A method for optimizing medical image data in a medical information system

Abstract

This invention discloses a method for optimizing medical image data in a medical information system, comprising: S1, acquiring data through the image acquisition module of the medical information system; S2, preprocessing the acquired raw medical image data; S3, compressing the preprocessed medical image data using a lossless compression algorithm; S4, storing the compressed medical image data in a database structure; S5, transmitting the compressed medical image data within the medical information system using a data transmission protocol; S6, intelligently processing the transmitted and stored medical image data using machine learning technology; S7, protecting the security of the medical image data during storage and transmission using data encryption and access control technologies; and S8, achieving comprehensive monitoring and auditing of access to the medical image data through a data monitoring and auditing module. This invention has advantages such as significantly improved image quality, high compression efficiency, and strong data security.

Description

Technical Field

[0001] This invention relates to the field of medical imaging technology, and in particular to a method for optimizing medical image data for medical information systems. Background Technology

[0002] With the rapid development of medical information technology, the optimization and management of medical imaging data has become one of the key technologies in medical information systems. Medical imaging data plays a vital role in diagnosis, treatment, and research, and its quality and efficiency directly affect the level of medical services and patient treatment outcomes. However, existing technologies still face many problems in the processing, storage, and transmission of medical imaging data, limiting the overall performance and reliability of medical information systems.

[0003] In existing technologies, traditional medical image data processing methods mainly rely on single image processing techniques or data compression algorithms. These methods exhibit the following shortcomings when faced with the ever-increasing volume of medical data and complex application requirements:

[0004] 1. Insufficient Image Quality Improvement: Traditional image processing techniques, such as simple denoising and enhancement methods, cannot effectively remove complex noise and improve image clarity, resulting in image quality that fails to meet the requirements of accurate diagnosis. Existing image processing methods lack the ability to extract and optimize complex image features, affecting the accuracy and reliability of diagnosis.

[0005] 2. Low data compression efficiency: Existing lossless and lossy compression algorithms have limited effectiveness in reducing storage space and transmission bandwidth usage, especially in high-resolution and large-volume image data processing, where they cannot significantly improve compression efficiency while maintaining image quality. These traditional algorithms often face problems of insufficient compression ratio and slow processing speed when dealing with large-scale data.

[0006] 3. Insufficient transmission speed and stability: Traditional data transmission protocols and methods suffer from slow speed, high latency, and poor stability in large-capacity medical image data transmission. Especially under network congestion and high load environments, the efficiency and reliability of data transmission are difficult to guarantee, affecting the real-time performance and effectiveness of the medical information system.

[0007] 4. Limited Intelligent Processing Capabilities: Most existing image data processing systems rely on simple machine learning algorithms or rules, lacking the application of deep learning and artificial intelligence technologies, resulting in low levels of intelligence in image recognition and diagnosis. Traditional methods have limitations in image feature extraction, classification, and anomaly detection, failing to achieve efficient and accurate intelligent processing.

[0008] 5. Insufficient Data Security and Privacy Protection: Existing medical imaging data security measures are mostly static encryption and access control, which cannot comprehensively protect against security threats during data storage and transmission. Traditional methods have technical deficiencies in data encryption, transmission encryption, and access control, making it difficult to guarantee comprehensive data security and the protection of patient privacy.

[0009] Therefore, how to provide a method for optimizing medical image data for medical information systems is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0010] One objective of this invention is to propose a method for optimizing medical image data in medical information systems. This invention fully utilizes image processing, data compression, data transmission, intelligent processing, and security protection technologies, and details the specific steps and algorithms for optimizing medical image data during storage, processing, and transmission. This method improves image quality by employing advanced image denoising, enhancement, segmentation, and registration techniques; significantly reduces storage space and transmission bandwidth usage through improved lossless and lossy compression algorithms; improves transmission speed and stability through optimized data transmission protocols and methods; enhances the accuracy and speed of image recognition and diagnosis by combining deep learning and artificial intelligence technologies; and ensures data security and patient privacy protection by introducing data encryption and access control technologies. This invention possesses advantages such as significantly improved image quality, high compression efficiency, fast transmission speed, precise intelligent processing, and strong data security, and can meet the requirements of modern medical information systems for high efficiency, real-time performance, and security in image data processing.

[0011] A method for optimizing medical image data in a medical information system according to an embodiment of the present invention is characterized by comprising the following steps:

[0012] S1. Data is collected through the image acquisition module of the medical information system to obtain raw medical image data;

[0013] S2. Preprocess the acquired raw medical image data, including image denoising, image enhancement, image segmentation, and image registration.

[0014] S3. Apply lossless compression algorithm to compress the preprocessed medical image data to reduce storage space requirements while ensuring that image quality is not significantly affected.

[0015] S4. Store the compressed medical image data in the database structure and manage it through the data storage management module to improve the storage efficiency and retrieval speed of the image data.

[0016] S5. Using data transmission protocols, the compressed medical image data is transmitted within the medical information system;

[0017] S6. By combining machine learning technology, intelligent processing of transmitted and stored medical image data is performed. By optimizing image data processing algorithms, the speed and accuracy of image recognition and diagnosis are improved.

[0018] S7. Data encryption and access control technologies are used to protect the security of medical image data during storage and transmission.

[0019] S8. Through the data monitoring and auditing module, comprehensive monitoring and auditing of access to medical imaging data can be achieved to prevent unauthorized access and data leakage.

[0020] Optionally, S1 specifically includes:

[0021] S11. Configure an image acquisition module in the medical information system. The image acquisition module includes an image acquisition device and a data interface to receive various types of raw medical image data.

[0022] S12. Convert the received raw medical image data into a unified DICOM format, converting image data from different sources and formats.

[0023] S13. Perform a preliminary quality check on the DICOM format image data:

[0024]

[0025] Where Q represents the image quality score, w i f represents the weighting coefficient. i (I) represents the i-th quality feature function, and n1 is the number of quality features;

[0026] S14. Timestamp and record metadata for image data that has passed quality inspection:

[0027] M = {P, T, E};

[0028] Where M represents the metadata, P represents patient information, T represents image acquisition time, and E represents device information;

[0029] S15. Transmit the DICOM format image data with timestamps and metadata to the buffer area of ​​the image acquisition module through the data interface.

