Medical image management method and system based on photo-electromagnetic hybrid hierarchical storage

Abstract

The invention relates to the technical field of medical information storage and management, and discloses a medical image management method and system based on photo-electromagnetic hybrid hierarchical storage, and the method comprises the steps: storing medical image data into a high-performance layer, and generating a redundancy check block and a storage index; and dynamically adjusting the storage level according to the storage priority of the image data, optimizing the storage position of the redundancy check block, and updating the storage index. And when the user accesses, the system queries the storage index, positions the image data and executes access operation according to the state of the redundancy check block. And when the data is changed, synchronously updating the storage index and the redundancy check block. Integrity detection is executed regularly, and if data damage is found, erasure codes are used for recovery. And executing a life cycle management strategy on the data which is not accessed for a long time, and adjusting the redundancy check block and updating the storage index during migration. The storage efficiency is improved, the data integrity is ensured and the long-term storage cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of medical information storage and management, and specifically to a medical image management method and system based on optical-electromagnetic hybrid hierarchical storage. Background Art

[0002] With the large-scale growth of medical image data, the demand of medical institutions for image storage systems has been increasing day by day. Medical image data mainly comes from examination equipment such as CT, MRI, X-ray, ultrasound, etc. These data usually have the characteristics of large data volume, high access demand, and long-life cycle storage. How to store and manage these data efficiently and securely has become a technical problem to be solved urgently. Most traditional medical image storage methods rely on disk storage systems, and such systems face certain challenges in terms of storage capacity and data access speed. Especially when storing large-scale and high-resolution image data, the efficiency and reliability of traditional storage methods cannot meet the requirements of the medical industry for data storage, fast access, and high availability. To solve this problem, in recent years, medical image management systems have gradually developed towards multi-level storage solutions. In particular, the hierarchical storage method combined with the optical-electromagnetic hybrid storage architecture has become an important development direction of storage systems.

[0003] Although the existing technologies have made improvements in storage levels and data management, there are still some deficiencies. Most existing storage systems rely on fixed storage priorities and thresholds to determine data migration, and these fixed strategies often cannot effectively adapt to the dynamic changes and access patterns of medical image data. Especially in the case of huge data volume, how to accurately predict the access requirements of data in real time and make efficient storage level adjustments based on this is still a technical problem. In addition, the existing data redundancy and recovery mechanisms are relatively simple and cannot cope with complex storage failures and data corruption situations, resulting in the system may not be able to recover data in time when a failure occurs, affecting the stability and reliability of medical services. Summary of the Invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the technical problems solved by the present invention are: the lack of intelligent optimization in the existing medical image data storage level, insufficient data integrity guarantee, inefficient storage migration strategy, and high long-term storage cost.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: a medical image management method based on optical-electromagnetic hybrid hierarchical storage, including: Receiving medical image data, generating redundant check blocks by using erasure code technology, and generating a storage index; Calculate the storage priority of medical image data according to the attributes of medical image data; the storage priority includes calculating the access priority using a time series prediction model, and calculating the storage priority by weighted summation of the access priority and clinical importance; Dynamically adjust the storage level of medical image data according to the prediction result, and synchronously adjust the redundant check blocks and update the storage index; Regularly detect the integrity of medical image data. When data corruption is detected, use the redundant check blocks to recover the corrupted medical image data; when the medical image data is changed, synchronously update the redundant check blocks; The dynamic adjustment includes adjusting using a migration scheduling strategy optimized by a reinforcement learning algorithm; in the migration scheduling strategy, evaluate the data state according to the storage priority, make a discrimination according to the set migration threshold and priority conditions, automatically trigger the migration scheduling, evaluate the initial state of the data using a reinforcement learning algorithm, randomly select migration actions and select the action with the maximum Q value through an exploration-exploitation strategy to determine the migration decision, and adaptively calculate the reward function according to the multi-factor migration effect, update the Q value and continuously optimize the migration decision.

[0007] As a preferred solution of the medical image management method based on optical-electromagnetic hybrid hierarchical storage according to the present invention, wherein: the medical image data includes original medical images, image metadata, image processing data, image-related documents, and image storage security information; The attributes of the medical image data include image type, creation time, access frequency, clinical importance, access permission; the hierarchical storage includes high-performance layer SSD solid-state storage, medium-performance layer HDD magnetic storage, and low-performance layer optical storage.

[0008] As a preferred solution of the medical image management method based on optical-electromagnetic hybrid hierarchical storage according to the present invention, wherein: the generating of the redundant check blocks using the erasure code technology includes dividing the original medical image data into data blocks ; Input the data blocks , encode the data blocks using a coding matrix over a finite field, and generate redundant check blocks ; Generate a storage index, record the storage locations of each data block and its redundant check block, generate unique identifiers for each data block and redundant check block by performing a hash calculation on each data block and redundant check block to generate UIDs; allocate storage paths for the data blocks and redundant check block storage paths, record the physical locations of the data blocks and redundant check blocks, and have unique paths in the three types of storage devices SSD, HDD, and optical storage. Update the UIDs, storage paths, storage levels, and data size information of the data blocks and redundant check blocks to the storage index table.

[0009] As a preferred solution of the medical image management method based on optical-electromagnetic hybrid hierarchical storage described in the present invention, wherein: calculating the storage priority of medical image data includes predicting the access frequency of the data, using an adaptive time series prediction model LSTM network to predict future access requirements based on historical access data; taking the historical access frequency of the data block as the input sequence. , LSTM predicts the future through past access patterns. The access frequency within the next step; calculate the storage priority, according to the access frequency predicted by LSTM. and the clinical importance of the medical image. , comprehensively calculate the storage priority. , It is expressed as the weighted sum of the access frequency and clinical importance.

[0010] As a preferred solution of the medical image management method based on optical-electromagnetic hybrid hierarchical storage described in the present invention, wherein: the migration scheduling strategy includes storage level division, setting migration conditions, and a delayed migration trigger mechanism; When the predicted access frequency of the data block. is higher than the threshold. , and the storage priority. reaches a predetermined trigger value. , the data will be migrated from the low-performance layer to the medium-performance layer; wherein, represents the access threshold. represents the data migration trigger value; When the storage priority of the data. reaches a predetermined trigger value. and the priority. , the data will be migrated from the low-performance layer or the medium-performance layer to the high-performance layer; wherein, represents the clinical urgency threshold; Set a threshold. , at each time step. Check whether the data needs to be migrated. If the access frequency and priority of the data exceed the set threshold within consecutive. time steps, trigger the migration scheduling; After triggering the migration scheduling, a reinforcement learning algorithm is used to optimize the migration decision according to the predicted access frequency and priority; the Q-learning model determines whether to perform migration by learning the migration benefits of the system.

