Method and system for converting medical data resources into data assets

Abstract

The invention relates to the field of data processing, in particular to a method and system for converting medical data resources into data assets. The method comprises the following steps: acquiring medical data from different sources and preprocessing the medical data to obtain preprocessed medical data; encrypting the preprocessed medical data by adopting an encryption algorithm to obtain encrypted medical data; a machine learning algorithm is adopted to analyze the encrypted medical data, so that valuable insight information is extracted, and the valuable insight information is information capable of generating actual guiding significance for clinical diagnosis and treatment, hospital management, scientific research or public health decision making; and a report is generated according to the valuable insight information for decision reference of medical institutions. By adopting the method provided by the invention, the data can be efficiently converted into the data assets on the basis of ensuring the data privacy.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing. More specifically, the present application relates to a method and system for transforming medical data resources into data assets. BACKGROUND

[0002] Data assets refer to those data resources that have been organized, classified and managed, and have potential economic or business value. They can be reused, shared, monetized, and even included in the balance sheet of an organization.

[0003] In the prior art, electronic health record (EHR) systems are the main way to store and manage patient information. These systems usually use data warehouse and data lake architectures to centrally manage and process medical data from different sources. In order to transform the medical data stored in electronic health records into data assets, it is usually necessary to clean, standardize and analyze the data. In the prior art, Apache Hadoop, Spark and other big data processing frameworks are often used in combination with Tableau, PowerBI and other visualization tools to analyze and process medical data, thereby transforming it into data assets. Although the above method can effectively analyze a large amount of medical data, there are still vulnerabilities in ensuring data privacy and preventing unauthorized access. In addition, the above method requires a large amount of time and computing resources when cleaning data in large-scale data sets, resulting in low efficiency in transforming medical data into data assets. SUMMARY

[0004] To solve the technical problem of poor data privacy and security in the process of transforming medical data into data assets in the prior art, the present application provides solutions in the following aspects.

[0005] In a first aspect, the present application provides a method for transforming medical data resources into data assets, comprising:

[0006] Collecting medical data from different sources and preprocessing them to obtain preprocessed medical data;

[0007] Encrypting the preprocessed medical data using an encryption algorithm to obtain encrypted medical data;

[0008] Analyzing the encrypted medical data using a machine learning algorithm to extract valuable insight information, which refers to information that can have practical guiding significance for clinical diagnosis and treatment, hospital management, scientific research or public health decision-making;

[0009] Generating a report based on the valuable insight information for reference by medical institutions in decision-making.

[0010] The method has the beneficial effects that: the method of the application firstly encrypts the medical data after collecting the medical data, and then automatically extracts information that can have practical guiding significance for clinical diagnosis and treatment, hospital management, scientific research or public health decision-making from the medical data by using a machine learning algorithm after encryption, thereby improving the privacy and security of medical data in the process of converting medical data into data assets.

[0011] Preferably, the source of the medical data includes: a hospital information system, a laboratory information system (LIS), and a picture archiving and communication system.

[0012] Preferably, the preprocessing of the medical data includes: cleaning and standardizing the medical data, the cleaning includes: realizing dynamic task slicing by a PowerJob engine, cutting the cleaning task into sub-tasks according to data sources and performing the sub-tasks in parallel; and the standardization includes: uniformly encoding data of different sources according to international standards.

[0013] The beneficial effects are that: by cutting the cleaning task into sub-tasks according to data sources and performing the sub-tasks in parallel, a millisecond-level scheduling mechanism is realized, thereby improving the NLP text analysis efficiency by 80% (such as structured processing of doctor's notes), automatically transferring faults to ensure 7x24 hours continuous processing, avoiding queue blocking of traditional batch processing, and improving the efficiency of converting medical data into data assets.

[0014] Preferably, the encryption of the preprocessed medical data by using an encryption algorithm includes:

[0015] selecting an encryption algorithm;

[0016] using a PowerJob system to perform workflow orchestration and encryption pipeline, and automatically distributing CPU-intensive tasks to special computing nodes;

[0017] establishing a secure key management system to ensure secure storage and distribution of keys.

