algorithm that does not change the pixel position of the image, such as the RSA algorithm to encrypt the medical image I.
- Step 106 Block the encrypted medical image based on the texture image to obtain multiple image blocks.
- the texture distance between the upper and lower equal non-overlapping sub-blocks B 1v and B 2v of the image block is calculated according to formulas (2) and (3):
- the value of the jth column of the row, ⁇ represents the texture type number, ⁇ represents the set of position indices of pixels with a pixel value of 0 in the mask image, B represents the set of position indices of all pixels in the image block B,
- the direction flag of the image block is h, first calculate the texture distance between the left and right sub-blocks and judge whether it is greater than a given difference threshold, if so, divide the image block into equal left and right sub-blocks, otherwise calculate the upper and lower sub-blocks Determine whether the texture distance between the two sub-blocks is greater than a given difference threshold, if so, divide the image block into equal upper and lower sub-blocks, otherwise keep the image block unchanged.
- All image blocks in the encrypted medical image E include sub-blocks and repeat the above-mentioned method of dividing blocks until all image blocks are inseparable (the texture distance between the upper and lower sub-blocks of the image block and the texture distance between the upper and lower sub-blocks of the image block are equal to each other). is not greater than a given difference threshold or the size of the image block is already a given minimum size).
- Step 107 Based on the mask image and the texture image, determine the restricted source area of the block to be repaired in each image block; the block to be repaired is composed of all the corresponding position values of 1 in the mask image in the image block pixel composition.
- the image processor reduces the patch search range, and first calculates each credible block to be repaired in the encrypted medical image E according to formulas (2) and (3) (There are pixels with a value of 1 at the corresponding position of the mask image in the image block and the number of such pixels accounts for less than half of the total number of pixels in the block) and the remaining trusted blocks B rel (corresponding to the mask image in the image block) The number of pixels with a position value of 0 accounts for more than half of the total number of pixels in the block) The texture distance between will be associated with the trusted block to be repaired texture distance
- the intact pixels in all credible blocks B rel smaller than a given difference threshold ⁇ (the pixel has a value of 0 in the corresponding position of the mask image) constitute the restricted source area of the block to be repaired.
- each untrusted repair block There are pixels with a value of 1 in the corresponding position of the mask image in the image block and the number of such pixels accounts for more than half of the total number of pixels in the block), its limited source area It is the intact pixel in the
- Step 108 Determine an optimal patch set in the restricted source region based on the JL transform encryption result image.
- the image processor divides the encrypted medical image into several lattices with a length and width of x pixels (the endpoint of the lattice falls on the pixel), and some lattices contain pixels to be repaired (the corresponding position of the pixel in the mask image value of 1), then each endpoint of these lattices is called a node.
- each endpoint of these lattices is called a node.
- ⁇ represents the node in the encrypted medical image
- ⁇ ⁇ represents the restricted source area of the image block where the node ⁇ is located
- ⁇ R represents the relative threshold, which is given
- (.) + represents the step function, when the parameter is greater than 0, it returns 1, otherwise it returns 0,
- V ⁇ (q) represents the cost, which is the local image block with a length and width of 2x-1 pixels centered on the node ⁇ the overhead between patches q, Represents the patch with the least overhead among all the patches in the restricted source area of the image block where node ⁇ is located, N( ⁇ ) represents the set of position indexes whose length and width are 2x-1 pixels centered on node ⁇ , E J (i, j) represents the value of row i and column j in the JL transform encryption result image, and the relative position of (i 2 , j 2 ) in the patch is consistent with the relative position of (i 1 , j 1 ) in the local image block.
- the texture weight overhead between the local image block of the node and all patches in the restricted source area of the image block where the node is located is calculated according to formulas (6) and (7) Only retain the texture weight overhead between the local image block of the node The smallest g patches are discarded, and the remaining patches are discarded. At this time, the node only has g patches for candidates, and then propagates the overhead to adjacent nodes according to formulas (8) and (9).
- B q represents the image block where the central pixel of the patch q is located
- B ⁇ represents the image block where the node ⁇ is located. Only keep the g patches with the smallest texture weight overhead between the local image blocks of the node, and discard the rest of the patches. At this time, the node only has g patches for candidates.