[0030] Optionally, S2 specifically includes:

[0031] S21. A Gaussian filter is used to reduce noise in the image, and image denoising processing is performed on the acquired raw medical image data:

[0032]

[0033] Among them, I denoised The image is denoised, σ is the standard deviation of the Gaussian filter, I(x,y) represents the original image data, and * represents the convolution operation.

[0034] S22. Histogram equalization is used to enhance image contrast; image enhancement processing is performed on the denoised medical image data.

[0035]

[0036] Among them, I enhanced This represents the enhanced image, where L represents the number of gray levels, and min(I) denoised ) and max(I denoised These represent the minimum and maximum gray values ​​of the image, respectively.

[0037] S23. The Otsu algorithm is used to automatically determine the segmentation threshold T, and image segmentation processing is performed on the enhanced medical image data:

[0038]

[0039] in, Let ω1(T) represent the inter-class variance, ω2(T) represent the probabilities of foreground and background pixels, and μ1(T) and μ2(T) represent the average gray values ​​of foreground and background pixels, respectively.

[0040] S24. A gradient descent-based mutual information method is used for registration to perform image registration processing on the segmented medical image data:

[0041]

[0042] Where MI(A,B) represents the mutual information between image A and image B, p(a,b) represents the joint probability distribution, and p(a) and p(b) represent the edge probability distributions of image A and image B, respectively;

[0043] S25. Transmit the medical image data after image denoising, enhancement, segmentation and registration to the image data processing module.

[0044] Optionally, S3 specifically includes:

[0045] S31. Using Huffman coding for lossless compression, the preprocessed medical image data is compressed using a lossless compression algorithm:

[0046]

[0047] Where C(I) represents the amount of compressed data, P(i) represents the probability of the i-th pixel value, and n2 is the total number of pixel values;

[0048] S32. Using Discrete Cosine Transform (DCT) and quantization processing, a lossy compression algorithm is applied to compress some medical image data:

[0049]

[0050] Where F(u,v) represents the DCT coefficients, I(x,y) represents the original image pixel values, N is the size of the image block, α(u) and α(v) are normalization coefficients, u represents the frequency index in the horizontal direction, and v represents the frequency index in the vertical direction;

[0051] S33. Quantize the lossy compressed data:

[0052]

[0053] Where Q(u,v) represents the quantized DCT coefficients, and QF(u,v) represents the coefficients of the quantization table;

[0054] S34. Using Inverse Discrete Cosine Transform (IDCT), the compressed medical image data is processed by inverse transformation to restore approximate values ​​of the original image:

[0055]

[0056] Where I(x,y) represents the recovered image pixel value, and F(u,v) represents the quantized DCT coefficients;

[0057] S35. Store the compressed medical image data in an optimized database structure and manage it through the data storage management module.

[0058] Optionally, S4 specifically includes:

[0059] S41. The database structure uses B+ tree indexing technology to store compressed medical image data in an optimized database structure. Data storage location L:

[0060]

[0061] Where B represents the branching factor of the B+ tree, h represents the height of the tree, and k i This represents the number of keywords in the i-th layer;

[0062] S42. A hash table structure is used to store metadata, enabling metadata management of image data and improving retrieval efficiency.

[0063] H(k1) = k1m;

[0064] Where H(k1) represents the hash value, k1 represents the key value of the metadata, and m represents the size of the hash table;

[0065] S43. Based on the different types and usage frequencies of the image data, the image data is partitioned and stored into hot data and cold data, which are then stored on different storage media:

[0066]

[0067] Where R represents the proportion of thermal data, D h D represents the amount of thermal data. c Indicates the amount of cold data;

[0068] S44. Based on the data size, the image data is segmented into several small segments for storage:

[0069]

[0070] Among them, S i Let n represent the size of the i-th slice, D1 represent the total size of the image data, and n3 represent the number of slices.

[0071] S45. RAID technology is used for redundant storage of image data, and storage redundancy management is implemented for image data.

[0072]

[0073] Where R represents the total amount of data after redundant storage, and D i This represents the i-th data fragment, and P represents redundant check data;

[0074] S46. Employ a combination of incremental and full backup methods for image data backup management:

[0075]

[0076] Where B1 represents the total amount of backup data, B full Indicates the total amount of data backed up, ΔB i Let m represent the data volume of the i-th incremental backup, and m be the number of incremental backups.

[0077] Optionally, S5 specifically includes:

[0078] S51. Transmit compressed medical image data within the medical information system using a data transmission protocol:

[0079]

[0080] Among them, R tT represents the transmission rate, D represents the total amount of data, and T represents the total data volume. t T represents the total transmission time. t D represents the total transmission time. i B represents the size of the i-th data block. i This represents the bandwidth of the i-th data block;

[0081] S52. Use cyclic redundancy check (CRC) to detect and correct errors in the data during transmission:

[0082] CRC = (D(x)·x) r G(x);

[0083]

[0084] G(x)=x r +g r-1 x r-1 +…+g1x+g0;

[0085] Where CRC represents Cyclic Redundancy Check, D(x) represents the data polynomial, x represents the shift operator, r represents the number of redundancy bits, G(x) represents the generator polynomial, and d i k2 represents the number of data bits, and g represents the number of data bits. i represents the coefficients of the generator polynomial, and r represents the number of redundant bits;

[0086] S53. The AES encryption standard algorithm is used to encrypt the data during transmission;

[0087] S54. Use the AES algorithm to decrypt the encrypted data;

[0088] S55. Divide the data D into several blocks, each block being M bits in size. For each data block D... i Perform an initial hash operation, and the hash value is represented as H. i , all initial hash values ​​H i Connect them to obtain the intermediate hash value H. mid For the intermediate hash value H mid Perform the final hash operation to obtain the final data digest H. Compare the generated data digest H with the expected hash value to verify data integrity.

[0089]

[0090]

[0091] if H == H expected Then data integrity verification passed;

[0092] Among them, Di Let H0 represent the i-th data block, n4 be the number of data blocks, H0 be the initial hash value, f be the hash function, and D be the hash function. i [j] represents the j-th bit in the i-th data block, H0 is the initial hash value, f is the hash function, and H mid [i] represents the i-th hash value in the intermediate hash values, H expected Indicates the expected hash value;

[0093] S56. Store the medical image data that has been transmitted and passed integrity verification in the central database.