[0018] The beneficial effects are that: the lightweight distributed processing engine based on PowerJob used when encrypting the medical data reduces the task scheduling delay to milliseconds (80% efficiency improvement compared with traditional batch processing mode) through its dynamic task slicing and real-time scheduling capabilities. The encryption speed of the medical data is greatly improved.

[0019] Preferably, the encryption algorithm is an advanced encryption standard or a homomorphic encryption technology.

[0020] Preferably, it further includes: deploying a smart contract to manage access permissions of the medical data.

[0021] By deploying the smart contract, it can be ensured that only authorized users can decrypt and view the data. Thus, unauthorized access to medical data by users is prevented, further improving the privacy and security of medical data.

[0022] Preferably, deploying the smart contract comprises:

[0023] Developing a smart contract based on a blockchain to set users allowed to access, and access rights and access conditions of the users allowed to access;

[0024] Taking the PowerJob task scheduling platform as a real-time scheduling engine of the smart contract;

[0025] Verifying the identity of the user by means of a digital certificate and a two-factor authentication method.

[0026] Preferably, it further comprises: recording access activities and modification activities of the medical data by means of a distributed ledger technology.

[0027] Preferably, the valuable insight information comprises one or more of association rules, clustering patterns, abnormal information, potential trends, disease development likelihoods, and trends or patterns of disease development of patients; the association rules are used to represent the corresponding relationship between different symptom combinations and diseases, the clustering patterns refer to clustering patients according to symptoms, genetic characteristics, and lifestyles to obtain susceptible populations of certain diseases, the abnormal information includes whether the hospitalization rate of patients is too high, whether the frequency of medical insurance reimbursement is too high, whether the frequency of drug side effects is too high, and whether the frequency of image examination application is too high; the potential trends are used to represent the directional changes exhibited by the medical data over time.

[0028] In a second aspect, the present application provides a system for converting medical data resources into data assets, comprising a processor and a memory, wherein the memory stores computer program instructions which, when executed by the processor, implement the method for converting medical data resources into data assets of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0029] The above and other objects, features and advantages of the exemplary embodiments of the present application will be readily understood through reading the detailed description of the exemplary embodiments of the present application below, with reference to the accompanying drawings. The exemplary embodiments of the present application are illustrated in the drawings, in which:

[0030] Figure 1 is a flowchart schematically showing a method for converting medical data resources into data assets according to an embodiment of the present application;

[0031] Figure 2is a schematic diagram illustrating a method of encrypting preprocessed medical data according to an embodiment of the present application;

[0032] Figure 3 is a schematic diagram of a system structure for converting medical data resources into data assets according to an embodiment of the present application. DETAILED DESCRIPTION

[0033] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0034] The specific embodiments of the present application will be described in detail below with reference to the drawings.

[0035] Embodiment of a method for converting medical data resources into data assets:

[0036] As shown in Figure 1 the method for converting medical data resources into data assets of the present application comprises:

[0037] S101, collecting medical data of different sources and preprocessing the medical data, thereby obtaining preprocessed medical data;

[0038] In this embodiment, the sources of medical data include: hospital information system, laboratory information system (LIS) and picture archiving and communication system. In other embodiments, the sources of medical data can also include other types of sources.

[0039] The types of collected data include but are not limited to electronic medical records (EMR), diagnosis reports, image data, genomic data, etc.

[0040] In this embodiment, preprocessing the medical data includes cleaning and standardizing the medical data. The cleaning includes: implementing dynamic task fragmentation through a PowerJob engine, cutting the cleaning task into subtasks according to data sources and executing the subtasks in parallel; the cleaning task includes outlier filling and time alignment.

[0041] By cutting the cleaning task into subtasks according to data sources and executing the subtasks in parallel, a millisecond-level scheduling mechanism is realized, thereby improving the NLP text analysis efficiency by 80% (such as doctor's note structured processing), automatically transferring faults to ensure 7x24 hours continuous processing, and avoiding queue blocking of traditional batch processing.

[0042] The standardization includes: uniformly encoding data of different sources according to international standards.

[0043] By uniformly encoding different sources of data, it can be ensured that all data follows the same format and naming rules for subsequent analysis.