[0011] As a preferred solution of the medical image management method based on optical-electromagnetic hybrid hierarchical storage according to the present invention, wherein: the reinforcement learning algorithm includes, represents the value of taking action in state , represents the immediate reward obtained after executing the action, represents the discount factor, represents the learning rate, represents that in the next state , select the action that can obtain the maximum Q value; At each time step , execute the migration decision through the following steps: State evaluation, evaluate the current state of the data according to the current access frequency , storage priority and clinical importance ; Determine the migration decision by selecting the action with the maximum Q value, using the exploration-exploitation strategy, represents the probability of performing exploration; use to select a random action, and select the action with the maximum Q value with a probability of ; Execute migration, and migrate the data to the corresponding storage level according to the selected action; After executing the action, dynamically calculate the reward according to performance improvement, resource saving and clinical priority ; Update the Q value according to the Q-learning update formula to learn the best migration strategy.

[0012] As a preferred solution of the medical image management method based on optical-electromagnetic hybrid hierarchical storage according to the present invention, wherein: the data recovery mechanism includes, when receiving a user's medical image data access request, reading the redundant check block and comparing it with the original data, and using erasure code technology to check the check result between the medical image data block and the redundant check block; calculating the hash value of the data block through the hash algorithm and comparing it with the hash value in the redundant check block to verify whether the data is damaged; When data corruption is detected, a redundancy recovery algorithm is used to recover the corrupted data blocks through the stored redundancy check blocks; for medical image data that has not been accessed for a long time, a lifecycle management policy will be executed to migrate the infrequently accessed data to low-cost storage and synchronously adjust the redundancy check blocks.

[0013] A medical image management system based on optical, electromagnetic, and hybrid hierarchical storage, wherein: A data reception and storage module that receives medical image data and stores it in the high-performance layer, generates redundancy check blocks using erasure coding technology, and generates a storage index; A storage priority calculation and dynamic layer adjustment module that calculates the storage priority of medical image data based on the image attributes of the medical image data, dynamically adjusts the storage layer of the medical image data using a hierarchical storage weight model, and synchronously adjusts the storage location of the redundancy check blocks and updates the storage index; A user access management module that receives a user access request, queries the storage index through the storage management control layer, locates the current storage location of the medical image data, verifies the status of the redundancy check block associated with the data, and executes the user request operation based on the status of the redundancy check block and its storage layer; An update module that, when the medical image data is changed, updates the storage index using the storage management control layer and synchronously updates the redundancy check blocks; A recovery module that periodically performs integrity detection on medical image data based on the storage index and the erasure coding redundancy policy, and when data corruption is detected, uses erasure coding technology to recover abnormal medical image data; A lifecycle management module that executes a lifecycle management policy on medical image data that has not been accessed for a long time, and synchronously adjusts the stored redundancy check blocks and updates the storage index during migration.

[0014] A computer device, comprising: a memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of any one of the methods of the present invention are implemented.

[0015] A computer-readable storage medium, on which a computer program is stored, characterized in that: when the computer program is executed by a processor, the steps of any one of the methods of the present invention are implemented.

[0016] Advantages of the present invention: The medical image management method based on optical, electromagnetic, and hybrid hierarchical storage provided by the present invention predicts the storage priority based on the LSTM model and combines the hierarchical storage weight model to achieve dynamic adjustment of the storage hierarchy of medical image data, avoid resource waste, and improve the intelligent level of storage management. Secondly, the erasure code technology is used to distribute redundant check blocks among cross-media storages to improve data integrity and ensure that image data can still be recovered in case of storage medium failure. For user access, the storage management control layer can intelligently query the storage index, quickly locate the image data, and optimize the access path in combination with the storage hierarchy to improve the efficiency of image retrieval. When the data changes, the system synchronously updates the storage index and redundant check blocks to ensure data consistency. By integrating intelligent prediction, dynamic storage, data protection, and access optimization technologies, an efficient, secure, and scalable medical image storage management solution is constructed. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is the overall flowchart of a medical image management method based on optical, electromagnetic, and hybrid hierarchical storage provided for the first embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0020] Example 1, referring to Figure 1 , which is an embodiment of the present invention, provides a medical image management method based on optical, electromagnetic, and hybrid hierarchical storage, including: S1: Receive medical image data and store it in the high-performance layer, generate redundant check blocks using erasure code technology, and generate a storage index.

[0021] Furthermore, the medical image data includes original medical images, image metadata, image processing data, image-related documents, and image storage security information; Medical image data attributes include image type, creation time, access frequency, clinical importance, and access permissions; hierarchical storage includes high-performance layer SSD solid-state storage, medium-performance layer HDD magnetic storage, and low-performance layer optical storage.

[0022] The system divides medical image data into multiple data blocks and uses erasure coding technology (such as Reed-Solomon coding) to generate redundant check blocks. These redundant check blocks are stored together with the original data blocks and can be used to recover data in case of data loss or corruption. Data block division and redundant block generation process Data division: The original medical image data is divided into data blocks , and the size of each data block is , that is . Redundant check block generation: Use the Reed-Solomon erasure coding algorithm to generate redundant check blocks , and each redundant check block is calculated by combining the contents of the data blocks. Reed-Solomon coding process: Input data blocks . Use the coding matrix over a finite field to encode these data blocks to generate redundant check blocks . The specific formula is as follows:

[0023] where RS(·) represents the process of encoding data blocks using the Reed-Solomon algorithm.

[0024] Relationship between redundant check blocks and data blocks: Redundant check blocks can recover the lost data when some data blocks are lost or damaged. For example, if and are damaged, the system can use the redundant check block to reconstruct.

[0025] Generate storage index After generating data blocks and redundant check blocks, the system needs to generate a storage index for these data. This index is used to record the storage locations of each data block and its redundant check blocks, so that the locations of data and redundant check blocks can be quickly located during subsequent data access or recovery.

