Periodic Reporting for period 2 - INCISIVE (A multimodal AI-based toolbox and an interoperable health imaging repository for the empowerment of imaging analysis related to the diagnosis, prediction and follow-up of cancer)
Okres sprawozdawczy: 2022-04-01 do 2023-03-31
INCISIVE is a 42-month project that aims at exploring the potential of novel AI tools for enhancing current imaging solutions for cancer cases thus supporting more effective decision-making for Healthcare Providers. It also addresses challenges related to data labelling and annotation, as well as availability and sharing of imaging data so that it can be used for training and validating AI tools for improved imaging methods.
The project targets two main results: A) an AI-based toolbox consisting of novel AI models, combined with a set of predictive, descriptive and prescriptive analytics, enabling the multi-modal exploration of the available data sources; this includes a Machine Learning (ML-) based automatic annotation system to produce data for the training of algorithms in AI research B) an interoperable pan-European federated repository of health images that allows the sharing of data in compliance with legal, ethical, privacy and security requirements, for AI-related training and experimentation.
The project solutions will be piloted on four types of cancer: lung, colorectal, breast and prostate cancer. The pilot activities will be carried out in 8 sites across 5 countries Greece, Italy, Spain, Cyprus and Serbia.
The project targets two main results: A) an AI-based toolbox consisting of novel AI models, combined with a set of predictive, descriptive and prescriptive analytics, enabling the multi-modal exploration of the available data sources; this includes a Machine Learning (ML-) based automatic annotation system to produce data for the training of algorithms in AI research B) an interoperable pan-European federated repository of health images that allows the sharing of data in compliance with legal, ethical, privacy and security requirements, for AI-related training and experimentation.
The project solutions will be piloted on four types of cancer: lung, colorectal, breast and prostate cancer. The pilot activities will be carried out in 8 sites across 5 countries Greece, Italy, Spain, Cyprus and Serbia.
During the project's second period (M19-Μ30) INCISIVE partners report good progress at all project levels and several important achievements.
Major emphasis was placed by the PM on establishing a fast start approach, to ensure the development of solid synergies among the various project tasks, common working procedures and mitigating risks related to critical project activities.
In WP1, communication channels, workflows and qualitative procedures were enhanced and sustained to cover administrative and technical aspects at the work package and project levels.
In WP2, partners finalized the user requirements and system requirements for the final INCISIVE architecture. Finally, activities were undertaken to improve the initial prototype and User Interface from the first period.
In WP3, data management and integration of data sources were prioritized. Technical requirements and business processes for the INCISIVE platform were defined by improving and adapting the infrastructure with DevOps and software integration. Software components from WP3, WP4 and WP5 were integrated and efforts were made to integrate all Data Providers’ machines into the infrastructure. Finally, AI architectural elements were designed and deployed for Federated Learning.
In WP4, cancer-related AI services were further developed at the AI model and software level. Platform's visual presentation was updated and improved. Further progress has been achieved on the development of the AI services by implementing new AI models utilizing open and INCISIVE data.
WP5, focused on completing the temporary infrastructure provided by FTSS and transferring all retrospective data to the main INCISIVE infrastructure prototype 2 (Hybrid infrastructure). Additionally, work focused on the development of the INCISIVE hybrid infrastructure including the set-up of the ‘permanent’ central repository hosted by MAG, the implementation of the search engine and the development a data sharing mechanism that informs and enables new Data Providers to share their data within the INCISIVE repository and authorized data users to search and reuse data shared in the INCISIVE repository in compliance with GDPR.
WP6 focused on the integration of INCISIVE’s Second Prototype, the training and validation activities of the main project results. The platform and infrastructure evolved to second version with a hybrid approach. Central and all the Federated Nodes were integrated. Data from the temporary infrastructure were migrated to the Central Node. Relevant tools/software necessary for the functionalities provided by the Central Node were deployed with CI/CD and GitOps automation. Pre-validation studies on data and image quality were conducted. Preparation for observational studies in multiple pilot sites began.
WP7 focused on activities to ensure legal and ethical compliance of the ongoing tasks. An amendment was made to the legal framework that was established during the initial period due to the transfer of data to the hybrid storage infrastructure from the temporary infrastructure. Amendment 2 to the data sharing agreement was prepared and signed by the partners. A description of security measures for the data storage and guidance for the hosting nodes, were circulated to relevant Data Providers. A second iteration of the Data Protection Impact Assessment was conducted and the recommendations, relating to hybrid and federated infrastructure, were shared. A legal framework for data donation was developed to enable external parties to contribute data to the INCISIVE repository. The INCISIVE data sharing model was named ‘Registered Access Plus’.
WP8 focused on the continuous validation and preliminary updating of the business plan, the development of a preliminary operational and deployment plan to support the launch and long-term sustainability of the image repository and updating the project’s Innovation Strategy for ensuring long-term exploitation and the sustainability of all outputs beyond the lifespan of the project.
WP9 focused on a wide range of communication and dissemination activities to target audiences the results and the advances of INCISIVE. Future work will continue all dissemination and communication work already undertaken with a special focus on the final INCISIVE prototype.
Major emphasis was placed by the PM on establishing a fast start approach, to ensure the development of solid synergies among the various project tasks, common working procedures and mitigating risks related to critical project activities.
In WP1, communication channels, workflows and qualitative procedures were enhanced and sustained to cover administrative and technical aspects at the work package and project levels.