- Markov energy can be expressed as:
- Step 109 Concatenate the adjacent patches in the optimal patch set to obtain the optimal patch set after concatenation.
- I W (i 1 , j 1 ) represents the value of row i 1 and column j 1 in the texture image.
- the cumulative texture error line by line in the overlapping area can be expressed as:
- (i, j) represents the index of the relative position in the overlapping area, and if it is the upper and lower adjacent nodes, the texture error will be accumulated column by column in the overlapping area. Find the boundary that minimizes the sum of texture errors in the overlapping area, and then connect two adjacent patches according to the boundary.
- Step 1010 Generate an encrypted image with embedded information based on the concatenated optimal patch set and the binary bit stream.
- the image processor encodes the ciphertext values and position information of all pixels to be repaired in the encrypted medical image E into a binary bit stream b, and then uses the previous step to complete the connection
- the patch replaces the local image blocks of all nodes in the encrypted medical image E to obtain the encrypted repair result E', and then uses the reversible information hiding technology with separable ciphertext domain to reversibly embed the binary bit stream b according to the embedding key ⁇ 1
- an encrypted image E" with embedded information is obtained, which is transmitted to the image database manager.
- the image database manager can store the encrypted image E" received in his own database, and when the doctor or patient needs to view the original image I, the image database manager will transmit the encrypted image E" stored in the database to the doctor or patient , the doctor or patient can extract the key ⁇ 2 according to the information, extract the binary bit stream b, and then decode the binary bit stream to obtain the ciphertext value and position information, and then use these ciphertext values to replace the corresponding positions in the encrypted image E′′ The ciphertext value, and finally according to the decryption private key Decrypt all the ciphertexts in the encrypted image E′′ to get the original image I, as follows:
Landscapes
- Health & Medical Sciences (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Radiology & Medical Imaging (AREA)
- Engineering & Computer Science (AREA)
- Epidemiology (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Image Processing (AREA)
Abstract
本发明公开了一种基于密文图像修复的医学图像隐私保护方法。该方法包括：通过医学图像生成掩码图像、成待修复图像、纹理图像、JL变换加密结果图像和加密的医学图像；基于纹理图像对加密的医学图像进行分块，得到多个图像块；基于掩码图像和纹理图像确定各图像块中待修复块的受限源区域；基于JL变换加密结果图像，确定受限源区域内的最佳补丁集合；对最佳补丁集合中相邻的补丁进行衔接，得到衔接后的最佳补丁集合；基于衔接后最佳补丁集合以及二进制比特流，生成带有嵌入信息的加密图像。本发明解决了现有医学图像隐私保护方法安全性不高和现有密文图像修复技术因修复效果不佳而无法应用于图像隐私保护的问题。
Description
本发明涉及医疗信息安全保护技术领域，特别是涉及一种基于密文图像修复的医学图像隐私保护方法。
现保护医学图像隐私的主要方法有：图像加密算法和基于明文图像修复技术。其中，图像加密算法多数无法有效抵抗选择明文攻击，黑客有可能破解加密的图像，所以在安全性方面不能达到很好的效果，而基于明文图像修复技术需要额外数据库用于存储剪切下的病灶图像，再将剪切掉病灶的图像直接传输给图像处理者，剪切掉病灶的图像中虽不包含病灶信息，但图像处理者仍然可从中推断出一些病情，存在泄露患者隐私的风险，所以这种方法的安全性不高。且目前的密文图像修复技术无法应用于保护医学图像的隐私。
发明内容
本发明的目的是提供一种基于密文图像修复的医学图像隐私保护方法。
为实现上述目的，本发明提供了如下方案：
一种基于密文图像修复的医学图像隐私保护方法，包括：
通过训练好的图像分割神经网络模型分割出医学图像中的病灶区域，生成掩码图像；
基于所述掩码图像剪切掉副本图像中的病灶区域，生成待修复图像；所述副本图像为所述医学图像的复制图像；
通过线性空间滤波器和K-均值聚类算法对所述待修复图像进行卷积和聚类，得到纹理图像；
对待修复图像中各局部图像块进行JL变换加密，得到JL变换加密结果图像；所述局部图像块由待修复图像中各像素以及与各像素相邻的s-1个像素组成；
通过图像加密算法对所述医学图像进行加密，得到加密的医学图像；
基于所述纹理图像对所述加密的医学图像进行分块，得到多个图像块；
基于所述掩码图像和所述纹理图像确定各图像块中待修复块的受限源区域；所述待修复块由所述图像块内所有在掩码图像相应位置值为1的像素组成；
基于所述JL变换加密结果图像，确定所述受限源区域内的最佳补丁集合；
对所述最佳补丁集合中相邻的补丁进行衔接，得到衔接后的最佳补丁集合；
基于衔接后最佳补丁集合以及二进制比特流，生成带有嵌入信息的加密图像。
根据本发明提供的具体实施例，本发明公开了以下技术效果：
本发明采用密文图像修复的技术手段，从而达到了图像处理者在完全不知图像内容信息的前提下完成受损图像修复工作的技术效果，提高了方法的安全性。
图1为基于密文图像修复的医学图像隐私保护方法的流程图。
下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述。
如图1所示，本发明提供的一种基于密文图像修复的医学图像隐私保护方法，包括以下步骤：
步骤101：通过训练好的图像分割神经网络模型分割出医学图像中的病灶区域，生成掩码图像。
像数据库管理者将患者诊断所拍的医学图像I如一张大小为240×240的脑部肿瘤切片图像经过训练好的图像分割神经网络自动地分割出图像中的病灶区域，生成掩码图像I
M(医学图像中属于病灶区域的所有像素在掩码图像中相应位置的值为1，医学图像中的其余像素在掩码图像中相应位置的值为 0)。
步骤102：基于所述掩码图像剪切掉副本图像中的病灶区域，生成待修复图像；所述副本图像为所述医学图像的复制图像。
像数据库管理者复制一个与医学图像I完全相同的副本图像，再基于步骤101的掩码图像I
M剪切掉副本图像中的病灶区域(对副本图像中每一个像素，若该像素在掩码图像中相应位置的值为1，那么该像素的像素值被改为4095(假设颜色深度为12位)，否则该像素的像素值保持不变)，生成待修复图像I
R。
步骤103：通过线性空间滤波器和K-均值聚类算法对所述待修复图像进行卷积和聚类，得到纹理图像。
像数据库管理者用18个不同方向(6个)和尺寸(3种)的Gabor滤波器卷积步骤102的待修复图像I
