Big data based digital scanner imaging management method

Abstract

The application discloses a digital scanner imaging management method based on big data, and relates to the technical field of imaging management, aiming at efficiently storing and managing medical images; medical images photographed by digital scanners of the same type and the same specification are acquired from a cloud database, the images are segmented and an associated pixel block set is extracted; then singular value decomposition is performed on each pixel block, a corresponding pixel fingerprint is generated, and a standard pixel block library is established; the similarity between the pixel blocks of a new image and the standard pixel block library is compared, redundant pixel blocks are screened out, and the redundant pixel blocks are replaced by storage pointers of standard blocks, so that the redundant storage of image data is realized; the storage efficiency of medical images can be optimized, and redundancy can be reduced, thereby providing an intelligent management scheme for medical image management.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of imaging management, and in particular relates to a digital scanner imaging management method based on big data. BACKGROUND

[0002] With the rapid economic development and continuous progress of information technology, the medical industry is gradually moving towards digitization and intelligentization. As an important tool for medical image acquisition, digital scanners are increasingly widely used in various fields such as diagnosis, treatment, and research.

[0003] In the field of medical image storage and management, the prior art usually relies on directly storing complete image files or using traditional compression algorithms such as JPEG or lossless formats to reduce storage occupancy. However, these methods have many shortcomings when facing a large number of similar medical images generated by the same type of digital scanner. Firstly, the prior art lacks the ability to identify fine-grained redundancies in image content, resulting in the repeated storage of a large number of pixel regions that are similar or highly similar, wasting cloud database storage space and increasing data management and transmission costs. Secondly, the prior art uses traditional feature extraction methods such as histograms or texture analysis, which cannot capture the inherent structural similarity of pixel blocks, resulting in low deduplication accuracy, either deleting important details or retaining too much redundancy, affecting storage efficiency and data integrity. Thirdly, the prior method often lacks a dynamic optimization mechanism in storage management. When new images are stored, it cannot intercept and process redundancies in real time, but relies on batch processing later, which increases computational delay and resource consumption, which is not conducive to real-time applications in a big data environment. Fourthly, traditional compression techniques can reduce file size, but they introduce information loss or are not suitable for the high-fidelity requirements of medical images.

[0004] To solve the above problems, the present application provides a digital scanner imaging management method based on big data. SUMMARY

[0005] In view of the shortcomings of the prior art, the present application provides a digital scanner imaging management method based on big data, which solves the problems of storage space waste, low processing efficiency, insufficient redundancy management, and poor economy of the prior art.

[0006] The object of the present application can be achieved by the following technical solutions:

[0007] The digital scanner imaging management method based on big data comprises:

[0008] Step one, obtaining a plurality of medical images of the same type and taken by the same type of digital scanner from a cloud database, performing segmentation operation on each medical image, and determining the pixel block set associated with each medical image;

[0009] Step two, based on the determined pixel block set, performing singular value decomposition on any pixel block to form a singular value vector associated with the corresponding pixel block, and generating a pixel fingerprint associated with the corresponding pixel block based on the determined singular value vector;

[0010] Step three, obtaining the pixel block set associated with each of the plurality of medical images and the pixel fingerprint associated with each pixel block, determining redundant pixel blocks, and performing de-duplication on the redundant pixel blocks and storing them in the pre-constructed standard pixel block library, marking the stored pixel blocks as standard pixel blocks, and extracting the storage pointers of the standard pixel blocks for display;

[0011] Step four, intercepting a medical image pre-stored in the cloud database, determining the pixel block set associated with the medical image and the pixel fingerprint of each pixel block, and based on the determination result of the pixel fingerprint and each pixel block in the standard pixel block library, screening redundant pixel blocks;

[0012] Extracting the storage pointers of the standard pixel blocks corresponding to the redundant pixel blocks in the standard pixel block library, replacing the redundant pixel blocks, and storing the medical image after the replacement operation in the cloud database.

[0013] As a further scheme of the present application, in step one, the specific way of determining the pixel block set associated with each medical image is:

[0014] Obtaining any one medical image Q, and determining the pixel size of the medical image;

[0015] Dividing the medical image Q into a plurality of pixel blocks of 8*8 pixels that do not overlap each other, counting the total number of pixel blocks, and denoting it as j;

[0016] Sorting the j pixel blocks in raster order, denoted as the pixel block set associated with the medical image Q, {q1, q2,..., qj};

[0017] Similarly, the pixel block sets associated with the remaining medical images are determined.