[0094] Optionally, S6 specifically includes:

[0095] S61. Convolutional Neural Networks (CNNs) are used to extract image features. Combined with machine learning techniques, intelligent processing of transmitted and stored medical image data is performed to extract features from the image data.

[0096] F = σ(W*I + b);

[0097] Where F represents the extracted features, σ represents the activation function, W represents the convolution kernel, I represents the input image, b represents the bias term, and * represents the convolution operation;

[0098] S62. Input the extracted image features into a Support Vector Machine (SVM) for classification:

[0099]

[0100] Where w represents the weight vector, C represents the penalty parameter, and ξ i n represents the slack variable, and n5 represents the sample size;

[0101] S63. Perform cluster analysis on the classification results, and use the k-means clustering algorithm to group similar images into the same class:

[0102]

[0103] Where J represents the clustering objective function, and k3 represents the number of clusters. Let μ represent the i-th sample in the j-th cluster. j Denotes the centroid of the j-th cluster;

[0104] S64. Use the Isolation Forest algorithm to identify abnormal image data and perform anomaly detection on the clustering results:

[0105]

[0106] Where s(x) represents the anomaly score of sample x, E(h(x)) represents the average path length of sample x, c(n) represents the adjustment coefficient, and n6 represents the total number of samples;

[0107] S65. The Structural Similarity Index (SSIM) is used for evaluation, and the intelligently processed image data is compared and analyzed with the original data:

[0108]

[0109] Where, μ x and μ y This represents the average of the x and y values ​​of the image. and σ represents the variance of the images x and y. xy Let C1 and C2 represent the covariance of the images x and y, and C2 be the stability constants.

[0110] Optionally, S7 specifically includes:

[0111] S71. Data encryption and access control technologies are used to protect the security of medical image data during storage and transmission.

[0112] S72. Employ the Role-Based Access Control (RBAC) model to control access to encrypted data, defining user roles R and permissions P:

[0113] A1={(r i ,p j )|r i ∈R,p j ∈P};

[0114] Where A1 represents the access control list, r i p represents the i-th user role. j This represents the j-th permission;

[0115] S73. Encrypt the data during transmission using a secure transmission protocol:

[0116]

[0117] Among them, C t This indicates that encrypted data is being transmitted. D represents the transmission encryption function. t k represents data in transmission. t Indicates the transmission of the encryption key;

[0118] S74. Using the Message Authentication Code (MAC) algorithm, the integrity of the transmitted encrypted data is verified, and a data integrity verification code M is generated:

[0119] M = H(D)t +K m );

[0120] Where M represents the message authentication code, H represents the hash function, and D... t K represents data in transmission. m Indicates the message authentication key;

[0121] S75. The data access control module manages and monitors the storage of encrypted and access-controlled medical image data in a central database.

[0122] P a ={(u i ,r i )|u i ∈U,r i ∈R};

[0123] Among them, P a This represents the user role-permission mapping list, u i Let r represent the i-th user. i This represents the role of the i-th user.

[0124] S76. Regularly perform security checks and audits on the stored data, and generate audit reports. a :

[0125]

[0126] Among them, R a E represents the audit report, and u represents the audit event. i Let a represent the i-th user. i t represents the access behavior of the i-th user. i Indicates the access time.

[0127] Optionally, S8 specifically includes:

[0128] S81. Continuously optimize medical image data by regularly updating compression algorithms, transmission protocols, and processing algorithms:

[0129]

[0130] Where C′ represents the optimized compressed data volume, P′(i) represents the probability of the i-th pixel value after optimization, and n2 is the total number of pixel values;

[0131] S82. Optimize the transmission protocol using transmission optimization techniques based on flow control and congestion control:

[0132]

[0133] Among them, Ropt T represents the optimized transmission rate, D represents the total data volume, and T represents the total data volume. opt This represents the optimized total transmission time;

[0134] S83. Employ image processing algorithms based on deep learning and artificial intelligence, optimize the processing algorithms, and improve the accuracy and speed of image recognition and diagnosis:

[0135] F opt =σ(W opt *I+b opt );

[0136] Among them, F opt Let W represent the optimized features, σ represent the activation function, and W represent the optimized features. opt This represents the optimized convolution kernel, where I represents the input image, and b represents the input image. opt This indicates the optimized bias term, and * indicates the convolution operation;

[0137] S84. Optimize the storage structure of image data by adopting a distributed storage system and data sharding technology:

[0138]

[0139] Where S′ represents the optimized slice size, D1 represents the total size of the image data, and n′ represents the optimized number of slices;

[0140] S85 employs advanced data encryption technology and multi-level access control to optimize data security.

[0141] C′=E k′ (P);

[0142] Where C′ represents the optimized ciphertext, E k ′ represents the optimized encryption function, P represents the plaintext data, and k′ represents the optimized encryption key;

[0143] S86. Use performance evaluation metrics to evaluate the optimization effect, and monitor and evaluate the optimization process:

[0144]

[0145] Where E represents the optimized evaluation result, This represents the i-th performance metric before optimization. Let represent the i-th performance metric after optimization, and n be the number of performance metrics.

[0146] The beneficial effects of this invention are:

[0147] (1) This invention provides comprehensive optimization of medical image data by combining image processing, data compression, data transmission, intelligent processing, and security protection technologies. Specifically, advanced image denoising, enhancement, segmentation, and registration techniques are employed to significantly improve image quality, making diagnosis more accurate and reliable. The application of these image processing technologies enables the system to process various complex image data, improving the accuracy and reliability of diagnosis.

[0148] (2) This invention achieves efficient data compression through improved lossless and lossy compression algorithms, significantly reducing the storage space and transmission bandwidth required. The optimized data compression technology can improve the efficiency of data storage and transmission while ensuring that image quality is not significantly affected, providing a solid foundation for the large-scale application of medical imaging data.

[0149] (3) This invention improves the transmission speed and stability of large-capacity medical image data by using optimized data transmission protocols and methods, and employing flow control and congestion control technologies. The improved transmission protocol can maintain efficient and stable data transmission under various network conditions, ensuring timely transmission of medical data to meet clinical needs.

[0150] (4) This invention combines deep learning and artificial intelligence technologies to intelligently process transmitted and stored medical image data. It employs convolutional neural networks (CNNs) to extract image features, classifies them using support vector machines (SVMs), and combines k-means clustering and isolated forest algorithms for anomaly detection, thereby improving the intelligence level of image recognition and diagnosis. These intelligent processing technologies can quickly and accurately identify and analyze image data, reducing manual intervention and improving diagnostic efficiency.