[0044] S102, encrypting the preprocessed medical data by using an encryption algorithm to obtain encrypted medical data;

[0045] S103, extracting valuable insight information, specifically including: using a machine learning algorithm to analyze the encrypted medical data, thereby extracting valuable insight information, the valuable insight information being information that can have actual guiding significance for clinical diagnosis and treatment, hospital management, scientific research or public health decision-making;

[0046] Valuable insight information is usually hidden in raw medical data and cannot be directly observed, but after mining by statistical modeling, machine learning and other methods, potential patterns, trends or correlation relationships can be revealed.

[0047] In this embodiment, the valuable insight information includes one or more of association rules, clustering patterns, trends or patterns of patient disease progression, abnormal information, disease development probability and potential trends.

[0048] Association rules are used to represent the correspondence between different symptom combinations and diseases, for example, the combination of cough and low oxygen saturation corresponds to pneumonia. Apriori algorithm can be used to extract association rules from raw medical data, for example: through the chief complaint and physical examination text in the electronic medical record, and then through the Apriori algorithm to obtain possible high-probability diseases.

[0049] Clustering patterns refer to clustering patients according to symptoms, genetic characteristics and lifestyle to obtain susceptible populations for certain diseases, such as collecting patient genetic sequencing data (SNP sites), lifestyle questionnaire (smoking / diet), clinical indicators (BMI, blood pressure) and other data, and identifying cardiovascular disease susceptible populations such as "high genetic risk + sedentary" through K-means clustering.

[0050] The trend or pattern of patient disease progression can be the progression path of the disease or the timeline of the treatment response. PrefixSpan algorithm can be used to extract the trend or pattern of patient disease progression from raw medical data, for example: through continuous follow-up records (such as ECG time series signals), hospital logs (inspection / treatment timestamps) and other data, the PrefixSpan algorithm can infer the conversion sequence of patients from "mild cognitive impairment → Alzheimer's disease", and after obtaining the sequence data, other complications and conversion time can be inferred, or continuous medication records can be added to infer the best treatment plan.

[0051] Abnormal information includes whether the patient's hospitalization rate is too high, whether the medical insurance reimbursement frequency is too high, whether the drug side effect frequency is too high, and whether the imaging examination application frequency is too high. The above abnormal information of the patient can be detected by the isolation forest algorithm.

[0052] The original medical data can be analyzed by a neural network model to extract the disease development possibility, such as whether the patient will develop into a chronic disease, how much probability, and how to avoid it. For example, the multi-year medical examination indicators (including blood glucose / lipid trend) of a pre-diabetic patient, family history, and pollution index of the place of residence are obtained, and then the LSTM model is used to predict the probability of the patient converting from pre-diabetes to type II diabetes.

[0053] Potential trends are used to characterize the directional changes exhibited by medical data over time. In the medical field, potential trends can include but are not limited to the following:

[0054] Changes in disease incidence, such as the inflection point of the change trend identified by the Joinpoint regression model, which can represent the incidence trend of cardiovascular disease. From 1990 to 2019, the age-standardized incidence of coronary heart disease showed an "up-plateau" curve, and fell to 197.4 / 100,000 person-years in 2019.

[0055] Changes in treatment effectiveness over time, such as real-world evidence (RWE) analysis, commonly known as clinical trials, to detect blood glucose in diabetic patients, monitor patient's dynamic blood glucose in real time, improve blood glucose compliance rate of diabetic patients, and reduce patient's hypoglycemic events.

[0056] Changes in patient health indicators over time, such as monitoring blood glucose fluctuations, which requires continuous glucose monitoring (CGM) to continuously record subcutaneous tissue fluid glucose concentration to generate high-precision data points, and then aligning multi-source time series data (such as blood glucose meter, smart bracelet data) using dynamic time warping (DTW) to determine the trend of patient's blood glucose level over time, and analyzing factors affecting patient's blood glucose rise and fall according to the trend.

[0057] Before analyzing encrypted medical data, Apache Hadoop, Spark, and other big data frameworks can be used to process large-scale medical data sets; when processing large-scale data sets, structured data can be stored by accessing multiple source heterogeneous data.

[0058] S104, generating a report according to the valuable insight information for reference of medical institution decision-making.

[0059] The report or the suggestion can be automatically generated periodically.