[0018] As a further scheme of the present application, in step two, the specific way of performing singular value decomposition on any pixel block based on the determined pixel block set to form a singular value vector associated with the corresponding pixel block is:

[0019] Obtaining the pixel block set {q1, q2,..., qj} of the medical image Q;

[0020] Extracting the pixel block qi, wherein i is a count index, 1≤i≤j;

[0021] Regarding the pixel block qi as an 8-row 8-column matrix, wherein each pixel point in the pixel block qi is regarded as a matrix element;

[0022] Perform singular value decomposition on pixel block qi to obtain qi = Ui * ∑i * Vi T Where Ui is an 8x8 orthogonal matrix, ∑i is an 8x8 diagonal matrix whose diagonal elements contain singular values, and the singular values ​​are arranged in descending order as σ1, σ2, ..., σ8, Vi T It is the transpose of another 8x8 orthogonal matrix Vi;

[0023] Obtain the singular values ​​σ1, σ2, ..., σ8 to form the singular value vector [σ1, σ2, ..., σ8] associated with the pixel block qi;

[0024] Similarly, determine the singular value vector of each pixel block in the pixel block set {q1,q2,...,qj}.

[0025] As a further aspect of the present invention, the specific method for generating the pixel fingerprint associated with the corresponding pixel block based on the determined singular value vector in step two is as follows:

[0026] Obtain the singular value vector [σ1,σ2,...,σ8] of pixel block qi;

[0027] use Calculate the Euclidean norm ‖σ‖;

[0028] Divide each element of the singular value vector [σ1,σ2,...,σ8] by the Euclidean norm ‖σ‖ to obtain the normalized singular value vector F_qi=[σ1 / ‖σ‖,σ2 / ‖σ‖,...,σ8 / ‖σ‖];

[0029] The normalized singular value vector is used as the pixel fingerprint associated with pixel block qi, and the pixel fingerprint of each pixel block is determined in this way.

[0030] As a further aspect of the present invention, in step three, the specific method for obtaining the set of pixel blocks associated with each of the several medical images and the pixel fingerprint associated with each pixel block, and for determining redundant pixel blocks, is as follows:

[0031] Obtain the set of pixel blocks for all medical images, randomly sort all pixel blocks, and denote the total set S = {p1, p2, ..., pn} consisting of all pixel block sets, where n is the total number of pixel blocks;

[0032] Extract the pixel fingerprint F_pm associated with any pixel block pm, where m is the counting index, 1≤m≤n;

[0033] Initialize an empty set, denoted as the redundant pixel block set H;

[0034] Initialize an empty list, denoted as the redundant pixel block list R;

[0035] acquire the pixel block pm and its associated pixel fingerprint F_pm from the total set S;

[0036] calculate the Euclidean distance between the pixel fingerprint F_pm and the pixel fingerprints of the rest of the pixel blocks in the total set S;

[0037] if the pixel fingerprint of any pixel block satisfies the condition that the Euclidean distance between the pixel fingerprint and the pixel fingerprint F_pm is less than θ, then mark this pixel block and the pixel block pm as redundant pixel blocks, and add them to the redundant pixel block list R together, wherein θ is a preset similarity threshold;

[0038] otherwise, do not process;

[0039] Similarly, determine the pixel blocks in the total set S whose pixel fingerprints have a Euclidean distance less than θ with the pixel fingerprint F_pm of the pixel block pm, and add them to the redundant pixel block list R, and combine all the pixel blocks in the redundant pixel block list R as a redundant pixel block grouping, and store it in the redundant pixel block set H;

[0040] perform a clearing operation on the redundant pixel block list R;

[0041] repeat the above steps to determine all redundant pixel block groupings.

[0042] As a further scheme of the present application, in step three, after determining all redundant pixel block groupings, extract any one redundant pixel block grouping from the redundant pixel block set H;

[0043] select a pixel block from the redundant pixel block grouping, wherein the pixel fingerprint of this pixel block has the smallest average Euclidean distance with the other pixel fingerprints in the redundant pixel block grouping;

[0044] store the selected pixel block in the pre-constructed standard pixel block library and mark it as a standard pixel block;

[0045] extract the storage pointer of the standard pixel block in the standard pixel block library, and store the storage pointer and the standard pixel block in the pre-constructed pointer mapping table in association, and display them;

[0046] repeat the above process to process all redundant pixel block groupings in the redundant pixel block set H.