[0151] (5) This invention ensures the security of medical image data and the protection of patient privacy during storage and transmission through data encryption and access control technologies. It employs the Advanced Encryption Standard (AES) algorithm and a Role-Based Access Control (RBAC) model to provide multi-layered security protection, preventing data leakage and unauthorized access. Attached Figure Description

[0152] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0153] Figure 1 This is an overall flowchart of a medical image data optimization method for a medical information system proposed in this invention.

[0154] Figure 2This invention presents a flowchart of image denoising, enhancement, segmentation, and registration processes for a medical image data optimization method for a medical information system.

[0155] Figure 3 This is a flowchart of the lossless and lossy compression algorithm processing of a medical image data optimization method for a medical information system proposed in this invention.

[0156] Figure 4 This is a flowchart of the data transmission protocol and optimization method for a medical image data optimization method for a medical information system proposed in this invention. Detailed Implementation

[0157] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0158] refer to Figures 1-4 A method for optimizing medical image data in a medical information system, characterized by comprising the following steps:

[0159] S1. Data is collected through the image acquisition module of the medical information system to obtain raw medical image data;

[0160] In this embodiment, S1 specifically includes:

[0161] S11. Configure an image acquisition module in the medical information system. The image acquisition module includes an image acquisition device and a data interface to receive various types of raw medical image data.

[0162] S12. Convert the received raw medical image data into a unified DICOM format, converting image data from different sources and formats.

[0163] S13. Perform a preliminary quality check on the DICOM format image data:

[0164]

[0165] Where Q represents the image quality score, w i f represents the weighting coefficient. i (I) represents the i-th quality feature function, and n1 is the number of quality features;

[0166] S14. Timestamp and record metadata for image data that has passed quality inspection:

[0167] M = {P, T, E};

[0168] Where M represents the metadata, P represents patient information, T represents image acquisition time, and E represents device information;

[0169] S15. Transmit the DICOM format image data with timestamps and metadata to the buffer area of ​​the image acquisition module through the data interface.

[0170] S2. Preprocess the acquired raw medical image data, including image denoising, image enhancement, image segmentation, and image registration.

[0171] In this embodiment, S2 specifically includes:

[0172] S21. A Gaussian filter is used to reduce noise in the image, and image denoising processing is performed on the acquired raw medical image data:

[0173]

[0174] Among them, I denoised The image is denoised, σ is the standard deviation of the Gaussian filter, I(x,y) represents the original image data, and * represents the convolution operation.

[0175] S22. Histogram equalization is used to enhance image contrast; image enhancement processing is performed on the denoised medical image data.

[0176]

[0177] Among them, I enhanced This represents the enhanced image, where L represents the number of gray levels, and min(I) denoised ) and max(I denoised These represent the minimum and maximum gray values ​​of the image, respectively.

[0178] S23. The Otsu algorithm is used to automatically determine the segmentation threshold T, and image segmentation processing is performed on the enhanced medical image data:

[0179]

[0180] in, Let ω1(T) represent the inter-class variance, ω2(T) represent the probabilities of foreground and background pixels, and μ1(T) and μ2(T) represent the average gray values ​​of foreground and background pixels, respectively.

[0181] S24. A gradient descent-based mutual information method is used for registration to perform image registration processing on the segmented medical image data:

[0182]

[0183] Where MI(A,B) represents the mutual information between image A and image B, p(a,b) represents the joint probability distribution, and p(a) and p(b) represent the edge probability distributions of image A and image B, respectively;

[0184] S25. Transmit the medical image data after image denoising, enhancement, segmentation and registration to the image data processing module.

[0185] S3. Apply lossless compression algorithm to compress the preprocessed medical image data to reduce storage space requirements while ensuring that image quality is not significantly affected.

[0186] In this embodiment, S3 specifically includes:

[0187] S31. Using Huffman coding for lossless compression, the preprocessed medical image data is compressed using a lossless compression algorithm:

[0188]

[0189] Where C(I) represents the amount of compressed data, P(i) represents the probability of the i-th pixel value, and n2 is the total number of pixel values;

[0190] S32. Using Discrete Cosine Transform (DCT) and quantization processing, a lossy compression algorithm is applied to compress some medical image data:

[0191]

[0192] Where F(u,v) represents the DCT coefficients, I(x,y) represents the original image pixel values, N is the size of the image block, α(u) and α(v) are normalization coefficients, u represents the frequency index in the horizontal direction, and v represents the frequency index in the vertical direction;

[0193] S33. Quantize the lossy compressed data:

[0194]

[0195] Where Q(u,v) represents the quantized DCT coefficients, and QF(u,v) represents the coefficients of the quantization table;

[0196] S34. Using Inverse Discrete Cosine Transform (IDCT), the compressed medical image data is processed by inverse transformation to restore approximate values ​​of the original image:

[0197]

[0198] Where I(x,y) represents the recovered image pixel value, and F(u,v) represents the quantized DCT coefficients;

[0199] S35. Store the compressed medical image data in an optimized database structure and manage it through the data storage management module.

[0200] S4. Store the compressed medical image data in the database structure and manage it through the data storage management module to improve the storage efficiency and retrieval speed of the image data.

[0201] In this embodiment, S4 specifically includes:

[0202] S41. The database structure uses B+ tree indexing technology to store compressed medical image data in an optimized database structure. Data storage location L:

[0203]

[0204] Where B represents the branching factor of the B+ tree, h represents the height of the tree, and k i This represents the number of keywords in the i-th layer;

[0205] S42. A hash table structure is used to store metadata, enabling metadata management of image data and improving retrieval efficiency.

[0206] H(k1) = k1m;

[0207] Where H(k1) represents the hash value, k1 represents the key value of the metadata, and m represents the size of the hash table;

[0208] S43. Based on the different types and usage frequencies of the image data, the image data is partitioned and stored into hot data and cold data, which are then stored on different storage media:

[0209]

[0210] Where R represents the proportion of thermal data, D h D represents the amount of thermal data. c Indicates the amount of cold data;

[0211] S44. Based on the data size, the image data is segmented into several small segments for storage:

[0212]

[0213] Among them, S i Let n represent the size of the i-th slice, D1 represent the total size of the image data, and n3 represent the number of slices.