[0060] The generated report includes a report of key indicators, trend prediction, etc. Customized report templates are supported to meet the needs of different departments.

[0061] In the embodiment, the report is generated only by using medical data, and in other embodiments, suggestions can also be generated. The generated suggestions can be suggestions for optimizing treatment plans or suggestions for improving service quality.

[0062] After generating the report or the suggestion, external data sources (such as public health statistical data) can be integrated to provide a more comprehensive perspective to help develop strategic planning.

[0063] The method of the present application first encrypts the medical data after collecting it, and then automatically extracts information that can have actual guiding significance for clinical diagnosis and treatment, hospital management, scientific research or public health decision-making from it using a machine learning algorithm, thereby efficiently converting it into data assets on the basis of ensuring data privacy.

[0064] In one embodiment, the process of training the neural network includes:

[0065] 1) Multi-modal network structure design:

[0066] 1.1) Design neural network models suitable for various types of data.

[0067] For image data in medical data, a 3D Swin Transformer and 3D-CNN dual-branch structure is used. This dual-branch structure can capture the global cross-slice correlation of the lesion and finely extract local features, and the actual lesion detection rate is improved by 9%.

[0068] 3D Swin Transformer is a Transformer architecture suitable for three-dimensional input (such as voxels, video frames, CT images) with a hierarchical window attention mechanism (Shifted Window Attention) as the core, balancing long-distance modeling capability and efficient computing performance.

[0069] For time series data in medical data, Hyena-LSTM neural network is used instead of traditional BiLSTM neural network, thereby significantly enhancing long-range dependency capturing capability, reducing disease course prediction error by 18%. GATv2 graph attention network is introduced for gene data, modeling gene-protein interaction network, and rare disease association discovery rate is improved by 23%.

[0070] 1.2) Data fusion: Adopt dynamic gated fusion layer to realize cross-modal feature adaptive weighting through learnable gating weights. At the same time, add semantic alignment constraints, use contrastive learning to ensure the consistency of multi-modal representations of the same patient, and make the EHR text-image feature similarity reach more than 0.92.

[0071] In this embodiment, EHR text refers to the free text part in the electronic health record (EHR), which usually includes unstructured text content input by medical staff such as doctors, nurses, and laboratory technicians during patient diagnosis and treatment.

[0072] 1.3) Medical data-specific initialization: Xavier initialization combined with spectral normalization is used for high-dimensional gene data to improve gradient stability by 40%; He initialization is used for medical images with channel attention weighting, and ReLU activation efficiency is improved by 35%; UMLS knowledge graph is used to fine-tune the BERT model for clinical text, and the F1-score of medical terminology recognition reaches 0.91.

[0073] Xavier initialization is a method for initializing weights in deep neural networks to address the problem of gradient vanishing or explosion that may occur at the beginning of network training. He initialization is a weight initialization method designed to address the problem of gradient vanishing / explosion caused by ReLU activation function.

[0074] 1.4) Adopt federated learning framework: Reduce communication overhead by 76% through sparse gradient transmission technology, while keeping model accuracy loss less than 2%.

[0075] 2) Design privacy protection training mechanism, including:

[0076] 2.1) Batch design: Combine differential privacy technology to realize adaptive noise injection, dynamically adjust noise intensity according to gradient norm, and fully comply with HIPAA compliance requirements.

[0077] HIPAA compliance (HIPAA Compliance) refers to the organization or individual handling medical-related data must comply with the privacy protection and security rules in the Health Insurance Portability and Accountability Act (Health Insurance Portability and Accountability Act) to protect patients' personal health information (PHI) from being leaked, misused, or illegally accessed.

[0078] 2.2) Design a clinically-oriented evaluation system: Introduce clinically interpretable indicators: such as 3D Grad-CAM++ technology to generate lesion positioning heat maps and output structured decision reports containing feature contribution.

[0079] 3D Grad-CAM++ is a technique for visualizing three-dimensional neural network models, with the goal of explaining the regions of voxels (i.e., three-dimensional regions in space) that the model focuses on when making a prediction.