[0047] As a further scheme of the present application, in step four, when a medical image is pre-stored in the cloud database, intercept the medical image and mark it as Q_new, determine the pixel block set {q_new1, q_new2,..., q_newj} of the medical image Q_new;

[0048] determining the pixel fingerprints associated with each pixel block in the pixel block set {q_new1, q_new2,..., q_newj};

[0049] initializing a redundant pixel block combination list L;

[0050] calculating the Euclidean distance between the pixel fingerprint of any pixel block q_newi and the pixel fingerprint of each standard pixel block in the standard pixel block library;

[0051] if the Euclidean distance between the pixel fingerprint of any standard pixel block and the pixel fingerprint of the pixel block q_newi is less than the similarity threshold θ, then the pixel block q_newi is marked as a redundant pixel block, combined with the standard pixel block, and added to the redundant pixel block combination list L;

[0052] otherwise, no processing is performed;

[0053] synchronously processing all pixel blocks in the pixel block set {q_new1, q_new2,..., q_newj} and outputting the redundant pixel block combination list L to determine all redundant pixel blocks in the medical image Q_new.

[0054] As a further scheme of the present application, in step four, after the redundant pixel block combination list L associated with the medical image Q_new is processed, any one redundant pixel block q_newi and its associated standard pixel block are obtained;

[0055] the storage pointer of this standard pixel block is obtained and denoted as ZZ_newi;

[0056] the redundant pixel block q_newi is replaced by the storage pointer ZZ_newi in the pixel block set {q_new1, q_new2,..., q_newj};

[0057] Similarly, the storage pointer of the standard pixel block is used to replace all redundant pixel blocks in the pixel block set {q_new1, q_new2,..., q_newj};

[0058] and the pixel block set associated with the medical image Q_new after the replacement operation is stored in the cloud database.

[0059] Advantages of the present application:

[0060] The present application obtains a large number of medical images of the same type and taken by the same specification digital scanner from the cloud database, first performs a segmentation operation to form a pixel block set, then generates a unique pixel fingerprint for each pixel block by singular value decomposition, thereby efficiently identifying redundant pixel blocks; secondly, a standard pixel block library is constructed and the redundant pixel blocks are stored after deduplication, only the storage pointers are retained, thereby reducing the storage space occupied by repeated data and reducing the storage cost of the cloud database; this method is suitable for big data scenarios, not only improves the storage and retrieval performance of medical images, but also maintains the integrity and usability of the data, thereby bringing an efficient and economical medical image data management solution to medical institutions;

[0061] The present application realizes efficient compression and stable representation of image features by segmenting medical images into pixel blocks, performing singular value decomposition to extract singular value vectors, and then generating pixel fingerprints through normalization; the core advantage lies in the inherent robustness of singular value decomposition, which can effectively resist noise, geometric deformation and compression interference, ensuring the reliability of feature extraction under medical image quality fluctuations; secondly, the influence of brightness and contrast changes is further eliminated through normalization processing, enhancing the consistency and adaptability of the fingerprint, so that it remains stable in complex clinical environments; in addition, the pixel blocks are converted into compact singular value vectors, significantly reducing the data dimension and improving the processing efficiency, which is suitable for real-time analysis of large-scale medical images, and the generated pixel fingerprints have high uniqueness and discrimination, thereby strengthening the security and integrity of medical data;

[0062] The present application identifies and groups similar pixel blocks into a redundant set by calculating the Euclidean distance between the pixel fingerprints of the medical image pixel blocks, thereby significantly reducing data redundancy, saving storage space and improving processing efficiency; secondly, the standard pixel block library is constructed by selecting the representative pixel block with the smallest average distance to other pixel fingerprints in each redundant group as the standard block for storage, and the compression and retrieval performance of large-scale medical image data is optimized through the method of fast association and access by pointer mapping table;

[0063] The present application quickly identifies similar redundant pixel blocks in the standard pixel block library by intercepting new medical images and comparing the Euclidean distance of pixel fingerprints, forms a redundant pixel block combination list, then replaces these redundant pixel blocks with the storage pointers of standard pixel blocks, and finally stores the modified pixel block set; this process reduces storage occupancy and cost by eliminating duplicate data, while reducing network transmission load, as storage pointers are smaller than actual pixel data, thereby optimizing bandwidth usage and ultimately bringing higher resource utilization and economy. BRIEF DESCRIPTION OF DRAWINGS

[0064] The present application will be further described below with reference to the accompanying drawings.

[0065] Figure 1 is a flowchart of the method described in the present application;

[0066] Figure 2 is a flowchart of the method described in the present application;

[0067] Figure 3 is a flowchart of the method described in the present application. DETAILED DESCRIPTION

[0068] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the protection scope of the present application.