[0214] S45. RAID technology is used for redundant storage of image data, and storage redundancy management is implemented for image data.

[0215]

[0216] Where R represents the total amount of data after redundant storage, and D i This represents the i-th data fragment, and P represents redundant check data;

[0217] S46. Employ a combination of incremental and full backup methods for image data backup management:

[0218]

[0219] Where B1 represents the total amount of backup data, B full Indicates the total amount of data backed up, ΔB i Let m represent the data volume of the i-th incremental backup, and m be the number of incremental backups.

[0220] S5. Using data transmission protocols, the compressed medical image data is transmitted within the medical information system;

[0221] In this embodiment, S5 specifically includes:

[0222] S51. Transmit compressed medical image data within the medical information system using a data transmission protocol:

[0223]

[0224] Among them, R t T represents the transmission rate, D represents the total amount of data, and T represents the total data volume. t T represents the total transmission time. t D represents the total transmission time. i B represents the size of the i-th data block. i This represents the bandwidth of the i-th data block;

[0225] S52. Use cyclic redundancy check (CRC) to detect and correct errors in the data during transmission:

[0226] CRC = (D(x)·x) r G(x);

[0227]

[0228] G(x)=x r +g r-1 x r-1 +…+g1x+g0;

[0229] Where CRC represents Cyclic Redundancy Check, D(x) represents the data polynomial, x represents the shift operator, r represents the number of redundancy bits, G(x) represents the generator polynomial, and d ik2 represents the number of data bits, and g represents the number of data bits. i represents the coefficients of the generator polynomial, and r represents the number of redundant bits;

[0230] S53. The AES encryption standard algorithm is used to encrypt the data during transmission;

[0231] S54. Use the AES algorithm to decrypt the encrypted data;

[0232] S55. Divide the data D into several blocks, each block being M bits in size. For each data block D... i Perform an initial hash operation, and the hash value is represented as H. i , all initial hash values ​​H i Connect them to obtain the intermediate hash value H. mid For the intermediate hash value H mid Perform the final hash operation to obtain the final data digest H. Compare the generated data digest H with the expected hash value to verify data integrity.

[0233] D = {D1, D2, ..., D} n4};

[0234]

[0235] H mid ={H1,H2,…,H n4};

[0236]

[0237] if H == H expected Then data integrity verification passed;

[0238] Among them, D i Let H0 represent the i-th data block, n4 be the number of data blocks, H0 be the initial hash value, f be the hash function, and D be the hash function. i [j] represents the j-th bit in the i-th data block, H0 is the initial hash value, f is the hash function, and H mid [i] represents the i-th hash value in the intermediate hash values, H expected Indicates the expected hash value;

[0239] S56. Store the medical image data that has been transmitted and passed integrity verification in the central database.

[0240] S6. By combining machine learning technology, intelligent processing of transmitted and stored medical image data is performed. By optimizing image data processing algorithms, the speed and accuracy of image recognition and diagnosis are improved.

[0241] In this embodiment, S6 specifically includes:

[0242] S61. Convolutional Neural Networks (CNNs) are used to extract image features. Combined with machine learning techniques, intelligent processing of transmitted and stored medical image data is performed to extract features from the image data.

[0243] F = σ(W*I + b);

[0244] Where F represents the extracted features, σ represents the activation function, W represents the convolution kernel, I represents the input image, b represents the bias term, and * represents the convolution operation;

[0245] S62. Input the extracted image features into a Support Vector Machine (SVM) for classification:

[0246]

[0247] Where w represents the weight vector, C represents the penalty parameter, and ξ i n represents the slack variable, and n5 represents the sample size;

[0248] S63. Perform cluster analysis on the classification results, and use the k-means clustering algorithm to group similar images into the same class:

[0249]

[0250] Where J represents the clustering objective function, and k3 represents the number of clusters. Let μ represent the i-th sample in the j-th cluster. j Denotes the centroid of the j-th cluster;

[0251] S64. Use the Isolation Forest algorithm to identify abnormal image data and perform anomaly detection on the clustering results:

[0252]

[0253] Where s(x) represents the anomaly score of sample x, E(h(x)) represents the average path length of sample x, c(n) represents the adjustment coefficient, and n6 represents the total number of samples;

[0254] S65. The Structural Similarity Index (SSIM) is used for evaluation, and the intelligently processed image data is compared and analyzed with the original data:

[0255]

[0256] Where, μ x and μ y This represents the average of the x and y values ​​of the image. and σ represents the variance of the images x and y.xy Let C1 and C2 represent the covariance of the images x and y, and C2 be the stability constants.

[0257] S7. Data encryption and access control technologies are used to protect the security of medical image data during storage and transmission.

[0258] In this embodiment, S7 specifically includes:

[0259] S71. Data encryption and access control technologies are used to protect the security of medical image data during storage and transmission.

[0260] S72. Employ the Role-Based Access Control (RBAC) model to control access to encrypted data, defining user roles R and permissions P:

[0261] A1={(r i ,p j )|r i ∈R,p j ∈P};

[0262] Where A1 represents the access control list, r i p represents the i-th user role. j This represents the j-th permission;

[0263] S73. Encrypt the data during transmission using a secure transmission protocol:

[0264]

[0265] Among them, C t This indicates that encrypted data is being transmitted. D represents the transmission encryption function. t k represents data in transmission. t Indicates the transmission of the encryption key;

[0266] S74. Using the Message Authentication Code (MAC) algorithm, the integrity of the transmitted encrypted data is verified, and a data integrity verification code M is generated:

[0267] M = H(D) t +K m );

[0268] Where M represents the message authentication code, H represents the hash function, and D... t K represents data in transmission. m Indicates the message authentication key;

[0269] S75. The data access control module manages and monitors the storage of encrypted and access-controlled medical image data in a central database.

[0270] P a ={(u i ,r i )|u i ∈U,r i ∈R};

[0271] Among them, P a This represents the user role-permission mapping list, u i Let r represent the i-th user. i This represents the role of the i-th user.

[0272] S76. Regularly perform security checks and audits on the stored data, and generate audit reports. a :

[0273]

[0274] Among them, R a E represents the audit report, and u represents the audit event. i Let a represent the i-th user. i t represents the access behavior of the i-th user. i Indicates the access time.