[0080] 2.3) Deployment test process: Knowledge distillation is used to compress the 3D model into a lightweight 2.5D network, reducing the inference delay to 28ms, while achieving a high throughput of 1200 req / s through the TensorRT quantization engine. The continuous learning mechanism adds a catastrophic forgetting suppression module to ensure stable maintenance of new disease recognition ability. Deployment API service encapsulates the model into a DICOM standard interface, which is connected to hospital PACS and other systems to output structured reports.

[0081] 2.4) Perform data analysis under privacy protection: Apply differential privacy technology to maximize the protection of individual privacy while ensuring data accuracy. For specific types of sensitive data, partial calculations can be performed on local devices before aggregating the results.

[0082] In one embodiment, as shown in Figure 2 the preprocessed medical data is encrypted using an encryption algorithm, including:

[0083] S201, selecting an encryption algorithm;

[0084] In this embodiment, the encryption algorithm is the Advanced Encryption Standard or homomorphic encryption technology. In other embodiments, other suitable encryption algorithms can also be used.

[0085] If the Advanced Encryption Standard is used, the AES-256 encryption algorithm can be selected. By using the Advanced Encryption Standard to encrypt medical data, the efficiency and security of encryption can be guaranteed.

[0086] For tasks that need to be executed without decryption, homomorphic encryption technology can be used.

[0087] S202, encryption acceleration processing: PowerJob system is used for workflow orchestration and encryption pipeline, and CPU-intensive tasks are automatically distributed to dedicated computing nodes;

[0088] The lightweight distributed processing engine based on PowerJob used in the encryption of medical data reduces the task scheduling delay to milliseconds through its dynamic task sharding and real-time scheduling capabilities, improving efficiency by 80% compared to traditional batch processing mode. The engine supports workflow orchestration and automatic fault transfer, and can handle 10 times the size of tasks in parallel under the same hardware resources, avoiding the waste of resources caused by queue accumulation in traditional architectures. According to the test comparison, the data processing throughput is 3.2 times that of existing technology, and the CPU / memory occupancy rate is reduced by 45%, the time consumption for encrypting 100GB of genomic data is reduced by 67%, greatly improving the encryption speed of medical data.

[0089] S203, Key Management: Establish a secure key management system to ensure secure storage and distribution of keys.

[0090] During the management of medical data, it is necessary to implement a regular key replacement strategy to enhance data protection.

[0091] In one embodiment, it further includes deploying a smart contract to manage access permissions for medical data.

[0092] By deploying a smart contract, only authorized users can decrypt and view data, preventing unauthorized access to medical data and further improving the privacy and security of medical data.

[0093] In one embodiment, deploying a smart contract to manage access permissions for medical data includes:

[0094] S301, Develop a smart contract based on blockchain to set users allowed to access and access permissions and conditions for users allowed to access;

[0095] The smart contract clearly specifies the granting and revoking processes of access permissions, as well as punishment measures for violations.

[0096] S302, Use the PowerJob task scheduling platform as the real-time scheduling engine of the smart contract;

[0097] By using the PowerJob real-time scheduling engine, the smart contract triggering efficiency is reduced from minutes to <20ms.

[0098] S303, Verify user identity using digital certificate method and two-factor authentication method.

[0099] By verifying the roles and responsibilities of users, appropriate access permissions can be assigned to users.

[0100] In one embodiment, it further includes using distributed ledger technology to record access and modification activities of medical data.

[0101] Due to the adoption of distributed ledger technology, each data interaction generates an unalterable log entry for audit trail use.

[0102] By recording the access activities and modification activities of medical data by using distributed ledger technology, the user can understand the historical changes of data through query, and the transparency of medical data is improved.

[0103] Embodiments of a system for converting medical data resources into data assets:

[0104] The present application also provides a system for converting medical data resources into data assets. As shown in Figure 3 The system for converting medical data resources into data assets includes a processor and a memory, and the memory stores computer program instructions, which, when executed by the processor, implement a method for converting medical data resources into data assets according to the above embodiments of the present application.

[0105] The system for converting medical data resources into data assets also includes a communication bus and a communication interface and other components familiar to those skilled in the art, the settings and functions of which are known in the art, and therefore will not be described here.