[0069] Embodiment 1

[0070] A digital scanner imaging management method based on big data, as shown in Figure 1 The method comprises the following steps:

[0071] The present scheme: the digital scanner imaging management method based on big data is mainly used for storing and managing medical images taken by a digital scanner used in a medical image cloud storage scene, and aims to solve the problem of medical image storage overhead through big data analysis technology and image processing technology.

[0072] Firstly, a plurality of medical images of the same type and taken by the same specification of digital scanner are obtained from a cloud database, for example, chest X-rays taken by the same specification of digital scanner, so as to ensure the feature deviation caused by equipment difference.

[0073] Then, any one of the medical images is obtained, and the pixel size thereof is analyzed. The medical images taken by the digital scanner of the same specification generally have uniform pixel size.

[0074] Then, the medical image is subjected to segmentation processing, and is divided into a plurality of non-overlapping pixel blocks. Then, the pixel blocks are sorted in a raster order, that is, from top to bottom and from left to right, to form a pixel block set associated with the medical image.

[0075] After the pixel block set is determined, a singular value decomposition operation is performed on any one of the pixel blocks in the pixel block set, to determine a singular value vector associated with the pixel block after the singular value decomposition operation, and to continue to normalize the singular value vector to determine a pixel fingerprint associated with the pixel block.

[0076] The normalization processing aims to ensure that the generated pixel fingerprints are scale-invariant and insensitive to illumination and contrast changes, so that similar blocks have similar pixel fingerprints.

[0077] Next, pixel fingerprints are generated for all pixel blocks in the determined set of pixel blocks, and the pixel fingerprints associated with each of the determined sets of pixel blocks and each of the pixel blocks in all the sets of pixel blocks are obtained. Based on the associated analysis of the pixel fingerprints, redundant pixel blocks are determined, and grouping operations are performed on the redundant pixel blocks to obtain a plurality of redundant pixel block groups. Then, each redundant pixel block group is stored in the standard pixel block library constructed by the operator in advance.

[0078] It should be explained that the pixel blocks stored in the standard pixel block library are all marked as standard pixel blocks, and the standard pixel block library also associates and saves the storage pointers of the standard pixel blocks.

[0079] When a new medical image is uploaded to the cloud database, it is intercepted and processed to apply deduplication optimization; by comparing the pixel fingerprints of the new image with the standard pixel block library, redundant blocks are identified and replaced with corresponding pointers;

[0080] After replacement, the image is stored as a hybrid structure of pixel block sets and pointers, where the pointers point to the standard pixel blocks in the standard library; in this way, the amount of stored data is reduced, while the integrity of the image is maintained, because the pointers can be restored to standard pixel blocks when decoding; this method is similar to dictionary encoding, but is based on image features and is suitable for batch management of medical images.

[0081] Embodiment 2

[0082] This embodiment further discloses a method for generating pixel fingerprints associated with each pixel block based on embodiment 1, as shown in Figure 2 The specific steps include the following:

[0083] This embodiment further explains and illustrates the pixel fingerprints described in embodiment 1, as follows:

[0084] First, all medical images need to be segmented to determine the set of pixel blocks associated with each medical image.

[0085] A plurality of medical images of the same type and taken by the same type of digital scanner are obtained from the cloud database, and any one of the medical images is selected for example processing, and the remaining medical images are processed in the same and synchronous manner.

[0086] Firstly, the selected medical image is marked as Q, and the medical image Q is preliminarily read by using big data analysis technology to determine the pixel size of the medical image Q, that is, the pixel number of the width and height of the medical image Q.

[0087] The pixel size is determined to facilitate subsequent calculation of the number of pixel blocks, to ensure that the segmentation process covers the entire image and avoids missing or overlapping areas.

[0088] Next, the medical image Q is uniformly segmented into a plurality of 8x8 pixel blocks, and these pixel blocks do not overlap each other, that is, each pixel belongs to only one pixel block;

[0089] It should be noted that the pixel block of 8x8 pixels is used as the standardization processing method in the present scheme, but it is not limited to 8x8 pixels, and the operator can determine it according to the actual situation, for example, set it to 16x16 pixels;

[0090] The total number of pixel blocks segmented from the medical image Q is counted and denoted as j, and then the j pixel blocks determined are sorted in raster order, that is, from top to bottom and from left to right in the medical image Q, to form an ordered pixel block set represented as: {q1, q2,..., qj}.

[0091] Similarly, the same operation is performed on the remaining medical images according to the above method to determine the pixel block set associated with each medical image.