[0275] S8. Through the data monitoring and auditing module, comprehensive monitoring and auditing of access to medical imaging data can be achieved to prevent unauthorized access and data leakage.

[0276] In this embodiment, S8 specifically includes:

[0277] S81. Continuously optimize medical image data by regularly updating compression algorithms, transmission protocols, and processing algorithms:

[0278]

[0279] Where C′ represents the optimized compressed data volume, P′(i) represents the probability of the i-th pixel value after optimization, and n2 is the total number of pixel values;

[0280] S82. Optimize the transmission protocol using transmission optimization techniques based on flow control and congestion control:

[0281]

[0282] Among them, R opt T represents the optimized transmission rate, D represents the total data volume, and T represents the total data volume. opt This represents the optimized total transmission time;

[0283] S83. Employ image processing algorithms based on deep learning and artificial intelligence, optimize the processing algorithms, and improve the accuracy and speed of image recognition and diagnosis:

[0284] F opt =σ(W opt *I+b opt );

[0285] Among them, F opt Let W represent the optimized features, σ represent the activation function, and W represent the optimized features. opt This represents the optimized convolution kernel, where I represents the input image, and b represents the input image. opt This indicates the optimized bias term, and * indicates the convolution operation;

[0286] S84. Optimize the storage structure of image data by adopting a distributed storage system and data sharding technology:

[0287]

[0288] Where S′ represents the optimized slice size, D1 represents the total size of the image data, and n′ represents the optimized number of slices;

[0289] S85 employs advanced data encryption technology and multi-level access control to optimize data security.

[0290] C′=E k′ (P);

[0291] Where C′ represents the optimized ciphertext, E k′ Let P represent the optimized encryption function, P represent the plaintext data, and k′ represent the optimized encryption key.

[0292] S86. Use performance evaluation metrics to evaluate the optimization effect, and monitor and evaluate the optimization process:

[0293]

[0294] Where E represents the optimized evaluation result, This represents the i-th performance metric before optimization. Let represent the i-th performance metric after optimization, and n be the number of performance metrics.

[0295] Example 1:

[0296] To verify the feasibility of this invention, we applied it to the medical information system of a large comprehensive hospital. This hospital generates a large amount of medical imaging data daily, including CT, MRI, X-ray, and ultrasound images. Due to the massive amount of data, how to effectively process, store, transmit, and protect this data has become a major problem for the hospital.

[0297] In this scenario, the hospital's radiology department acquires raw medical image data through the image acquisition module of the medical information system. The system preprocesses the acquired raw medical image data, including image denoising, image enhancement, image segmentation, and image registration. To improve image quality, a Gaussian filter is used for image denoising, and histogram equalization is employed to enhance image contrast, ensuring the clarity and accuracy of the image data. After preprocessing, the system applies a lossless compression algorithm to compress the image data, reducing storage space requirements while ensuring that image quality is not significantly affected.

[0298] In terms of data storage, compressed medical image data is stored in an optimized database structure, employing B+ tree indexing technology to improve storage efficiency and retrieval speed. The system partitions and shards the image data, using a distributed storage system and data sharding technology to enhance storage efficiency and retrieval speed. Image data is categorized as hot and cold data and stored on different storage media to improve data access efficiency.

[0299] During data transmission, the system utilizes an optimized data transmission protocol, employing the TCP / IP protocol stack to ensure the stability and reliability of data transmission. To verify the effectiveness of the optimized transmission protocol, we conducted transmission tests on large-capacity medical image data in both intranet and extranet environments within the hospital. The results show that the optimized transmission speed is significantly improved, reducing the transmission time from 10 minutes to 2 minutes, while the data packet loss rate during transmission decreases from 0.5% to 0.1%.

[0300] The system combines machine learning techniques to intelligently process transmitted and stored medical image data. Convolutional Neural Networks (CNNs) are used to extract image features, and Support Vector Machines (SVMs) are used for classification. Cluster analysis is performed on the classification results, using the k-means clustering algorithm to group similar images into the same class, and the Isolation Forest algorithm to identify abnormal image data. This intelligent processing significantly improves the accuracy and speed of image recognition and diagnosis. Experimental data shows that, through the intelligent processing algorithm, the image data recognition accuracy increased from 85% to 98%, and the processing speed increased by approximately three times.

[0301] To ensure data security, the system employs advanced data encryption technology and a role-based access control (RBAC) model to protect medical image data during storage and transmission. Encryption uses the Advanced Encryption Standard (AES) algorithm to ensure data security during transmission and storage. For access control, the system defines detailed user roles and permissions, encrypts data during transmission using a secure transmission protocol, and verifies data integrity using a Message Authentication Code (MAC) algorithm. Experimental results show that the data leakage rate during transmission is 0 after security processing, ensuring patient privacy and data security.

[0302] To evaluate the overall performance of the system, we monitored and assessed its data processing efficiency, transmission speed, and security over different time periods. The results showed a significant improvement in data processing efficiency, with the daily data volume increasing from 500GB to 1TB; a substantial increase in data transmission speed, reducing transmission time by 80%; and a significant enhancement in data security, with zero data breaches and unauthorized access incidents.

[0303] Table 1 Performance Comparison of Hospital Medical Information System Before and After Optimization

[0304] Data categories Before optimization After optimization Daily data processing volume 500GB 1TB Data transmission time 10 minutes 2 minutes Data packet loss rate 0.5％ 0.1％ Recognition accuracy 85％ 98％ Processing speed 1x 3 times Number of data breaches 3 times / month 0 times / month Number of unauthorized access events 5 times / month 0 times / month

[0305] As shown in Table 1, after implementing the medical information system optimization method of this invention, the hospital achieved significant improvements in data processing, transmission, and security. Daily data processing volume increased from 500GB to 1TB, data transmission time decreased from 10 minutes to 2 minutes, and the data packet loss rate decreased from 0.5% to 0.1%. The recognition accuracy of the intelligent processing algorithm increased from 85% to 98%, and the processing speed increased threefold. Data leakage incidents and unauthorized access incidents were significantly reduced, ensuring the security and reliability of medical image data. These data demonstrate that this invention not only improves the efficiency and accuracy of image data processing but also significantly enhances the stability and security of data transmission, providing strong support for the hospital's medical information system.