R，再对其输出的振幅应用K-均值聚类算法(K＝16)，记录待修复图像I
R中每一个像素的聚类结果，得纹理图像I
W。
步骤104：对待修复图像中各局部图像块进行JL变换加密，得到JL变换加密结果图像；所述局部图像块由待修复图像中各像素以及与各像素相邻的s-1个像素组成。
像数据库管理者根据JL变换算法的加密密钥κ
J，生成大小为7*25随机矩阵
和随机生成的7维噪声向量σ，用于根据公式(1)加密步骤102的待修复图像I
R中每个像素以及与它相邻的的s-1个像素所组成的局部图像块G，对于图像边界上的像素，可以使用周围像素填补图像块内的缺失像素。
式中，G(i，j)代表待修复图像中第i行第j列像素以及与它相邻的s-1个像素所组成的局部图像块内所有像素值组成的矩阵。
收集所有对应的加密结果以形成JL变换加密结果图像E
J。
步骤105：通过图像加密算法对所述医学图像进行加密，得到加密的医学图像。
像数据库管理者利用不改变图像像素位置的图像加密算法如RSA算法加密医学图像I，图像所有者首先选择两个大素数p
1和p
2来生成加密公钥κ＝e和 解密私钥
其中e是满足gcd(e，(p
1-1)(p
2-1))＝的大整数，
n是任意的正整数，然后利用加密公钥κ＝e加密医学图像I中每个像素，得加密的医学图像E，如下式：
E(i，j)＝I(i，j)
emod(p
1·p
2)
最后将加密的医学图像E、JL变换加密结果图像E
J、纹理图像I
W和掩码图像I
M都传输给图像处理者。
步骤106：基于所述纹理图像对所述加密的医学图像进行分块，得到多个图像块。
像处理者自适应分块加密的医学图像E，首先将加密的医学图像E分成近似均等不重叠的四个图像块，并给每个图像块依据它们更长的维度分配一个方向标志δ＝v或h。
若图像块的方向标志为v，则根据公式(2)和(3)计算图像块上下两均等不重叠子块B
1v和B
2v之间的纹理距离：
式中，
代表图像块的纹理，是图像块B内第η类纹理的统计；(i，j)代表图像中第i行第j列像素的位置索引，I
W(i，j)代表纹理图像中第i行第j列的值，η代表纹理类型号，Φ代表掩码图像中像素值为0的像素的位置索引的集合，B代表图像块B中所有像素的位置索引的集合，|.|代表返回集合内元素数量的函数，ξ代表指标函数，当参数为真返回1，否则返回0。
当纹理距离
大于给定的差异阈值τ＝0.15时，将图像块分为上下两子块，否则再根据公式(2)和(3)计算左后均等两子块B
1h和B
2h之间的纹理距离，当纹理距离
大于给定的差异阈值τ，将图像块分为左右两子块，否则保持图像块不变。
若图像块的方向标志为h，则优先计算左右两子块之间的纹理距离并判断 是否大于给定的差异阈值，若是，则将图像块分为均等的左右两子块，否则再计算上下两子块的纹理距离并判断是否大于给定的差异阈值，若是，则将图像块分为均等的上下两子块，否则保持图像块不变。
当图像块被分成两子块后，再给这两子块都分配一个与父块方向标志相反的方向标志。加密的医学图像E内所有图像块包括子块一直重复上述方式分块，直到所有图像块已不可分(该图像块的上下两子块之间的纹理距离和上下两子块之间的纹理距离都不大于给定的差异阈值或该图像块的尺寸已是给定的最小尺寸)。
步骤107：基于所述掩码图像和所述纹理图像确定各图像块中待修复块的受限源区域；所述待修复块由所述图像块内所有在掩码图像相应位置值为1的像素组成。
图像处理者缩减补丁搜索范围，首先根据公式(2)和(3)计算加密的医学图像E中每一个可信待修复块
(该图像块内有在掩码图像相应位置值为1的像素且这样的像素数量占块内总像素数量的比列小于一半)与其余可信块B
rel(该图像块内在掩码图像相应位置值为0的像素的数量占块内总像素数量一半以上)之间的纹理距离
将与该可信待修复块
之间纹理距离
小于给定的差异阈值τ的所有可信块B
rel内的完好像素(该像素在掩码图像相应位置值为0)组成该待修复块的受限源区域
然后对于每一个不可信修复块
(该图像块内有在掩码图像相应位置值为1的像素且这样的像素数量占块内总像素数量的一半以上)，它的受限源区域
是该块内完好像素、该块相邻可信待修复块的受限源区域和该块相邻不可信块(图像块内完好像素数量与块内总像素数量的比例小于0.5)内完好像素所组成的区域，最后加密的医学图像E中所有待修复块(该图像块内有在掩码图像相应位置值为1的像素)都有各自的受限源区域
步骤108：基于所述JL变换加密结果图像，确定所述受限源区域内的最佳补丁集合。
像处理者将加密的医学图像分成若干个长宽为x个像素的晶格(晶格的端点落在像素上)，一些晶格内包含待修复像素(该像素在掩码图像内的相应位置值为1)，那么这些晶格的每一个端点都被称为节点。根据公式(4)、(5)和(6) 计算每个节点λ的优先级P(λ)并按照优先级P(λ)由大到小的顺序逐个访问每个节点。
式中，λ代表加密的医学图像中的节点，q代表长宽为2x-1个像素(s＝(2x-1)2)的补丁，Λ
λ代表节点λ所在图像块的受限源区域内所有补丁组成的集合，τ
R代表相对阈值，是给定的，(.)