[0092] After determining the pixel block set associated with the medical image Q, singular value decomposition can be performed on all pixel blocks in the pixel block set to construct the singular value vector associated with the pixel block, and based on the normalization operation on the singular value vector, the pixel fingerprint associated with the medical image Q is determined, as follows:

[0093] From the pixel block set {q1, q2,..., qj} associated with any one of the determined medical images Q, any one pixel block qi is extracted, where i is a count index, and i takes a value in the range of 1 to j;

[0094] At this time, the pixel block qi is an 8x8 pixel region, which can be regarded as an 8x8 matrix, where each matrix element corresponds to each pixel point in the pixel block qi.

[0095] Next, singular value decomposition is performed on the pixel block qi to obtain the decomposition form as follows:

[0096] qi=Ui*∑i*Vi T ;

[0097] wherein Ui is an 8-by-8 orthogonal matrix representing left singular vectors, describing the row-wise features of the pixel block, ∑i is an 8-by-8 diagonal matrix whose diagonal elements contain singular values, and the singular values are arranged in descending order, denoted as: σ1, σ2,..., σ8, satisfying σ1≥ σ2≥,...,≥ σ8, and the non-diagonal elements of the diagonal matrix ∑i are zeros, Vi T is the transpose of another 8-by-8 orthogonal matrix Vi representing right singular vectors, describing the column-wise features of the pixel block.

[0098] The main features of the pixel block qi are captured by the singular value decomposition operation, and the singular values σ1 to σ8 represent the energy distribution of the pixel block: the larger singular values correspond to the main structure or texture information of the image block, and the smaller singular values correspond to the details, and the singular value decomposition operation has certain invariance to rotation and scaling, and is suitable for medical images.

[0099] Then, the singular values σ1, σ2,..., σ8 are sequentially extracted from the diagonal matrix ∑i, and a singular value vector [σ1, σ2,..., σ8] is formed, which is a compact representation of the pixel block qi and describes the feature descriptor of the pixel block, and is a dimension reduction processing of the pixel block qi.

[0100] Then, the singular value vector [σ1, σ2,..., σ8] associated with the pixel block qi is obtained, and the Euclidean norm ‖σ‖ of the singular value vector is calculated by using The Euclidean norm ‖σ‖ of the singular value vector associated with the pixel block qi is calculated, which reflects the total contrast and intensity variation of the pixel block, and the calculation of the Euclidean norm is the premise of normalization, which is used to eliminate the influence of absolute numerical value.

[0101] Each element in the singular value vector is divided by the Euclidean norm ‖σ‖, to obtain a normalized singular value vector after normalization, denoted as: F_qi=[σ1 / ‖σ‖,σ2 / ‖σ‖,...,σ8 / ‖σ‖].

[0102] Then, the normalized singular value vector F_qi=[σ1 / ‖σ‖,σ2 / ‖σ‖,...,σ8 / ‖σ‖] is taken as the pixel fingerprint associated with the pixel block qi.

[0103] The above steps are repeated to process all pixel blocks in the pixel block set {q1, q2,..., qj} in the same way, and the pixel fingerprints associated with all pixel blocks are determined.

[0104] Example 3

[0105] The embodiment continues to disclose a method for determining redundant pixel blocks based on pixel fingerprints and further determining standard pixel blocks based on the embodiment 2, as shown in the following: Figure 3 Specifically, the method comprises the following steps:

[0106] According to the embodiment 2 and the content described in the embodiment 1, the pixel block set associated with each of the extracted medical images can be obtained, all the pixel blocks in the pixel block sets are randomly sorted, and the results after the sorting are summarized. The summarized results are recorded as a total set S={p1, p2,..., pn} associated with the medical images, wherein p1 to pn represent the first to the n-th pixel blocks, and n is the total number of pixel blocks.

[0107] It should be explained that the random sorting of all the pixel blocks in the pixel block sets ensures that each pixel block is treated equally, prevents the algorithm from preferentially processing some pixel blocks due to a fixed order, and improves the fairness and comprehensiveness of the redundancy detection.

[0108] Then, any one pixel block pm is extracted from the total set S={p1, p2,..., pn}, and the pixel fingerprint F_pm associated with the pixel block pm is obtained, wherein m is a count index, and 1≤m≤n.

[0109] The following steps are an example of processing the pixel block pm, and all the pixel blocks in the total set S={p1, p2,..., pn} are processed in the same way.

[0110] First, an empty set is initialized and recorded as a redundant pixel block set H, which is used to group redundant pixel blocks, and each group is actually a small set.

[0111] Then, an empty list is initialized and recorded as a redundant pixel block list R, which is used to temporarily store the currently discovered redundant pixel blocks in the iteration.