[0026] Generate unique identifier (UID): Generate a unique identifier (UID) for each data block and redundant check block. The UID can be generated by performing a hash calculation on each data block or redundant check block. Assume the hash value of each data block is , and the hash value of each redundant check block is , then:

[0027] Among them, represents performing a hash operation on a data block or a redundant check block to generate a unique identifier.

[0028] Generating a storage path: The storage location of each data block and redundant check block needs to be registered in the storage system. Assuming that each storage device (SSD, HDD, optical storage, etc.) has a unique path, the system will allocate a storage path for each data block and redundant check block and , that is:

[0029] The storage path records the physical locations of the data block and the redundant check block.

[0030] Updating the storage index table: Update information such as the UID, storage path, storage level, and data size of each data block and redundant check block into the storage index table. The storage index table contains detailed information about each data block and its redundant check block.

[0031] The location storage of the redundant check block is also recorded in the storage index. This ensures that when data is lost or damaged, the location of the redundant check block can be quickly located to recover the data.

[0032] Storage index table update: The storage index table records detailed information about each data block and redundant check block, including UID, storage path, storage level, data size, etc. Generating a storage index includes: generating a unique identifier UID and generating a storage index.

[0033] Generating the unique identifier UID includes: extracting medical image data elements from medical image data, using a hash algorithm to calculate the unique hash value of the medical image data, and splicing the hash value with the medical image data elements to generate the UID.

[0034] Generating a storage index includes: creating a new medical image data entry in the storage index table and recording the unique identifier (UID) of the medical image data, storage level, physical storage path, medical image data size, creation time, medical image data format, and the storage location of the redundant check medical image data.

[0035] It should be noted that by combining hierarchical storage with erasure coding technology, the efficient storage, reliability, and long-term availability of medical image data are ensured. The security and recoverability of long-term storage are ensured. The cross-storage medium distribution of medical image data blocks improves fault tolerance. Even if a storage medium fails, the medical image data can still be recovered using redundant check blocks, reducing the risk of medical image data loss. At the same time, the storage index mechanism manages medical image data through unique identifiers (UIDs), records the storage path, format, and redundant information, making the access, migration, and recovery of medical image data more efficient and accurate, and further enhancing the system's medical image data management capabilities and reliability.

[0036] S2: According to the attributes of medical image data, use an adaptive time series prediction model to calculate the storage priority of medical image data, and perform hierarchical storage according to the priority.

[0037] The influencing attributes include: access frequency, age of medical image data, clinical value score, regulatory requirements, and storage space pressure.

[0038] Calculating the storage priority of medical image data includes calculating the storage priority of medical image data according to the attributes of medical image data.

[0039] Predict the access frequency of data using the adaptive time series prediction model LSTM network. This model can predict future access requirements based on historical access data. Design the LSTM network. The input data is that the system takes the access frequency of each data block within a certain period of history as the input sequence , for example, the access frequency within the past time steps. The prediction model formula is expressed as:

[0040] Among them, represents the access frequency at time , is the prediction step size. LSTM predicts the access frequency within the future steps based on past access patterns. Calculate the storage priority. According to the access frequency predicted by LSTM and the clinical importance of medical images , comprehensively calculate the storage priority. The formula is expressed as:

[0041] Among them, and represent the importance of access frequency and clinical importance respectively. Usually, the clinical importance has a greater impact on emergency data (such as emergency images) and will be set to a higher value.

[0042] The system status includes the storage layer capacity utilization rate and the storage level where the current medical image data is located. The access log records the access times and the last access time of each medical image data at different moments.

[0043] Extract the access records from the access log, group them by UID, and count the access times in the past week. Obtain the creation time and the current time of the medical image data, and calculate the age of the medical image data. The clinical value is obtained by manual scoring by the directly using doctor. Obtain the available capacity of the current storage layer, and calculate the occupancy ratio of this medical image data to get the current storage layer utilization rate.

[0044] When the medical image data is archived to the optical storage, the number of redundant check blocks is reduced to the minimum standard. When the medical image data is archived to the medium performance layer, the number of redundant check blocks is reduced to the medium standard. When the medical image data is archived to the high performance layer, the number of redundant check blocks remains at the high standard.

[0045] Furthermore, by comprehensively analyzing the access frequency, age, clinical value score, regulatory requirements, and storage space pressure of medical image data, calculate the storage priority, and combine with the LSTM model to predict future storage requirements, so as to make an intelligent decision on the storage level of medical image data, avoid waste of storage resources and access latency. Use the Sigmoid transformation and linear scaling to ensure the stability of the predicted value, and adopt a hierarchical storage weight model to dynamically adjust the storage location to achieve efficient management of medical image data. At the same time, through the migration conditions and the mechanism to prevent frequent migration, avoid frequent changes of medical image data between storage levels, reduce the system I / O load, and improve the stability of medical image data management. In addition, adjust the redundant check block strategy, maintain a high redundancy in the high performance layer to ensure the reliability of medical image data, reduce redundancy in the low performance layer to occupy storage space, and finally optimize the storage cost, improve the long-term availability of medical image data and the overall performance of the storage system.

[0046] S3: According to the hierarchical storage result, dynamically adjust the storage level of the medical image data, and synchronously adjust the redundant check blocks and update the storage index; the dynamic adjustment includes using a reinforcement learning algorithm to make adjustments through a delayed migration trigger mechanism.

[0047] The delayed migration trigger mechanism includes that the dynamic adjustment of the storage level is based on the predicted storage priority and the predicted access frequency , and the system dynamically adjusts the storage level. The specific migration decision is as follows: Migrate from the optical storage to the medium performance layer (HDD magnetic storage) when the predicted access frequency of a certain data block is higher than the threshold , and the storage priority When the predetermined trigger value is reached, the data will be migrated from optical storage to HDD magnetic storage.

[0048]

[0049] Among them, and are threshold values, representing the conditions for migrating data from optical storage to medium-performance storage, depending on the change in access frequency and the importance of the data.

[0050] Migration from low-performance and medium-performance layers to high-performance layer (SSD solid-state storage): When the clinical importance of the data increases and the storage priority of the data reaches the predetermined trigger value and the priority is met, the data will be migrated from the low-performance and medium-performance layers to the high-performance layer (SSD).