[0306] During the application process, radiologists in hospitals reported a significant improvement in the clarity and quality of image data. Particularly during diagnosis, the accurate presentation of image details greatly enhanced diagnostic accuracy. Through image denoising and enhancement processing, noise in the images was effectively removed, and image contrast was significantly enhanced. Especially in the identification of complex lesions and subtle structures, doctors were able to see image details more clearly, leading to more accurate diagnoses.

[0307] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for optimizing medical image data in a medical information system, characterized in that, Includes the following steps: S1. Data is collected through the image acquisition module of the medical information system to obtain raw medical image data; S2. Preprocess the acquired raw medical image data, including image denoising, image enhancement, image segmentation, and image registration. S3. Apply lossless compression algorithm to compress the preprocessed medical image data to reduce storage space requirements while ensuring that image quality is not significantly affected. S4. Store the compressed medical image data in the database structure and manage it through the data storage management module; S5. Using data transmission protocols, the compressed medical image data is transmitted within the medical information system; S6. Combine machine learning technology to intelligently process transmitted and stored medical image data; S6 specifically includes: S61. Convolutional Neural Networks (CNNs) are used to extract image features. Combined with machine learning techniques, intelligent processing of transmitted and stored medical image data is performed to extract features from the image data. F=σ1(W*I+b 1 ); Where F represents the extracted features, σ1 represents the activation function, W represents the convolution kernel, I represents the input image, and b 1 * indicates the bias term, and * indicates the convolution operation; S62. Input the extracted image features into a Support Vector Machine (SVM) for classification: Where w represents the weight vector, C represents the penalty parameter, and ξ i Let n represent the slack variable and n represent the total number of samples. S63. Perform cluster analysis on the classification results, and use the k-means clustering algorithm to group similar images into the same class: Where J represents the clustering objective function, and k3 represents the number of clusters. Let μ represent the i-th sample in the j-th cluster. j Denotes the centroid of the j-th cluster; S64. Use the Isolation Forest algorithm to identify abnormal image data and perform anomaly detection on the clustering results: Where s(x2) represents the anomaly score of sample x2, E(h(x2)) represents the average path length of sample x2, c(n) represents the adjustment coefficient, and n represents the total number of samples; S65. The Structural Similarity Index (SSIM) is used for evaluation, and the intelligently processed image data is compared and analyzed with the original data: Where, μ x and μ y This represents the average of the x and y values ​​of the image. and σ represents the variance of the images x and y. xy Let C1 and C2 represent the covariance of the images x and y, and C2 be the stationary constants. S7. Data encryption and access control technologies are used to protect the security of medical image data during storage and transmission. Specifically, S7 includes: S71. Data encryption and access control technologies are used to protect the security of medical image data during storage and transmission. S72. Employ the Role-Based Access Control (RBAC) model to control access to encrypted data, defining user roles R and permissions P: Where A1 represents the access control list, This represents the i8th user role. This indicates the j3rd permission; S73. Encrypt the data during transmission using a secure transmission protocol: Among them, C t This indicates that encrypted data is being transmitted. D represents the transmission encryption function. t k represents data in transmission. t Indicates the transmission of the encryption key; S74. Using the Message Authentication Code (MAC) algorithm, the integrity of the transmitted encrypted data is verified, and a data integrity verification code M is generated: M=H(D t +K m ); Where M represents the message authentication code, H represents the hash function, and D... t K represents data in transmission. m This represents the message authentication key, and m represents the size of the hash table; S75. The data access control module manages and monitors the storage of encrypted and access-controlled medical image data in a central database. Among them, P a This represents the user role-permission mapping list. This represents the i-th user. This represents the i8th user role; S76. Regularly perform security checks and audits on the stored data, and generate audit reports. a : Among them, R a E represents the audit report, and E represents the audit event. This represents the i-th user. This represents the access behavior of the i8th user. Indicates the access time; S8. Through the data monitoring and auditing module, comprehensive monitoring and auditing of access to medical imaging data can be achieved to prevent unauthorized access and data leakage.

2. The method for optimizing medical image data in a medical information system according to claim 1, characterized in that, S1 specifically includes: S11. Configure an image acquisition module in the medical information system. The image acquisition module includes an image acquisition device and a data interface to receive various types of raw medical image data. S12. Convert the received raw medical image data into a unified DICOM format, converting image data from different sources and formats. S13. Perform a preliminary quality check on the DICOM format image data: Where Q represents the image quality score. Indicates the weighting coefficient. Indicates the i-th 1 There are 1 quality characteristic functions, where n1 is the number of quality characteristics; S14. Timestamp and record metadata for image data that has passed quality inspection: M1={P 1 ,T 1 ,E 1 }; Where M1 represents the meta-dataset, P 1 Indicates patient information, T 1 E represents the image acquisition time. 1 Indicates device information; S15. Transmit the DICOM format image data with timestamps and metadata to the buffer area of ​​the image acquisition module through the data interface.

3. The method for optimizing medical image data in a medical information system according to claim 2, characterized in that, S2 specifically includes: S21. A Gaussian filter is used to reduce noise in the image, and image denoising processing is performed on the acquired raw medical image data: Among them, I denoised I(x) represents the denoised image, σ is the standard deviation of the Gaussian filter, and I(x) is the standard deviation of the Gaussian filter. q ,y q () represents the original image data, and * represents the convolution operation; S22. Histogram equalization is used to enhance image contrast; image enhancement processing is performed on the denoised medical image data. Among them, I enhanced L represents the enhanced image. 1 This represents the number of gray levels, min(I denoised ) and max(I denoised These represent the minimum and maximum gray values ​​of the image, respectively. S23. The Otsu algorithm is used to automatically determine the segmentation threshold T, and image segmentation processing is performed on the enhanced medical image data: in, Let ω1(T) represent the inter-class variance, ω2(T) represent the probabilities of foreground and background pixels, and μ1(T) and μ2(T) represent the average gray values ​​of foreground and background pixels, respectively. S24. A gradient descent-based mutual information method is used for registration to perform image registration processing on the segmented medical image data: Where MI(A,B) represents the mutual information between image A and image B, p(a,b) represents the joint probability distribution, and p(a) and p(b) represent the edge probability distributions of image A and image B, respectively; S25. Transmit the medical image data after image denoising, enhancement, segmentation and registration to the image data processing module.