[0106] In the present application, the aforementioned memory can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or apparatus. For example, the computer readable storage medium can be any suitable magnetic storage medium or magneto-optical storage medium, such as resistive random access memory RRAM (Resistive Random Access Memory), dynamic random access memory DRAM (Dynamic Random Access Memory), static random access memory SRAM (Static Random-Access Memory), enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), high bandwidth memory HBM (High-Bandwidth Memory), hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store desired information and can be accessed by an application, module or both. Any such computer storage medium can be part of the device or accessible or connectable to the device. Any application or module described in the present application can be implemented using computer readable / executable instructions that can be stored or otherwise held by such computer readable medium.

[0107] In the description of the present specification, the meaning of "a plurality of", "several" is at least two, such as two, three or more, etc., unless otherwise explicitly specifically limited.

[0108] While the present specification has shown and described a number of embodiments of the application, it is to be understood that such embodiments are merely illustrative of the many possible embodiments thereof. Numerous modifications, adaptations, and variations will be apparent to those skilled in the art in view of the foregoing description. It is to be understood that, in the course of practicing the application, various alternative arrangements can be adopted by those skilled in the art.

Claims

1. A method for transforming medical data resources into data assets, characterized in that, The method comprises the following steps: Collecting medical data from different sources and preprocessing the medical data to obtain preprocessed medical data; Encrypting the preprocessed medical data using an encryption algorithm to obtain encrypted medical data; Analyzing the encrypted medical data using a machine learning algorithm to extract valuable insight information, which refers to information that can provide practical guidance for clinical diagnosis and treatment, hospital management, scientific research, or public health decision-making; Generating a report based on the valuable insight information for reference by medical institutions.

2. The method for transforming medical data resources into data assets of claim 1, wherein, The sources of the medical data include a hospital information system, a laboratory information system (LIS), and a picture archiving and communication system.

3. The method for transforming medical data resources into data assets of claim 1, wherein, The preprocessing of the medical data includes cleaning and standardizing the medical data, wherein the cleaning includes implementing dynamic task partitioning through a PowerJob engine to divide cleaning tasks into subtasks according to data sources and perform the subtasks in parallel; and the standardization includes uniformly encoding data from different sources according to international standards.

4. The method for transforming medical data resources into data assets of claim 1, wherein, The encryption of the preprocessed medical data using an encryption algorithm includes: Selecting an encryption algorithm; Using a PowerJob system to arrange a workflow and an encryption pipeline, and automatically assigning CPU-intensive tasks to dedicated computing nodes; Establishing a secure key management system to ensure secure storage and distribution of keys.

5. The method for transforming medical data resources into data assets of claim 4, wherein, The encryption algorithm is the advanced encryption standard or homomorphic encryption technology.

6. The method for transforming medical data resources into data assets of claim 1, wherein, The method further comprises: Deploying a smart contract to manage access permissions for the medical data.

7. The method for transforming medical data resources into data assets of claim 6, wherein, The deployment of the smart contract includes: Developing a smart contract based on a blockchain to set users allowed to access and access permissions and conditions for the users allowed to access; Using a PowerJob task scheduling platform as a real-time scheduling engine for the smart contract; Verifying the identity of a user using a digital certificate method and a two-factor authentication method.

8. The method for transforming medical data resources into data assets of claim 1, wherein, The method further comprises: Recording access activities and modification activities of the medical data using a distributed ledger technology.

9. The method for transforming medical data resources into data assets of any of claims 1-8, wherein, The valuable insight information includes one or more of association rules, clustering patterns, abnormal information, potential trends, disease development likelihoods, and trends or patterns of disease progression for patients; the association rules are used to represent the corresponding relationship between different symptom combinations and diseases, the clustering patterns refer to clustering patients according to symptoms, genetic characteristics, and lifestyles to obtain susceptible populations for a certain disease, the abnormal information includes whether the hospitalization rate of a patient is too high, whether the frequency of medical insurance reimbursement is too high, whether the frequency of drug side effects is too high, and whether the frequency of image examination application is too high; and the potential trends are used to represent the directional changes exhibited by the medical data over time.

10. A system for transforming medical data resources into data assets, comprising a processor and a memory, the memory storing computer program instructions, wherein, When the computer program instructions are executed by the processor, the method for converting medical data resources into data assets according to any one of claims 1-9 is implemented.

Images