In WP2, partners finalized the user requirements and system requirements for the final INCISIVE architecture. Finally, activities were undertaken to improve the initial prototype and User Interface from the first period.
In WP3, data management and integration of data sources were prioritized. Technical requirements and business processes for the INCISIVE platform were defined by improving and adapting the infrastructure with DevOps and software integration. Software components from WP3, WP4 and WP5 were integrated and efforts were made to integrate all Data Providers’ machines into the infrastructure. Finally, AI architectural elements were designed and deployed for Federated Learning.
In WP4, cancer-related AI services were further developed at the AI model and software level. Platform's visual presentation was updated and improved. Further progress has been achieved on the development of the AI services by implementing new AI models utilizing open and INCISIVE data.
WP5, focused on completing the temporary infrastructure provided by FTSS and transferring all retrospective data to the main INCISIVE infrastructure prototype 2 (Hybrid infrastructure). Additionally, work focused on the development of the INCISIVE hybrid infrastructure including the set-up of the ‘permanent’ central repository hosted by MAG, the implementation of the search engine and the development a data sharing mechanism that informs and enables new Data Providers to share their data within the INCISIVE repository and authorized data users to search and reuse data shared in the INCISIVE repository in compliance with GDPR.
WP6 focused on the integration of INCISIVE’s Second Prototype, the training and validation activities of the main project results. The platform and infrastructure evolved to second version with a hybrid approach. Central and all the Federated Nodes were integrated. Data from the temporary infrastructure were migrated to the Central Node. Relevant tools/software necessary for the functionalities provided by the Central Node were deployed with CI/CD and GitOps automation. Pre-validation studies on data and image quality were conducted. Preparation for observational studies in multiple pilot sites began.
WP7 focused on activities to ensure legal and ethical compliance of the ongoing tasks. An amendment was made to the legal framework that was established during the initial period due to the transfer of data to the hybrid storage infrastructure from the temporary infrastructure. Amendment 2 to the data sharing agreement was prepared and signed by the partners. A description of security measures for the data storage and guidance for the hosting nodes, were circulated to relevant Data Providers. A second iteration of the Data Protection Impact Assessment was conducted and the recommendations, relating to hybrid and federated infrastructure, were shared. A legal framework for data donation was developed to enable external parties to contribute data to the INCISIVE repository. The INCISIVE data sharing model was named ‘Registered Access Plus’.
WP8 focused on the continuous validation and preliminary updating of the business plan, the development of a preliminary operational and deployment plan to support the launch and long-term sustainability of the image repository and updating the project’s Innovation Strategy for ensuring long-term exploitation and the sustainability of all outputs beyond the lifespan of the project.
WP9 focused on a wide range of communication and dissemination activities to target audiences the results and the advances of INCISIVE. Future work will continue all dissemination and communication work already undertaken with a special focus on the final INCISIVE prototype.
The below results of the period span across the entire spectrum of technical development actions foreseen in the project. They form the solid basis on which the final project period will consolidate the final two main project results: INCISIVE imaging and other health data repository and the AI toolbox.
1. The delivery of the second update of the full system design, addressing the users’ requirements.
2. Τhe delivery of the second version of the AI services covering all 4 cancer types addressed by the project and combining several AI models per AI service. The AI toolbox and its services involve the implementation of segmentation/classification algorithms, prognostic models and image-to-report transformation, explainable AI features, clinical concept data pipelines, the implemented approach on federated learning and the definition/implementation of presentation features (GUI and AR-enhanced visualizations). All INCISIVE AI services may deliver significant assistance to medical professionals, thus having a positive impact on the clinical workflow.
3. The implementation and release of the 2nd INCISIVE integrated platform, building on the first prototype and adding all planned functionalities based on the INCISIVE platform architecture, including the public part of the platform with the information on the INCISIVE data sharing mechanism and platform functionalities requiring prior registration. The latter covers both AI training-related functionalities and the delivery of AI services to Health care professionals. A major part of the work was devoted to transforming the federated first prototype into a hybrid (federated/centralised) platform by integrating MAG's central data node.
4. The delivery of a GDPR-compliant data sharing mechanism, which enables the new data providers to share their data within the INCISIVE repository and allows data users to search and reuse INCISIVE data.
1. The delivery of the second update of the full system design, addressing the users’ requirements.
2. Τhe delivery of the second version of the AI services covering all 4 cancer types addressed by the project and combining several AI models per AI service. The AI toolbox and its services involve the implementation of segmentation/classification algorithms, prognostic models and image-to-report transformation, explainable AI features, clinical concept data pipelines, the implemented approach on federated learning and the definition/implementation of presentation features (GUI and AR-enhanced visualizations). All INCISIVE AI services may deliver significant assistance to medical professionals, thus having a positive impact on the clinical workflow.
3. The implementation and release of the 2nd INCISIVE integrated platform, building on the first prototype and adding all planned functionalities based on the INCISIVE platform architecture, including the public part of the platform with the information on the INCISIVE data sharing mechanism and platform functionalities requiring prior registration. The latter covers both AI training-related functionalities and the delivery of AI services to Health care professionals. A major part of the work was devoted to transforming the federated first prototype into a hybrid (federated/centralised) platform by integrating MAG's central data node.
4. The delivery of a GDPR-compliant data sharing mechanism, which enables the new data providers to share their data within the INCISIVE repository and allows data users to search and reuse INCISIVE data.