[0051]

[0052] Among them, is the set critical priority, representing the condition that data needs to be migrated immediately to ensure fast reading. At this time, the clinical urgency of the data is relatively high, so faster access speed is required.

[0053] Storage layer division, migration conditions, and prevention of frequent migrations; To avoid overloading the storage system and reducing system performance caused by frequent storage migrations, this system implements a delayed migration trigger mechanism to reduce unnecessary migration operations and ensure a smooth transition of data between storage layers. 2.1 Continuous migration conditions and trigger strategies Migrations will not occur at each time step. Instead, a threshold is set, and migrations will only be triggered when the data continuously meets the migration conditions within a certain number of consecutive time steps. Continuous migration trigger conditions: The system will check at each time step whether the data needs to be migrated. If the access frequency and priority of the data exceed the set threshold within consecutive

[0054] If the conditions are true for consecutive steps, the system will perform data migration.

[0055] Predictive migration scheduling To further optimize migration decisions, Q-learning will help the system learn the best storage migration actions to take in a given state. The update formula of the Q-learning algorithm is as follows:

[0056] Among them, Indicates taking an action in the state with value. Indicates the immediate reward obtained after executing an action. Indicates the discount factor, representing the importance of future rewards. The higher the value, the more the system values future rewards.

[0057] Indicates the learning rate, controlling the speed of Q-value update. Indicates in the next state to select the action that can obtain the maximum Q-value.

[0058] At each time step , the migration decision is made through the following steps: State evaluation, evaluating the current state of the data according to the current access frequency , storage priority and clinical importance of the data.

[0059] Action selection, determining the migration decision by selecting the action with the maximum Q-value, using the epsilon-greedy strategy to balance exploration and exploitation, and selecting a random action with a probability of and selecting the action with the maximum Q-value with a probability of . Among them, exploration includes that the system randomly selects an action with a probability of

[0060] Here is the action at the current time , random indicates selecting a random action. Exploitation includes selecting the action with the maximum Q-value with a probability of , which is expressed as:

[0061] Here indicates selecting the action that maximizes the Q-value , that is, selecting the estimated optimal migration action in the current Q-value table.

[0062] Execute migration, migrating the data to the corresponding storage level according to the selected action (such as migrating from optical storage to HDD, or from HDD to SSD).

[0063] Reward feedback, after executing the action, the system calculates the reward according to factors such as performance improvement, resource conservation, and clinical priority .

[0064] Q-value update: Update the Q-value according to the Q-learning update formula to learn the optimal migration strategy.

[0065] Furthermore, dynamically adjust the storage hierarchy in combination with the access frequency of medical image data, automatically migrate frequently accessed medical image data to high-speed storage (such as SSD), optimize the user experience, and avoid waste of storage resources. Overall, this solution significantly improves the access efficiency, fault tolerance, and storage optimization ability of medical image data, reduces the risk of medical image data loss, and ensures the long-term reliability and availability of medical image data.

[0066] It should be noted that the storage index is intelligently queried through the storage management control layer to ensure the efficient retrieval and access of medical image data. At the same time, the integrity verification of medical image data is carried out in combination with the status of redundant check blocks to improve the availability and reliability of medical image data. When the medical image data is complete, the optimal access path can be provided according to the storage hierarchy to achieve fast reading. When part of the medical image data is lost, the system automatically calls the erasure code algorithm to recover the lost medical image data to ensure the availability of medical image data. When the loss of medical image data exceeds the recovery ability, the system immediately triggers the administrator notification mechanism to reduce the risk of irrecoverability.

[0067] S4: Regularly perform integrity detection on medical image data according to the storage index. When data corruption is detected, use erasure code technology to recover the corrupted medical image data.

[0068] Query the storage index to determine the storage hierarchy and the location of redundant check blocks of medical image data: Access the storage index table to obtain the storage hierarchy (SSD / HDD / optical storage) and the storage location of redundant check blocks of medical image data.

[0069] Record the distribution of medical image data blocks and redundant check blocks to ensure the availability of medical image data blocks.

[0070] Access the location of the redundant check blocks recorded in the index table and check whether all redundant check blocks are still available: Traverse the storage location of redundant check blocks to detect the availability of redundant check blocks.

[0071] Read the medical image data of the redundant check blocks and compare it with the records in the storage index to determine whether there is damage or missing.

[0072] When the redundant check blocks are complete, if all redundant check blocks are available, the medical image data is complete, and the user request operation (such as reading, downloading, analyzing medical image data, etc.) is directly executed.

[0073] Medical image data is accessed according to the storage hierarchy: High-performance layer (SSD): Directly reads medical imaging data and returns it to the user. Medium-performance layer (HDD): Reads medical imaging data and decodes redundant check blocks. Low-performance layer (optical storage): If medical imaging data is stored in optical storage, schedules the medical imaging data to the buffer area to improve access speed.

[0074] Verify the status of the redundant check blocks associated with the data, and perform user-requested operations according to the status of the redundant check blocks and their storage levels, including accessing the positions of the redundant check blocks recorded in the index table and checking whether all redundant check blocks are still available; when the redundant check blocks are complete, the medical imaging data is complete, and the user-requested operation is directly executed.

[0075] Regularly perform integrity detection and recover abnormal data. To ensure the long-term integrity and reliability of medical imaging data, the system regularly performs integrity detection on the data stored. This detection process includes: The system triggers data integrity detection according to a preset time interval or access frequency (e.g., monthly, quarterly, or each time data is read). Use erasure coding technology to check the verification results between medical imaging data blocks and redundant check blocks. Calculate the hash value of the data block through the hash algorithm and compare it with the hash value in the redundant check block to verify whether the data has been damaged during storage.

[0076] When data corruption is detected (e.g., the data block does not match the redundant check block), the system will recover the abnormal medical imaging data through erasure coding: Use Reed-Solomon or other redundant recovery algorithms to recover the damaged data block through the stored redundant check blocks. If the redundant check block itself is also damaged or lost, the system will try to recover the check block from other storage media (e.g., from off-site backup) according to the preset redundant backup strategy.

[0077] After the data recovery is completed, the system will recalculate the redundant check blocks of the data blocks and synchronously update the storage index table to ensure that future access requests can find the recovered data.