+代表阶跃函数，当参数大于0则返回1，否则返回0，
代表相对开销，是节点λ所在图像块的受限源区域内补丁q的相对开销，V
λ(q)代表开销，是以节点λ为中心的长宽为2x-1个像素的局部图像块与补丁q之间的开销，
代表节点λ所在图像块的受限源区域内所有补丁中开销最小的补丁，N(λ)代表以节点λ为中心的长宽为2x-1个像素的位置索引的集合，E
J(i，j)代表JL变换加密结果图像中第i行第j列的值，(i
2，j
2)在补丁内的相对位置与(i
1，j
1)在局部图像块内的相对位置一致。
当访问节点时，根据公式(6)和(7)计算该节点的局部图像块与该节点所在图像块的受限源区域内所有补丁之间的纹理权重开销
只保留与该节点的局部图像块之间纹理权重开销
最小的g个补丁，其余补丁丢弃，此时该节点仅有g个补丁用于候选，再根据公式(8)和(9)传播开销至相邻节点。
式中，B
q代表补丁q的中心像素所在的图像块，B
λ代表节点λ所在的图像块。 只保留与该节点的局部图像块之间纹理权重开销最小的g个补丁，其余补丁丢弃，此时该节点仅有g个补丁用于候选，
代表与优先级更高的节点λ相邻的节点，
代表节点
的集合
中的一个补丁，t代表更新步数，
代表成对潜力，是补丁q与补丁
假设分别覆盖节点λ和
的局部图像块时补丁q内重叠区域和补丁
内重叠区域之间的开销，
代表当补丁q与补丁
假设分别覆盖节点λ和
的局部图像块时补丁q内重叠区域的像素位置索引的集合，
代表当补丁q与补丁
假设分别覆盖节点λ和
的局部图像块时补丁
内重叠区域的像素位置索引的集合，(i
2，j
2)在补丁
内重叠区域的相对位置与(i
1，j
1)在补丁q内重叠区域的相对位置一致。
当所有节点被访问，此时每个节点都只有g个用于候选的补丁，从中选出使马尔科夫能量最小的补丁组合，最后每个节点只剩一个用于修复的补丁。马尔科夫能量可表示为：
步骤109：对所述最佳补丁集合中相邻的补丁进行衔接，得到衔接后的最佳补丁集合。
在实际应用中，当每个节点的局部图像块用一个补丁覆盖时，相邻节点之间的补丁存在重叠区域，此时图像处理者需要找到重叠区域内最小纹理误差S的边界来衔接两相邻节点的补丁，两相邻节点的重叠区域内两像素的纹理误差可表示为
式中，I
W(i
1，j
1)代表纹理图像中第i
1行第j
1列的值。
若是左右相邻的节点，则在重叠区域内逐行累计纹理误差可表示为：
式中，(i，j)代表重叠区域内相对位置的索引，若是上下相邻的节点，则在重叠区域内逐列累计纹理误差。在重叠区域内找到使纹理误差总和最小的边界，再根据边界衔接两相邻补丁。
步骤1010：基于衔接后最佳补丁集合以及二进制比特流，生成带有嵌入信息的加密图像。
在实际应用中，当所有相邻节点的补丁衔接完之后，图像处理者将加密的医学图像E中所有待修复像素的密文值和位置信息编码成二进制比特流b，再用上一步衔接好的补丁代替加密的医学图像E中所有节点的局部图像块，得加密的修复结果E'，再利用密文域可分离的可逆信息隐藏技术根据嵌入密钥κ
1将二进制比特流b可逆地嵌入在加密的修复结果E'中，得带有嵌入信息的加密图像E″，将其传输给图像数据库管理者。
图像数据库管理者将收到的加密图像E″可存储在自己的数据库中，当医生或者患者需要查看原始图像I时，图像数据库管理者把存储在数据库中的加密图像E″传输给医生或者患者，医生或者患者可根据信息提取密钥κ
2，提取出二进制比特流b，再将二进制比特流译码出密文值和位置信息，然后用这些密文值代替加密图像E″中相应位置的密文值，最后根据解密私钥
解密加密图像E″中的所有密文，即可得原始图像I，如下式：
I(i，j)＝E″(i，j)
dmod(p
1·p
2))
本说明书内容不应理解为对本发明的限制。
Claims (4)
- 一种基于密文图像修复的医学图像隐私保护方法，其特征在于，包括：通过训练好的图像分割神经网络模型分割出医学图像中的病灶区域，生成掩码图像；基于所述掩码图像剪切掉副本图像中的病灶区域，生成待修复图像；所述副本图像为所述医学图像的复制图像；通过线性空间滤波器和K-均值聚类算法对所述待修复图像进行卷积和聚类，得到纹理图像；对待修复图像中各局部图像块进行JL变换加密，得到JL变换加密结果图像；所述局部图像块由待修复图像中各像素以及与各像素相邻的s-1个像素组成；通过图像加密算法对所述医学图像进行加密，得到加密的医学图像；基于所述纹理图像对所述加密的医学图像进行分块，得到多个图像块；基于所述掩码图像和所述纹理图像确定各图像块中待修复块的受限源区域；所述待修复块由所述图像块内所有在掩码图像相应位置值为1的像素组成；基于所述JL变换加密结果图像，确定所述受限源区域内的最佳补丁集合；对所述最佳补丁集合中相邻的补丁进行衔接，得到衔接后的最佳补丁集合；基于衔接后最佳补丁集合以及二进制比特流，生成带有嵌入信息的加密图像。
- 根据权利要求1所述的基于密文图像修复的医学图像隐私保护方法，其特征在于，所述基于所述纹理图像对所述加密的医学图像进行分块，得到多个图像块，具体包括：对所述加密的医学图像进行分块，得到多个初始图像块；基于各初始图像块最长的维度确定方向标志；基于所述纹理图像，计算不同方向标志的各初始图像块中各子图像块之间 的纹理距离；基于所述纹理距离对各所述初始图像块进行分块。
- 根据权利要求1所述的基于密文图像修复的医学图像隐私保护方法，其特征在于，所述基于所述JL变换加密结果图像，确定所述受限源区域内的最佳补丁集合，具体包括：计算所述受限源区域内所有节点的优先级；按照所述优先级访问各节点，并基于所述JL变换加密结果图像计算各节点的局部图像块与各节点所在图像块的受限源区域内所有补丁之间的纹理权重开销；根据所述纹理权重开销确定最佳补丁集合。