[0112] Next, the pixel blocks are selected one by one from the total set S, and the pixel fingerprints of the pixel blocks are obtained, and the pixel block pm is processed:

[0113] The Euclidean distance between the pixel fingerprint of the selected pixel block from the total set S and the pixel fingerprint of the pixel block pm is calculated.

[0114] The calculated Euclidean distance is compared with the similarity threshold θ preset by the operator, and if there is any pixel block in the total set S whose pixel fingerprint satisfies that the Euclidean distance with the pixel fingerprint F_pm is less than θ, then the pixel block and the pixel block pm are both marked as redundant pixel blocks and added to the redundant pixel block list R.

[0115] If the Euclidean distance between the pixel fingerprint of any one pixel block and the pixel fingerprint F_pm is greater than or equal to θ, the processing of this pixel block is skipped, and the next pixel block is searched.

[0116] According to the above method, all the pixel blocks in the overall set S with the Euclidean distance less than θ from the pixel fingerprint F_pm of the pixel block pm can be determined and added to the redundant pixel block list R.

[0117] At this time, the redundant pixel block list R includes several redundant pixel blocks, and all of them are redundant pixel blocks associated with the pixel block pm, and the pixel block pm is also located in it.

[0118] All the redundant pixel blocks in the redundant pixel block list R at this time are combined as a redundant pixel block group (that is, a small set) and stored in the redundant pixel block set H (after storing the current redundant pixel block group in the redundant pixel block set H, the redundant pixel block list R is emptied to prepare for processing the next pixel block).

[0119] By repeating the above steps, all the redundant pixel block groups in the overall set S can be determined, and it should be noted that any one pixel block can only be a redundant pixel block in one redundant pixel block group.

[0120] After determining all the redundant pixel block groups in the overall set S, any one redundant pixel block group in the redundant pixel block set H is extracted, and a pixel block is selected from the redundant pixel block group.

[0121] The selection method is that the average Euclidean distance between the pixel fingerprint of the pixel block and other pixel fingerprints in the redundant pixel block group is the smallest.

[0122] Then, the pixel block is stored in the standard pixel block library pre-constructed by the operator and is marked as a standard pixel block, which is the only standard for the corresponding redundant pixel block group.

[0123] Then, the storage pointer (storage address of the standard pixel block in the standard pixel block library) of the standard pixel block in the standard pixel block library is obtained.

[0124] The extracted storage pointer is stored in the pre-constructed pointer mapping table in association with the corresponding standard pixel block and is displayed. In subsequent operations, the pointer mapping table can be used to quickly locate and replace redundant pixel blocks, improving data access efficiency.

[0125] The pointer mapping table is displayed to facilitate the operator to understand the redundancy removal process and support auditing and debugging.

[0126] Secondly, the pointer mapping table allows dynamic updates, such as when the standard pixel block library changes, only the storage pointer needs to be adjusted, without affecting the reference of the original data.

[0127] Finally, by repeating the above steps, it is ensured that all redundant pixel block groups in the redundant pixel block set H are processed, and the pointer mapping table and the standard pixel block library are perfected.

[0128] It should be noted that the standard pixel block library is not immutable, and the operator can set the update period, for example, update once a week, obtain the medical image data to be stored in the cloud database (do not repeat the medical image data that has been obtained), and determine the new standard pixel block.

[0129] Embodiment 4

[0130] This embodiment continues to disclose a method for de-duplication storage of pre-stored medical images to reduce storage pressure based on embodiment 3, which specifically includes the following:

[0131] After determining the standard pixel block library and the pointer mapping table, the pre-stored medical images in the cloud database are intercepted in real time, and any intercepted medical image is denoted as Q_new, and the following operations are performed on the medical image Q_new. It should be noted that the remaining medical images are operated in the same way.

[0132] According to the above embodiment, the pixel block set associated with the medical image Q_new can be determined, which is denoted as {q_new1, q_new2,..., q_newj}, and the pixel fingerprint associated with each pixel block in the pixel block set {q_new1, q_new2,..., q_newj} is determined.

[0133] Next, an empty list is initialized, denoted as a redundant pixel block combination list L, which is used to store redundant pixel block combinations.

[0134] All pixel fingerprints of the standard pixel blocks in the standard pixel block library are traversed and compared with the pixel fingerprint of the pixel block q_newi, wherein the pixel block q_newi is any one of the pixel blocks in the pixel block set {q_new1, q_new2,..., q_newj}.

[0135] If the Euclidean distance between the pixel fingerprint of any one of the standard pixel blocks in the standard pixel block library and the pixel fingerprint of the pixel block q_newi is less than the similarity threshold θ preset by the operator, the pixel block q_newi is marked as a redundant pixel block, and the redundant pixel block q_newi is combined with the standard pixel block associated with it and added to the pixel block combination list L.