4. The method for optimizing medical image data in a medical information system according to claim 3, characterized in that, S3 specifically includes: S31. Using Huffman coding for lossless compression, the preprocessed medical image data is compressed using a lossless compression algorithm: Where C(I) represents the compressed data size, P(i) 2 ) represents the i-th 2 The probability of a pixel value, where n2 is the total number of pixel values; S32. Using Discrete Cosine Transform (DCT) and quantization processing, a lossy compression algorithm is applied to compress some medical image data: Where F(u,v) represents the DCT coefficients, I(x) q ,y q ) represents the original image pixel value, N is the size of the image patch, α(u) and α(v) are normalization coefficients, u represents the frequency index in the horizontal direction, and v represents the frequency index in the vertical direction; S33. Quantize the lossy compressed data: Where Q(u,v) represents the quantized DCT coefficients, QF(u,v) represents the coefficients of the quantization table, and F(u,v) represents the quantized DCT coefficients. S34. Using Inverse Discrete Cosine Transform (IDCT), the compressed medical image data is processed by inverse transformation to restore approximate values ​​of the original image: Where I(x) q ,y q ) represents the restored image pixel values, and F(u,v) represents the quantized DCT coefficients; S35. Store the compressed medical image data in an optimized database structure and manage it through the data storage management module.

5. A method for optimizing medical image data in a medical information system according to claim 4, characterized in that, S4 specifically includes: S41. The database structure uses B+ tree indexing technology to store compressed medical image data in an optimized database structure. Data storage location L: Where B represents the branching factor of the B+ tree, and h represents the height of the tree. This indicates the number of keywords in the i9th layer; S42. A hash table structure is used to store metadata, enabling metadata management of image data and improving retrieval efficiency. H(k1) = k1m; Where H(k1) represents the hash value, k1 represents the number of keys in the first level, and m represents the size of the hash table; S43. Based on the different types and usage frequencies of the image data, the image data is partitioned and stored into hot data and cold data, which are then stored on different storage media: Among them, R 1 D represents the proportion of thermal data. hot D represents the amount of thermal data. c Indicates the amount of cold data; S44. Based on the data size, the image data is segmented into several small segments for storage: in, Indicates the i-th 3 The size of each data slice, where D1 represents the total size of the image data and n8 represents the number of slices; S45. RAID technology is used for redundant storage of image data, and storage redundancy management is implemented for image data. Among them, R 2 This represents the total amount of data after redundant storage. Indicates the i-th 3 Data shards, P 2 This indicates redundant check data; S46. Employ a combination of incremental and full backup methods for image data backup management: Where B1 represents the total amount of backup data, B full This indicates the amount of data in the full backup. This represents the data volume of the i4th incremental backup, and m1 is the number of incremental backups.

6. A method for optimizing medical image data in a medical information system according to claim 5, characterized in that, S5 specifically includes: S51. Transmit compressed medical image data within the medical information system using a data transmission protocol: Among them, R t T represents the transmission rate, D represents the total amount of data, and T represents the total data volume. t Indicates the total transmission time. Indicates the i-th 5 The size of each data block Indicates the i-th 5 Bandwidth per data block; S52. Use cyclic redundancy check (CRC) to detect and correct errors in the data during transmission: CRC=(D(x q )·x q r )G(x q ); Where CRC represents Cyclic Redundancy Check code, D(x q ) represents a data polynomial, x q This represents the shift operator, r represents the number of redundant bits, and G(x) represents the bit shift operator. q ) represents the generator polynomial. k1 represents the number of data bits, and k2 represents the number of data bits. represents the coefficients of the generator polynomial, and r represents the number of redundant bits; S53. The AES encryption standard algorithm is used to encrypt the data during transmission; S54. Use the AES algorithm to decrypt the encrypted data; S55. Divide the data D into several blocks, each block being Q bits in size. For each data block... Perform the initial hash operation, and the hash value is represented as: All initial hash values Connect them to obtain the intermediate hash value H. mid For the intermediate hash value H mid Perform the final hash operation to obtain the final data digest H. 1 The generated data digest H 1 Compare the hash value with the expected hash value to verify data integrity: ifH=H expected Then data integrity verification passed; in, Indicates the i-th 5 There are n5 data blocks, where n5 is the number of data blocks, H0 is the initial hash value, and H is the hash function. Indicates the i-th 5 The j4th bit in the data block, H mid [i 7 ] represents the i-th intermediate hash value. 7 H hash values expected Indicates the expected hash value; S56. Store the medical image data that has been transmitted and passed integrity verification in the central database.

7. A method for optimizing medical image data in a medical information system according to claim 1, characterized in that, S8 specifically includes: S81. Continuously optimize medical image data by regularly updating compression algorithms, transmission protocols, and processing algorithms: Among them, C ′ P represents the optimized compressed data volume. ′ (i 2 ) represents the optimized i-th 2 The probability of a pixel value, where n2 is the total number of pixel values; S82. Optimize the transmission protocol using transmission optimization techniques based on flow control and congestion control: Among them, R opt T represents the optimized transmission rate, D represents the total data volume, and T represents the total data volume. opt This represents the optimized total transmission time; S83. Employ image processing algorithms based on deep learning and artificial intelligence, optimize the processing algorithms, and improve the accuracy and speed of image recognition and diagnosis: F opt =σ1(W opt *I+b opt ); Among them, F opt Let W represent the optimized features, σ1 represent the activation function, and W represent the optimized features. opt This represents the optimized convolution kernel, where I represents the input image, and b represents the input image. opt This indicates the optimized bias term, and * indicates the convolution operation; S84. Optimize the storage structure of image data by adopting a distributed storage system and data sharding technology: Among them, S ′ This indicates the optimized slice size, where D1 represents the total size of the image data, and n... ′ Indicates the number of shards after optimization; S85 employs advanced data encryption technology and multi-level access control to optimize data security. C ′ =E k′ (P 3 ); Among them, C ′ E represents the optimized ciphertext. k′ P represents the optimized encryption function. 3 k represents plaintext data. ′ This represents the optimized encryption key; S86. Use performance evaluation metrics to evaluate the optimization effect, and monitor and evaluate the optimization process: Among them, E 2 This indicates an optimized evaluation result. This represents the i9th performance metric before optimization. This represents the i-th performance metric after optimization, and n3 is the number of performance metrics.

Images