- 根据权利要求3所述的基于密文图像修复的医学图像隐私保护方法，其特征在于，所述对所述最佳补丁集合中相邻的补丁进行衔接，得到衔接后的最佳补丁集合，具体包括：基于所述纹理图像，确定相邻节点的重叠区域内最小纹理误差的边界；根据所述边界衔接相邻的补丁，得到衔接后的最佳补丁集合。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2021/132861 WO2023092342A1 (zh) | 2021-11-24 | 2021-11-24 | 一种基于密文图像修复的医学图像隐私保护方法 |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2021/132861 WO2023092342A1 (zh) | 2021-11-24 | 2021-11-24 | 一种基于密文图像修复的医学图像隐私保护方法 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2023092342A1 true WO2023092342A1 (zh) | 2023-06-01 |
Family
ID=86538674
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2021/132861 WO2023092342A1 (zh) | 2021-11-24 | 2021-11-24 | 一种基于密文图像修复的医学图像隐私保护方法 |
Country Status (1)
| Country | Link |
|---|---|
| WO (1) | WO2023092342A1 (zh) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117156182A (zh) * | 2023-10-27 | 2023-12-01 | 北京健滨医药科技有限公司 | 一种医疗远程智能监测方法及系统 |
| CN117499551A (zh) * | 2024-01-02 | 2024-02-02 | 天津中荣印刷科技有限公司 | 一种基于二进制信号的加密防伪印刷方法 |
| CN117973496A (zh) * | 2024-02-19 | 2024-05-03 | 苏州慧维智能医疗科技有限公司 | 局部病变医疗图像生成模型的训练方法、系统及应用 |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160182920A1 (en) * | 2014-12-18 | 2016-06-23 | Konkuk University Industrial Cooperation Corp | Error concealment method using spatial interpolation and exemplar-based image inpainting |
| CN108846816A (zh) * | 2018-06-20 | 2018-11-20 | 上海理工大学 | 一种密文域图像修复方法 |
| CN110659511A (zh) * | 2019-09-25 | 2020-01-07 | 中国人民解放军61646部队 | 一种遥感图像的信息隐藏处理方法及装置 |
-
2021
- 2021-11-24 WO PCT/CN2021/132861 patent/WO2023092342A1/zh unknown
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160182920A1 (en) * | 2014-12-18 | 2016-06-23 | Konkuk University Industrial Cooperation Corp | Error concealment method using spatial interpolation and exemplar-based image inpainting |
| CN108846816A (zh) * | 2018-06-20 | 2018-11-20 | 上海理工大学 | 一种密文域图像修复方法 |
| CN110659511A (zh) * | 2019-09-25 | 2020-01-07 | 中国人民解放军61646部队 | 一种遥感图像的信息隐藏处理方法及装置 |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117156182A (zh) * | 2023-10-27 | 2023-12-01 | 北京健滨医药科技有限公司 | 一种医疗远程智能监测方法及系统 |
| CN117156182B (zh) * | 2023-10-27 | 2024-01-23 | 北京健滨医药科技有限公司 | 一种医疗远程智能监测方法及系统 |
| CN117499551A (zh) * | 2024-01-02 | 2024-02-02 | 天津中荣印刷科技有限公司 | 一种基于二进制信号的加密防伪印刷方法 |