[0136] According to the above method, all the pixel blocks in the pixel block set {q_new1, q_new2,..., q_newj} are processed in the same way to determine all the redundant pixel blocks in the pixel block set {q_new1, q_new2,..., q_newj} (medical image Q_new), and output the redundant pixel block combination list L.

[0137] Then, any one redundant pixel block q_newi and the standard pixel block associated with the redundant pixel block q_newi are extracted from the redundant pixel block combination list L.

[0138] Based on the pointer mapping table, the storage pointer of the standard pixel block in the standard pixel block library is obtained, and is recorded as ZZ_newi (indicating association with the redundant pixel block q_newi). The pixel block set {q_new1, q_new2,..., q_newj} is operated as follows:

[0139] The redundant pixel block q_newi in the pixel block set {q_new1, q_new2,..., q_newj} is replaced by the storage pointer ZZ_newi to obtain a pixel block set of a multi-data structure.

[0140] Similarly, all the redundant pixel blocks in the pixel block set {q_new1, q_new2,..., q_newj} are replaced by the storage pointers of the standard pixel blocks in the above manner, and the pixel block set obtained after the replacement operation is stored in the cloud database as the pixel block set associated with the medical image Q_new.

[0141] Some data in the formulas described above are dimensionless for numerical calculation, and the contents not described in detail in the specification all belong to the prior art known to those skilled in the art.

[0142] The above is only an example and description of the present application. Those skilled in the art can make various modifications or supplements to the described specific embodiments or use similar ways to replace them, as long as they do not deviate from the invention or exceed the scope defined by the present claims, and should belong to the protection scope of the present application.

[0143] It should be declared that all the user data collected in the present application is collected with the consent and authorization of the user. The use of user data is legal and compliant, and the use and processing of user data comply with relevant laws, regulations and standards in the relevant region.

Claims

1. A digital scanner imaging management method based on big data, characterized in that, The method includes: Step 1: Obtain several medical images Q of the same type and taken by the same type of digital scanner from the cloud database, perform segmentation operation on each medical image Q, and determine the set of pixel blocks {q1,q2,...,qj} associated with each medical image; Step two, based on the determined set of pixel blocks {q1,q2,...,qj}, performs singular value decomposition on any pixel block to construct the singular value vector associated with the corresponding pixel block. The specific method is as follows: Obtain the set of pixel blocks {q1,q2,...,qj} of the medical image Q; Extract pixel block qi, where i is the counting index, 1≤i≤j; View the pixel block qi as an 8x8 matrix, where each pixel in the pixel block qi is considered a matrix element; Performing singular value decomposition on pixel block qi yields qi = Ui * ∑i * Vi T Where Ui is an 8x8 orthogonal matrix, ∑i is an 8x8 diagonal matrix whose diagonal elements contain singular values, and the singular values ​​are arranged in descending order as σ1, σ2, ..., σ8, Vi T It is the transpose of another 8x8 orthogonal matrix Vi; Obtain the singular values ​​σ1, σ2, ..., σ8 to form the singular value vector [σ1, σ2, ..., σ8] associated with the pixel block qi; Similarly, determine the singular value vector of each pixel block in the pixel block set {q1,q2,...,qj}; The specific method for generating the pixel fingerprint associated with the corresponding pixel block based on the determined singular value vector is as follows: Obtain the singular value vector [σ1,σ2,...,σ8] of pixel block qi; use Calculate the Euclidean norm ‖σ‖; Divide each element of the singular value vector [σ1,σ2,...,σ8] by the Euclidean norm ‖σ‖ to obtain the normalized singular value vector F_qi=[σ1 / ‖σ‖,σ2 / ‖σ‖,...,σ8 / ‖σ‖]; The normalized singular value vector is used as the pixel fingerprint associated with pixel block qi, and the pixel fingerprint of each pixel block is determined in this way. Step 3: Obtain the set of pixel blocks associated with each of the medical images and the pixel fingerprint associated with each pixel block. The specific method for identifying redundant pixel blocks is as follows: Obtain the set of pixel blocks for all medical images, randomly sort all pixel blocks, and denote the total set S = {p1, p2, ..., pn} consisting of all pixel block sets, where n is the total number of pixel blocks; Extract the pixel fingerprint F_pm associated with any pixel block pm, where m is the counting index, 1≤m≤n; Initialize an empty set, denoted as the redundant pixel block set H; Initialize an empty list, denoted as the redundant pixel block list R; Obtain the pixel block pm and its associated pixel fingerprint F_pm from the overall set S; Calculate the Euclidean distance between the pixel fingerprint F_pm and the pixel fingerprints of the remaining pixel blocks in the overall set S; If the pixel fingerprint of any pixel block satisfies that the Euclidean distance between it and the pixel fingerprint F_pm is less than θ, then this pixel block and pixel block pm are marked as redundant pixel blocks and added together to the list of redundant pixel blocks R, where θ is a preset similarity threshold. Conversely, no action is taken; Similarly, identify all pixel blocks in the overall set S whose Euclidean distance between the pixel fingerprint and the pixel fingerprint F_pm of pixel block pm is less than θ, add them to the redundant pixel block list R, and combine all pixel blocks in the redundant pixel block list R into a redundant pixel block group, which is stored in the redundant pixel block set H. Perform a clear operation on the list of redundant pixel blocks R; Repeat the above steps to determine the grouping of all redundant pixel blocks; Redundant pixel blocks are deduplicated and stored in a pre-built standard pixel block library. The stored pixel blocks are marked as standard pixel blocks, and the storage pointer of the standard pixel block is extracted and displayed. Step 4: Intercept medical images pre-stored in the cloud database, determine the set of pixel blocks associated with the medical image and the pixel fingerprint of each pixel block, and filter redundant pixel blocks based on the judgment results of each pixel block in the standard pixel block library using the pixel fingerprints. Extract the storage pointers of the standard pixel blocks corresponding to the redundant pixel blocks in the standard pixel block library, replace the redundant pixel blocks, and store the medical images after the replacement operation in the cloud database.

2. The method according to claim 1, characterized in that, In step one, the specific method for segmenting each medical image and determining the set of pixel blocks associated with each medical image is as follows: Obtain any medical image Q and determine the pixel size of this medical image; Divide the medical image Q into several non-overlapping 8*8 pixel blocks, count the total number of pixel blocks, and denote it as j; Sort the j pixel blocks according to raster order, and denote it as the set of pixel blocks {q1,q2,...,qj} associated with the medical image Q; Similarly, determine the set of pixel blocks associated with the remaining medical images.

3. The method according to claim 2, characterized in that, In step three, after determining all redundant pixel block groups, any one redundant pixel block group is extracted from the redundant pixel block set H. Select a pixel block from the redundant pixel block group, where the pixel fingerprint of this pixel block has the smallest average Euclidean distance with the pixel fingerprints of other pixels in the redundant pixel block group. The selected pixel blocks are stored in a pre-built library of standard pixel blocks and marked as standard pixel blocks; Extract the storage pointer of the standard pixel block in the standard pixel block library, associate the storage pointer with the standard pixel block and store it in a pre-built pointer mapping table, and then display it; Repeat the above process to process all redundant pixel block groups in the redundant pixel block set H.

4. The method according to claim 3, characterized in that, In step four, when a medical image is pre-stored in the cloud database, the medical image is intercepted and recorded as Q_new, and the set of pixel blocks {q_new1,q_new2,...,q_newj} of the medical image Q_new is determined; Determine the pixel fingerprint associated with each pixel block in the pixel block set {q_new1,q_new2,...,q_newj}; Initialize the list of redundant pixel block combinations L; Calculate the Euclidean distance between the pixel fingerprint of any pixel block q_newi and the pixel fingerprints of each standard pixel block in the standard pixel block library; If the Euclidean distance between the pixel fingerprint of any standard pixel block and the pixel fingerprint of pixel block q_newi is less than the similarity threshold θ, then pixel block q_newi is marked as a redundant pixel block, associated with the standard pixel block, and added to the redundant pixel block combination list L. Conversely, no action is taken; Synchronize all pixel blocks in the pixel block set {q_new1,q_new2,...,q_newj} and output a list L of redundant pixel block combinations to determine all redundant pixel blocks in the medical image Q_new.

5. The method according to claim 4, characterized in that, In step four, after the redundant pixel block combination list L associated with the medical image Q_new is processed, any one of the redundant pixel blocks q_newi and its associated standard pixel block is obtained. Obtain the storage pointer of this standard pixel block, denoted as ZZ_newi; In the pixel block set {q_new1,q_new2,...,q_newj}, the redundant pixel block q_newi is replaced with the storage pointer ZZ_newi; Similarly, replace all redundant pixel blocks in the pixel block set {q_new1,q_new2,...,q_newj} with the storage pointers of the standard pixel blocks; The set of pixel blocks associated with the medical image Q_new after the replacement operation is stored in the cloud database.

Images