[0078] When the medical imaging data changes due to operations such as adding annotations, image processing, or other modifications, the storage management control layer receives a trigger signal for the modification of the medical imaging data.

[0079] According to the modification operation, generate the updated medical imaging data and assign new version information to it. The SMCL uses the information of the new version of the medical imaging data to regenerate the storage index. For the entire medical imaging data, calculate a new check matrix using finite field operations and generate redundant check blocks.

[0080] Store the updated redundant check blocks on the specified medium according to the original strategy, and record the storage location and related parameters of the new redundant check blocks in the storage index.

[0081] After the update is completed, perform a consistency check on the medical image data to verify the matching of the new medical image data with the redundancy check block, ensure that the update process is error-free, and thus guarantee the integrity and recoverability of the medical image data.

[0082] SMCL records the detailed log of this update operation and feedbacks the information of successful update to the user or administrator, ensuring that subsequent access requests are executed based on the latest medical image data and redundancy information.

[0083] It should be noted that through the version management mechanism, the medical image data can be accurately traced, and at the same time, medical risks caused by the coverage or loss of medical image data can be avoided. Using erasure code technology, the redundancy strategy is optimized for different storage levels (SSD, HDD, optical storage), so that the medical image data still has high recovery ability after update. Even if the storage medium is damaged, the content of the medical image data can be restored through the redundancy check block. In addition, the updated storage index can ensure that the latest version of the medical image data is always retrieved when the user accesses, avoiding the use of outdated or incorrect information from affecting clinical decisions. The synchronized consistency check of medical image data further guarantees the reliability of the updated medical image data and reduces the occurrence of potential storage errors or inconsistencies. Finally, SMCL records a complete update log, improves traceability and the security of medical image data, and enhances the stability and reliability of medical image data storage management.

[0084] S5: When receiving a user access request, locate the storage location of the medical image data by querying the storage index, execute the user request operation, and verify the status of the redundancy check block.

[0085] If the access frequency of the medical image data increases, automatically adjust the storage level: If the medical image data was originally stored in the HDD or optical storage layer and is accessed frequently within a short period of time, prefetch the medical image data to a higher performance layer (SSD) to optimize the user access experience.

[0086] Record the medical image data access log, adjust the storage priority and update the comprehensive priority, and optimize the medical image data storage in combination with the dynamic storage adjustment strategy. The maximum redundancy is judged according to experience.

[0087] The Storage Management Control Layer (SMCL) performs integrity checks on the storage index and medical image data according to the set integrity detection period: Trigger mechanism: Periodic detection (weekly). Storage level strategy: High-performance layer (SSD): Detection frequency per day to ensure high reliability of accessing and storing medical image data. Medium-performance layer (HDD): Detection weekly to ensure storage stability. Low-performance layer (optical storage): Detection every six months to prevent damage to medical image data caused by the aging of optical media for long-term storage.

[0088] Access the storage index table, check the storage path, redundant check block location, medical image data size, and integrity of the original medical image data for each medical image data to ensure that the index information is not damaged or lost.

[0089] Query the index table to obtain the storage locations of medical image data blocks and redundant check blocks. Read the medical image data of the redundant check blocks and detect whether all stored redundant check blocks are still available: If all redundant check blocks are intact, the medical image data is complete and directly passes the integrity check. If some redundant check blocks cannot be accessed, enter the medical image data recovery process.

[0090] Calculate the hash value of the current medical image data block using hash verification and compare it with the pre-stored value in the storage index: If the hash matches, it indicates that the medical image data is not damaged and the integrity verification passes. If the hash does not match, it indicates that the medical image data may be damaged and enter the recovery process.

[0091] Some redundant check blocks are damaged (the number of damaged blocks < maximum redundancy), and the medical image data can still be recovered. Invoke the erasure code algorithm (Reed - Solomon code) to reconstruct the lost medical image data: Read the remaining k intact medical image data blocks and m redundant check blocks. Calculate the missing medical image data blocks or redundant check blocks through finite field operations: where G represents the erasure code generation matrix, D represents the medical image data block, and P represents the redundant check block. Reconstruct the lost medical image data and store it back to the original storage medium or a new storage medium. Update the storage index to record the locations of the recovered medical image data blocks and redundant check blocks.

[0092] The medical image data block is damaged and can be recovered (the number of damaged blocks < maximum redundancy). Read the remaining medical image data blocks and redundant check blocks. Reconstruct the missing medical image data blocks through the Reed - Solomon code:

[0093] Rewrite the recovered medical image data to the storage medium to ensure the integrity of the medical image data.

[0094] When the medical image data cannot be reconstructed, trigger the administrator notification mechanism: Record the UID of the damaged medical image data and mark the status of the medical image data as "lost".

[0095] Check if there is a historical backup: If there is a historical version, try to recover from the historical medical image data and update the index. If there is no historical backup, notify the administrator for manual recovery. If a user accesses the medical image data, return a "Medical image data unavailable" prompt and provide recovery suggestions.

[0096] After the medical image data is restored, recalculate m redundant check blocks and store them in the corresponding storage layer. Update the index table to ensure the integrity of the medical image data structure. Perform hash verification again to ensure that the restored medical image data is consistent with the original medical image data and guarantee the restoration quality.

[0097] It should be noted that through regular integrity detection, redundant check block verification, and erasure code recovery mechanisms, the long-term availability and security of medical image data are ensured. A hierarchical detection strategy is adopted, with frequent checks on the high-performance layer (SSD), regular verification on the medium-performance layer (HDD), and long-term maintenance on the low-performance layer (optical storage), effectively reducing the risk of medical image data loss caused by the aging of storage media. The combination of hash verification and erasure code technology enables the automatic recovery of medical image data even in the case of partial damage. Compared with traditional RAID solutions, the recovery efficiency is higher, reducing the probability of medical image data loss. At the same time, for the loss of medical image data beyond the recoverable range, a historical backup recovery and administrator notification mechanism is provided to ensure that the integrity of medical image data is not affected.

[0098] S6: When the medical image data needs to be changed, locate the storage location of the medical image data according to the storage index, change the medical image data, and synchronously update the redundant check blocks.