| CN117499551B (zh) * | 2024-01-02 | 2024-04-23 | 天津中荣印刷科技有限公司 | 一种基于二进制信号的加密防伪印刷方法 |
| CN117973496A (zh) * | 2024-02-19 | 2024-05-03 | 苏州慧维智能医疗科技有限公司 | 局部病变医疗图像生成模型的训练方法、系统及应用 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2023092342A1 (zh) | 一种基于密文图像修复的医学图像隐私保护方法 | |
| DE102016002792B4 (de) | Biometrie-Geheimnis-Bindungsschema mit verbessertem Datenschutz | |
| Al-Haj et al. | Secured telemedicine using region-based watermarking with tamper localization | |
| CN105871543B (zh) | 多数据拥有者背景下基于属性的多关键字密文检索方法 | |
| CN110337000B (zh) | 基于双二叉树扩展和公钥加密的加密域可逆信息隐藏方法 | |
| Ustubioglu et al. | A new medical image watermarking technique with finer tamper localization | |
| Kandar et al. | Visual cryptography scheme for color image using random number with enveloping by digital watermarking | |
| Pavithra et al. | A survey on the techniques of medical image encryption | |
| Hashim et al. | Based on iot healthcare application for medical data authentication: Towards a new secure framework using steganography | |
| CN106709854A (zh) | 基于猫脸变换和混沌的图像信息融合加密方法 | |
| CN113987553A (zh) | 一种基于密文图像修复的医学图像隐私保护方法及系统 | |
| CN110289060A (zh) | 一种基于随机Merkle树的链下个人医疗数据存储方法 | |
| Liu et al. | Polynomial-based extended secret image sharing scheme with reversible and unexpanded covers | |
| Wang et al. | TPE-ISE: approximate thumbnail preserving encryption based on multilevel DWT information self-embedding | |
| CN115311119B (zh) | 一种抗几何攻击的立体图像零水印嵌入与提取方法 | |
| CN109347833A (zh) | 基于属性加密的用于机器学习环境下的访问控制方法和系统 | |
| CN106875324B (zh) | 基于sbde的无损图像信息隐藏方法 | |
| CN113382128B (zh) | 一种位平面压缩的加密图像可逆信息隐藏方法 | |
| CN114912631A (zh) | 一种基于区块链医疗数据共享的联邦学习方法 | |
| Paikaray et al. | Reversible selective embedding for DICOM image security and integrity using visual cryptography | |
| Ustubioglu et al. | IWT-MDE based reversible thermal image watermarking enhanced with secret sharing mechanism | |
| Bayari et al. | A novel security framework for medical data in IoT ecosystems | |
| Malipatil et al. | An efficient 3D watermarking algorithm for 3D mesh models | |
| Sajitha et al. | Analysis of Various Visual Cryptographic Techniques and Their Issues Based on Optimization Algorithms | |
| CN113206926B (zh) | 一种基于三像素点加密以平衡图像隐私性和可用性的方法 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 21965080 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |