CN-113987553-B - Medical image privacy protection method and system based on ciphertext image restoration

CN113987553BCN 113987553 BCN113987553 BCN 113987553BCN-113987553-B

Abstract

The invention discloses a medical image privacy protection method and system based on ciphertext image restoration. The method comprises the steps of generating a mask image, an image to be repaired, a texture image, a JL transformation encryption result image and an encrypted medical image through the medical image, dividing the encrypted medical image into a plurality of image blocks based on the texture image, determining a limited source area of the block to be repaired in each image block based on the mask image and the texture image, determining an optimal patch set in the limited source area based on the JL transformation encryption result image, connecting adjacent patches in the optimal patch set to obtain a connected optimal patch set, and generating an encrypted image with embedded information based on the connected optimal patch set and a binary bit stream. The invention solves the problems that the existing medical image privacy protection method is low in safety and the existing ciphertext image restoration technology cannot be applied to image privacy protection due to poor restoration effect.

Inventors

KONG PING
WU TAO
LI AN
ZHOU LIANG
ZHOU YANLI
ZHANG JIANQING
CHEN LIFAN
WANG HONGJIE

Assignees

上海健康医学院
上海健康医学院

Dates

Publication Date: 20260421
Application Date: 20211124
Priority Date: 20211124

Claims (9)

1. A medical image privacy protection method based on ciphertext image restoration is characterized by comprising the following steps: dividing focus areas in the medical image through the trained image division neural network model to generate mask images; cutting off focus areas in duplicate images based on the mask images to generate images to be repaired, wherein the duplicate images are duplicate images of the medical images; convoluting and clustering the image to be repaired by using a linear spatial filter and a K-means clustering algorithm to obtain a texture image; Performing JL transformation encryption on each local image block in the image to be repaired to obtain a JL transformation encryption result image, wherein each local image block consists of each pixel in the image to be repaired and s-1 pixels adjacent to each pixel; Encrypting the medical image through an image encryption algorithm to obtain an encrypted medical image; Partitioning the encrypted medical image based on the texture image to obtain a plurality of image blocks; Determining a limited source area of a block to be repaired in each image block based on the mask image and the texture image, wherein the block to be repaired consists of all pixels with corresponding position values of 1 in the mask image in the image block; determining an optimal patch set in the limited source area based on the JL conversion encryption result image; linking adjacent patches in the optimal patch set to obtain a linked optimal patch set; an encrypted image with embedded information is generated based on the concatenated optimal patch set and the binary bit stream.
2. The medical image privacy protection method based on ciphertext image restoration of claim 1, wherein the partitioning the encrypted medical image based on the texture image to obtain a plurality of image blocks specifically comprises: Partitioning the encrypted medical image to obtain a plurality of initial image blocks; determining a direction mark based on the longest dimension of each initial image block; Calculating texture distances among all sub-image blocks in all initial image blocks of different direction marks based on the texture image; and partitioning each initial image block based on the texture distance.
3. The method of claim 1, wherein the restricted source region of the block to be repaired comprises a restricted source region of a trusted block to be repaired and a restricted source region of an untrusted block to be repaired.
4. A medical image privacy protection method based on ciphertext image restoration as defined in claim 3, wherein the determination of the restricted source region of the trusted block to be restored comprises: Comparing texture distances between each trusted block to be repaired in the encrypted medical image and other trusted blocks, wherein the trusted block to be repaired consists of all pixels which have corresponding position values of 1 in the mask image and the number of which is less than half of the total number of pixels in the image block, and the other trusted blocks consist of all pixels which have corresponding position values of 0 in the mask image and the number of which is more than half of the total number of pixels in the image block; And forming a good pixel in all the trusted blocks with texture distances smaller than a given difference threshold value between the trusted blocks and the trusted blocks to be repaired into a limited source region of the trusted blocks, wherein the good pixel is a pixel with a value of 0 at a corresponding position of a mask image.
5. The medical image privacy protection method based on ciphertext image restoration of claim 4, wherein the limited source area of the untrusted to-be-restored block is an area formed by intact pixels in the untrusted to-be-restored block, limited source areas of adjacent trusted to-be-restored blocks of the untrusted to-be-restored block and intact pixels in adjacent untrusted blocks of the untrusted to-be-restored block, and the untrusted to-be-restored block is formed by all pixels with corresponding position values of 1 in a mask image in an image block and more than half of the total number of pixels in the block.
6. The medical image privacy protection method based on ciphertext image restoration according to claim 1, wherein the determining an optimal patch set in the limited source area based on the JL transform encryption result image specifically comprises: Calculating the priority of all nodes in the limited source area; Accessing each node according to the priority, and calculating texture weight overhead between a local image block of each node and all patches in a limited source region of an image block where each node is located based on the JL transformation encryption result image; and determining the optimal patch set according to the texture weight overhead.
7. The medical image privacy protection method based on ciphertext image restoration of claim 6, wherein the linking the adjacent patches in the optimal patch set to obtain a linked optimal patch set specifically comprises: Determining a boundary of a minimum texture error in an overlapping region of adjacent nodes based on the texture image; and connecting adjacent patches according to the boundary to obtain the connected optimal patch set.
8. The medical image privacy protection method based on ciphertext image restoration according to claim 1, wherein the generating an encrypted image with embedded information based on the joined optimal patch set and a binary bit stream specifically comprises: encoding ciphertext values and position information of all pixels to be repaired in the encrypted medical image into a binary bit stream; replacing local image blocks of all nodes in the encrypted medical image with the optimal patch set after connection to obtain an encrypted restoration result image And reversibly embedding the binary bit stream into the encrypted restoration result image by a reversible information hiding technology with separable ciphertext domains to obtain an encrypted image with embedded information.
9. A medical image privacy protection system based on ciphertext image restoration, comprising: the segmentation module is used for segmenting focus areas in the medical image through the trained image segmentation neural network model to generate mask images; The image generation module to be repaired is used for cutting off focus areas in duplicate images based on the mask images to generate images to be repaired; The texture image generation module is used for carrying out convolution and clustering on the image to be repaired through a linear spatial filter and a K-means clustering algorithm to obtain a texture image; The JL conversion encryption result image generation module is used for carrying out JL conversion encryption on each local image block in the image to be repaired to obtain a JL conversion encryption result image, wherein the local image block consists of each pixel in the image to be repaired and s-1 pixels adjacent to each pixel; The medical image generation module is used for encrypting the medical image through an image encryption algorithm to obtain an encrypted medical image; The blocking module is used for blocking the encrypted medical image based on the texture image to obtain a plurality of image blocks; The limiting source region determining module is used for determining a limiting source region of a block to be repaired in each image block based on the mask image and the texture image, wherein the block to be repaired consists of all pixels with the corresponding position value of 1 in the mask image in the image block; the optimal patch set determining module is used for determining an optimal patch set in the limited source area based on the JL transformation encryption result image; the linking module is used for linking adjacent patches in the optimal patch set to obtain a linked optimal patch set; and the encrypted image generation module with the embedded information is used for generating an encrypted image with the embedded information based on the optimal patch set after the linking and the binary bit stream.

Description

Medical image privacy protection method and system based on ciphertext image restoration Technical Field The invention relates to the technical field of medical information security protection, in particular to a medical image privacy protection method and system based on ciphertext image restoration. Background With the endless cases caused by privacy disclosure, people's awareness of privacy protection is gradually enhanced. However, many medical institutions currently store medical images such as X-rays or B-rays taken during patient diagnosis directly in their separate databases without protection, and the method of storing patient images is convenient to manage and recall, but is in fact a safety risk. Although the medical data information of many medical institutions is stored in the internal local area network of the medical institutions and is basically not connected to the external internet, the medical data information seems to be very safe, with the development of medical informatization, the medical institutions need to implement networking to share medical information, remotely conduct medical diagnosis and other changes conforming to the medical informatization in the future. The medical data information which is seemingly safe is due to the sealing property, and once the medical institution networking situation is entered, the originally safe medical information is exposed on the public internet, the safety problem of the medical data which does not take protective measures on the internet cannot be guaranteed, and hackers can attack and illegally steal the valuable medical information. Medical images are one of the important medical data information for which it is necessary to take safety precautions. The main methods for protecting the privacy of medical images include an image encryption algorithm and a clear text image restoration technology. Most of image encryption algorithms cannot effectively resist attack of selecting plaintext, and a hacker is likely to crack the encrypted image, so that a good effect cannot be achieved in terms of safety, an extra database is needed for storing the cut focus image based on the plaintext image restoration technology, the cut focus image is directly transmitted to an image processor, the cut focus image does not contain focus information, but the image processor can still deduce some illness states from the cut focus image, and the risk of revealing privacy of a patient exists, so that the safety of the method is not high. The ciphertext image restoration technology is to reconstruct a damaged area (an area to be restored) in a ciphertext image by utilizing a perfect area in the ciphertext image (the image encrypted by an image encryption algorithm), and the restored area after the ciphertext image restoration result is decrypted achieves a visually plausible effect. The method has the characteristics that damaged image restoration work can be completed on the premise that an image processor is completely unaware of image content information, dishonest image processors can be prevented, and the safety defect of the existing method is overcome. Existing methods of ciphertext image restoration can be categorized into diffusion-based and patch-based methods. The ciphertext image restoration algorithm based on diffusion is realized by spreading information inwards from the neighborhood of the area to be restored along the direction of equal illumination, but for a thick area to be restored, the deeper the information is, the lower the reliability is, so that the restoration effect is poor, while the ciphertext image restoration algorithm based on patches is to search patches with good similarity with the blocks to be restored in images to fill the blocks to be restored, but because the existing algorithm does not consider coordination among the patches, obvious visual conflict exists between adjacent patches of restoration results, and the aim of difficultly perceiving that the image is modified cannot be achieved, so that the current ciphertext image restoration technology cannot be applied to protecting the privacy of medical images. Disclosure of Invention The invention aims to provide a medical image privacy protection method and a system based on ciphertext image restoration, which are used for solving the problems that the existing medical image privacy protection method is low in safety and the existing ciphertext image restoration technology cannot be applied to image privacy protection due to poor restoration effect. In order to achieve the above object, the present invention provides the following solutions: A medical image privacy protection method based on ciphertext image restoration comprises the following steps: dividing focus areas in the medical image through the trained image division neural network model to generate mask images; cutting off focus areas in duplicate images based on the mask images to generate images to be repaired, whe