[0099] Synchronously adjusting the storage location of the redundant check blocks includes that when the medical image data is archived to optical storage, the number of redundant check blocks is reduced to the minimum standard; receiving a user access request and querying the storage index When the system receives a user's medical image data access request, it will first verify and parse the request to confirm whether the user has access rights and ensure the legitimacy of the request. The specific operations include: the system confirms the user's access rights through a user authentication mechanism (such as role-based access control RBAC). The data identifier (such as UID or image ID) contained in the request will be extracted and used for subsequent operations.

[0100] The storage management control layer queries the storage index. The main goal of querying the storage index is to determine the actual location of the medical image data in the storage system and obtain relevant metadata (such as storage level, redundant check block location, etc.). After receiving the user request, the storage management control layer queries the storage index table based on the data identifier (UID). The storage index table records information such as the storage location of each medical image data, the location of the redundant check blocks, and the storage level. Through the query, the system can locate the current storage location of the image data and the associated redundant check blocks.

[0101] By querying the storage index table, locate the storage path of the data block and the location of the redundant check blocks.

[0102] Verify the status of the redundant check block. Once the system locates the storage location, it will next check the status of the redundant check block to ensure data integrity. The system will read the redundant check block and compare it with the original data to verify whether the redundant check block conforms to the expected error correction code. If the status of the redundant check block is normal, the user request operation will continue to be executed. If there are problems with the redundant check block (such as damage or loss), the system will trigger the data recovery mechanism.

[0103] Update the storage index and redundant check block when medical image data is changed. When the user requests to change medical image data (such as modifying, updating, relabeling the image, etc.), the system first needs to confirm whether the change requires a storage level adjustment or redundant update. The changes submitted by the user will be parsed and corresponding operations will be executed (such as image editing, metadata update, etc.). The system generates the updated image data block and recalculates the redundant check block for the new image data block.

[0104] Update the storage index table to record the new storage location and the location of the redundant check block. The new storage index needs to include the actual storage location of the updated data block and the redundant check block information. The system updates the storage index table to ensure that the new data block and the location of the redundant check block are correctly recorded. For example, if the data is migrated to a new storage level (such as from HDD to SSD), the storage index will also be updated accordingly.

[0105] For each data change, the redundant check block must also be updated synchronously. The specific operations are as follows: The system calculates the new redundant check block, generates the new redundant check block through the ReedSolomon erasure code algorithm. Store the new redundant check block at the predetermined storage level (according to the redundancy policy, such as SSD or HDD) and update the storage path. Re-verify the validity of the redundant check block to ensure the reliability of data recovery. The lifecycle management policy is periodically executed by the Storage Management Control Layer (SMCL) to detect long-unaccessed medical image data and determine its storage optimization strategy.

[0106] If the medical image data has not been accessed for a long time (6 months), judge whether migration is required according to the migration conditions. Before migrating the medical image data, check the integrity of the medical image data: Read the medical image data block and the redundant check block, perform a hash check to ensure that the medical image data is not damaged. If damage is detected, call the erasure code algorithm to recover the medical image data to ensure the integrity of the medical image data during migration.

[0107] When migrating from SSD to HDD: The erasure code policy is adjusted from high redundancy to medium redundancy.

[0108] When migrating from HDD to optical storage: The erasure code policy is adjusted from medium redundancy to low redundancy to improve long-term storage reliability.

[0109] During the migration of data, the system not only migrates the data itself but also synchronously adjusts the storage location of the redundant check blocks to ensure that the data can still receive complete redundant protection in the low-performance storage layer (such as optical storage). By using erasure coding technology, new redundant check blocks are recalculated and generated to ensure that the migrated data can be effectively recovered in the future.

[0110] Update the storage index. During the migration process, the system updates the storage index table, recording the new storage location of the data and the storage path of the redundant check blocks. The migrated data can still be quickly located through the updated storage index.

[0111] Update the new storage path, storage level, and location of the redundant check blocks of the medical image data in the index table.

[0112] It should be noted that through the life cycle management strategy, the storage level of long-unaccessed medical image data is optimized to ensure the efficient use of storage resources. Through intelligent detection and LSTM prediction, the future access requirements of medical image data can be accurately evaluated, and the storage level can be dynamically adjusted to reduce the occupancy of the high-performance storage layer (SSD) and relieve the storage pressure. In addition, through the storage index update mechanism, the system effectively reduces the I / O overhead, ensures the safe migration of medical image data between different storage layers, and avoids the loss or access delay of medical image data. Compared with the traditional static storage management, this method improves the scalability, reliability, and economy of the storage system, reduces the long-term storage cost, and at the same time ensures the integrity and availability of medical image data at any time.

[0113] Embodiment 2 is an embodiment of the present invention, providing a medical image data management system based on optical-electromagnetic hybrid hierarchical storage, including: Medical image data receiving and storing module, which receives medical image data and stores it in the high-performance layer, generates redundant check blocks by using erasure coding technology, and generates a storage index.

[0114] Storage priority calculation and dynamic level adjustment module, which calculates the storage priority of medical image data according to the attributes of medical image data, dynamically adjusts the storage level of medical image data by using the hierarchical storage weight model, synchronously adjusts the storage location of the redundant check blocks, and updates the storage index.

[0115] User access management module, which receives user access requests, queries the storage index through the storage management control layer, locates the current storage location of the medical image data, verifies the status of the redundant check blocks associated with the medical image data, and executes the user request operation according to the status of the redundant check blocks and their storage levels.

[0116] An update module, when medical image data is changed, updates the storage index by using the storage management control layer and synchronously updates the redundant check blocks.

[0117] A recovery module, according to the storage index and the erasure code redundancy strategy, periodically performs integrity detection on medical image data, and when it detects that the medical image data is damaged, uses erasure code technology to recover the abnormal medical image data.

[0118] A lifecycle management module, executes the lifecycle management strategy for medical image data that has not been accessed for a long time, and when migrating, synchronously adjusts the redundant check blocks of the storage and updates the storage index.

[0119] Embodiment 3, an embodiment of the present invention, which is different from the previous two embodiments in that: If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs, etc., which can store program codes.

[0120] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch instructions from the instruction execution system, apparatus, or device and execute the instructions), or used in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

[0121] More specific examples (nonexhaustive list) of computer-readable media include the following: electrical connections (electronic devices) having one or more wirings, portable computer disk cartridges (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber devices, and portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0122] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGA), field programmable gate arrays (FPGA), etc.

[0123] Example 4, an embodiment of the present invention, provides a medical image management method and system based on optical electromagnetic hybrid hierarchical storage. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0124] This experiment was carried out in the medical image storage system of a certain tertiary first-class hospital, aiming to verify the advantages of the present invention in terms of image storage, access speed, data integrity, and storage cost optimization. The experiment used an Intel Xeon Gold 6226R processor, 128 GB of memory, 8 TB of NVMe SSD (high-performance layer), 50 TB of HDD (medium-performance layer), and a 100 TB Blu-ray disc storage library (low-performance layer). The data source was the medical images of the hospital in the past 12 months, totaling 10.2 TB, covering various medical images such as CT, MRI, and ultrasound.

[0125] Step 1: In the process of medical image data storage and index generation experiment, the hospital adds 1.2 TB of image data every day. These data are first stored in the high-performance layer (SSD), and Reed-Solomon erasure code (4 + 2) is used for data protection to generate 6 data blocks (4 data blocks + 2 redundant check blocks), which are respectively stored in the SSD and HDD. At the same time, a storage index is created to record the data UID, storage path, and the location of the redundant check blocks.

[0126] Step 2: Storage Priority Calculation and Dynamic Adjustment of Storage Hierarchy The storage system calculates the storage priority of images by combining factors such as access frequency, data creation time, image importance, storage cost, and regulatory requirements, and predicts the future image access trend based on the LSTM prediction model. After the priority calculation, if the comprehensive priority is higher than 80, the image data is stored in the high-performance layer (SSD). If the priority is between 40 and 80, it is stored in the medium-performance layer (HDD). If the priority is lower than 40, the data is archived to the low-performance layer (optical storage). During the data migration process, the storage management system automatically adjusts the storage location of the redundant check blocks and updates the storage index to ensure data availability.

[0127] Step 3: User Access to Image Data After the image data is stored, users can access the data through the storage management control layer. The system first queries the storage index, locates the data storage hierarchy, and performs integrity verification. If the redundant check blocks are complete, the data is directly read. If some redundant check blocks are missing, the erasure code algorithm is called to recover. If the damaged redundant check blocks exceed the recoverable range, the administrator is notified for manual recovery.

[0128] Step 4: Image Data Change and Synchronous Update of Redundant Check Blocks When the user modifies the image data (such as adding diagnostic annotations), the system automatically calculates a new hash value UID, updates the storage index, and regenerates the redundant check blocks to ensure data consistency.

[0129] Step 5: Periodic Integrity Detection and Data Recovery The storage management control layer performs data integrity detection every 30 days, uses hash verification to compare whether the data blocks are damaged. If an anomaly is found, the erasure code algorithm is called to recover the data to ensure the integrity of the image data throughout its life cycle.

[0130] Step 6: Lifecycle Management Strategy For image data that has not been accessed for a long time (no access record for more than 12 months), the system executes the lifecycle management strategy, archives the data to optical storage, and reduces the proportion of redundant check blocks (adjusted from 4+2 to 6+3) to reduce storage space occupancy. At the same time, if the image data is accessed again, the system will automatically migrate back to HDD or SSD to improve access efficiency.

[0131] The experiment ran for 6 months. Comparing with the traditional HDD storage scheme, the results are as follows: In terms of image access speed, the average time for SSD to read an image is 3.12 seconds, the average time for HDD to read an image is 6.43 seconds, and the average read time of optical storage has dropped from 21.35 seconds in the traditional method to 9.87 seconds. Compared with the traditional HDD storage scheme, the image access speed has increased by 62.2%.

[0132] In terms of data integrity check, the present invention adopts a 30-day integrity detection cycle. Compared with the traditional HDD storage with a 60-day detection cycle, the success rate of recovering damaged image data has increased from 87.4% to 99.1%, and the probability of data loss has been significantly reduced.

[0133] In terms of optimizing storage costs, due to the introduction of the intelligent hierarchical storage strategy, only 15.3% of the image data is stored in the high-performance layer (SSD), and the rest of the data is migrated to HDD or optical storage. The storage cost is reduced by 34.5%. Among them, the cost of optical storage is only 40% of that of HDD storage, greatly reducing the operating expenses for long-term storage.

[0134] In terms of optimizing system energy consumption, since optical storage has the characteristic of zero-power standby compared with HDD, combined with the data life cycle management strategy, the energy consumption of the overall image storage system is reduced by 28.6%, and the storage efficiency is further improved.

[0135] The experimental data fully demonstrate the superiority of the present invention in medical image storage management. Compared with the traditional HDD storage solution, the present invention effectively improves the access efficiency, reliability, and storage cost optimization ability of image data through intelligent storage priority calculation, LSTM prediction migration, erasure code data protection, and life cycle management.

[0136] (1) The storage hierarchical strategy optimizes the access speed of image data. The experimental results show that the average time for SSD to read image data is 3.12 seconds, which is 62.2% faster than the HDD solution, and the access time of optical storage is shortened to 9.87 seconds, indicating that the intelligent storage hierarchical strategy effectively reduces the access latency of image data and improves the system response ability.

[0137] (2) The data integrity is significantly improved. The present invention uses the erasure code algorithm to perform redundant protection on image data and regularly performs integrity detection, so that the success rate of recovering damaged data reaches 99.1%. Compared with the traditional storage solution (87.4%), the risk of data loss is reduced, and the reliability of the system is significantly improved.

[0138] (3) The storage resource allocation is more efficient. The intelligent storage level migration strategy enables the high-performance SSD to store only 15.3% of the image data, and the rest of the image data is migrated to the HDD and optical storage layers. This strategy reduces the storage cost by 34.5% while ensuring that the data can still be accessed at any time, greatly reducing the hospital's operating expenses.

[0139] (4) The management efficiency of long-term image data is improved. Since the data that has not been accessed for a long time is migrated to optical storage, and the unit cost of optical storage is 60% lower than that of HDD, combined with the data life cycle management strategy, the energy consumption is reduced by 28.6%, making the image storage system more energy-saving and environmentally friendly while ensuring the long-term availability of data.

[0140] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A medical image management method based on optical and electromagnetic hybrid hierarchical storage, characterized in that: include: Receive medical imaging data, generate redundant check blocks using erasure coding technology, and generate storage indexes; According to the attributes of medical image data, the storage priority of medical image data is calculated using an adaptive time series prediction model; the storage priority includes calculating the access priority using the time series prediction model, and calculating the storage priority based on the weighted sum of the access priority and clinical importance; According to the prediction results, the storage level of medical imaging data is dynamically adjusted, and the redundant check blocks are adjusted synchronously, and the storage index is updated; Regularly perform integrity checks on medical imaging data. When data corruption is detected, use redundant check blocks to restore the damaged medical imaging data. When medical imaging data is changed, the redundant check blocks are synchronously updated. The dynamic adjustment includes adjusting the migration scheduling strategy optimized by a reinforcement learning algorithm; In the migration scheduling strategy, the data status is evaluated according to the storage priority, and the migration scheduling is automatically triggered according to the set migration threshold and priority conditions. The reinforcement learning algorithm is used to evaluate the initial state of the data. The migration decision is determined by randomly selecting the migration action and the action with the maximum Q value through the exploration-utilization strategy, and the reward function is adaptively calculated according to the multi-factor migration effect, and the Q value is updated to continuously optimize the migration decision.

2. The medical image management method based on optical and electromagnetic hybrid hierarchical storage as claimed in claim 1, characterized in that: The medical imaging data includes original medical images, imaging metadata, imaging processing data, imaging-related documents, and imaging storage security information; The medical image data attributes include image type, creation time, access frequency, clinical importance, and access rights; the hierarchical storage includes high-performance layer SSD solid-state storage, medium-performance layer HDD magnetic storage, and low-performance layer optical storage.

3. The medical image management method based on optical and electromagnetic hybrid hierarchical storage as claimed in claim 2, characterized in that: The method of using erasure coding technology to generate redundant check blocks includes converting the original medical image data Divide into Data blocks , the size of each data block is ; Input data block , use the coding matrix on the finite field to encode the data block, and use the erasure code algorithm to generate Redundancy check block ; Generate a storage index, record the storage location of each data block and its redundant check block, generate a unique identifier for each data block and redundant check block, and generate a UID by performing a hash calculation on each data block and redundant check block; assign a storage path and a redundant check block storage path to the data block, record the physical location of the data block and redundant check block, and have a unique path in three types of storage devices: SSD, HDD, and optical storage; Update the UID, storage path, storage level, and data size information of the data block and redundancy check block to the storage index table.

4. The medical image management method based on optical-electromagnetic hybrid hierarchical storage as claimed in claim 3, characterized in that: The calculation of the priority of medical image data storage includes predicting the access frequency of data, using an adaptive time series prediction model LSTM network to predict future access needs based on historical access data; using the historical access frequency of data blocks as an input sequence , LSTM predicts the future through past access patterns Access frequency within the step; calculate storage priority based on the access frequency predicted by LSTM and the clinical importance of medical imaging , comprehensive computing storage priority , Expressed as a weighted sum of visit frequency and clinical importance.

5. The medical image management method based on optical-electromagnetic hybrid hierarchical storage as claimed in claim 4, characterized in that: The migration scheduling strategy includes storage tier division, setting migration conditions and delayed migration triggering mechanism; When the predicted access frequency of a data block Above threshold , and store the priority Reaching the predetermined trigger value When , data will be migrated from the low-performance layer to the medium-performance layer; represents the access threshold, Indicates the data migration trigger value; When data storage priority Reaching the predetermined trigger value And priority When the data is migrated from the low-performance layer or the medium-performance layer to the high-performance layer, represents the clinical urgency threshold; Setting Thresholds , at each time step Check whether the data needs to be migrated. If the access frequency and priority within a time step exceed the set threshold, migration scheduling is triggered; After the migration scheduling is triggered, the reinforcement learning algorithm is used to optimize the migration decision based on the predicted access frequency and priority; the Qlearning model decides whether to migrate by learning the migration benefits of the system.

6. The medical image management method based on optical-electromagnetic hybrid hierarchical storage as claimed in claim 5, characterized in that: The reinforcement learning algorithm includes: Indicates in status Take action The value of Indicates the immediate reward obtained after executing the action. represents the discount factor, represents the learning rate, Indicates that in the next state Next, select the action that can obtain the maximum Q value; At each time step , perform the migration decision by following these steps: Status evaluation, based on current access frequency , Storage Priority and clinical importance Evaluate the current state of your data; The migration decision is determined by selecting the action with the maximum Q value, using an explore-exploit strategy. represents the probability of performing exploration; use Choose a random action to The probability of selecting the action with the maximum Q value; Execute the migration and migrate the data to the corresponding storage tier according to the selected action; After the action is performed, rewards are dynamically calculated based on performance improvement, resource savings, and clinical priority ; Update the Q value according to the Q-learning update formula and learn the best migration strategy.

7. The medical image management method based on optical-electromagnetic hybrid hierarchical storage as claimed in claim 6, characterized in that: The data recovery mechanism includes, when receiving a user's medical imaging data access request, reading the redundant check block and comparing it with the original data, using the erasure code technology to check the verification result between the medical imaging data block and the redundant check block; calculating the hash value of the data block by the hash algorithm and comparing it with the hash value in the redundant check block to verify whether the data is damaged; When data corruption is detected, a redundant recovery algorithm is used to recover the damaged data block through the stored redundant check blocks. For medical imaging data that has not been accessed for a long time, a lifecycle management strategy will be implemented to migrate infrequently accessed data to low-cost storage and adjust the redundant check blocks synchronously.

8. A medical image management system based on optical-electromagnetic hybrid hierarchical storage using the method according to any one of claims 1 to 7, characterized in that: The data receiving and storage module receives medical imaging data and stores it in the high-performance layer, uses erasure coding technology to generate redundant check blocks, and generates storage indexes; A storage priority calculation module calculates the storage priority of the medical image data according to the image attributes of the medical image data, adjusts the storage position of the redundant check block, and updates the storage index; The user access management module receives user access requests, queries the storage index through the storage management control layer, and locates the current storage location of the medical imaging data; The update recovery module uses the storage management control layer to update the storage index when the medical imaging data is changed, and regularly performs integrity checks on the medical imaging data.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the medical image management method based on optical and electromagnetic hybrid hierarchical storage as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the medical image management method based on optical-electromagnetic hybrid hierarchical storage according to any one of claims 1 to 7 are implemented.

Images