Periodic Reporting for period 1 - BIGPICTURE (Central Repository for Digital Pathology)
Berichtszeitraum: 2021-02-01 bis 2022-07-31
Artificial intelligence (AI) has the potential to enhance speed, precision, quality and effectiveness of many processes across medical fields, especially in pathology. Traditionally, pathologists diagnose and understand disease by examining tissues under a microscope or using digitised microscopy images. To leverage AI in pathology, AI algorithms need to be developed by training in recognizing patterns on a large variety of images. Bigpicture's mission is to collect and share an unprecedented amount of quality-controlled pathology images, metadata and AI algorithms, in order to facilitate such development and research. (https://www.linkedin.com/company/bigpicture-project/)
Building on existing assets such as ELIXIR infrastructure, Bigpicture establishes the first European ethical and regulatory compliant platform connecting pathologists, researchers, AI developers, patients, and industry. Our vision is to become a catalyst in the digital transformation of Pathology, allowing AI to reach maturity in Pathology: Enabling the development and validation of trustworthy technologies to help diagnose and predict a wide range of diseases and improve the quality and efficiency of toxicity studies, in turn boosting drug development for the benefit of patients. By engaging and building stakeholder consensus, Bigpicture will contribute to a regulatory framework for digital pathology and AI-based methods. Finally, Bigpicture envisions sustainability of its platform through a community-based model that creates value and reciprocity in use.
Building on existing assets such as ELIXIR infrastructure, Bigpicture establishes the first European ethical and regulatory compliant platform connecting pathologists, researchers, AI developers, patients, and industry. Our vision is to become a catalyst in the digital transformation of Pathology, allowing AI to reach maturity in Pathology: Enabling the development and validation of trustworthy technologies to help diagnose and predict a wide range of diseases and improve the quality and efficiency of toxicity studies, in turn boosting drug development for the benefit of patients. By engaging and building stakeholder consensus, Bigpicture will contribute to a regulatory framework for digital pathology and AI-based methods. Finally, Bigpicture envisions sustainability of its platform through a community-based model that creates value and reciprocity in use.
Workpackage (WP) 1 implemented a management structure to oversee and monitor the project. WP1 completed anticipated deliverables and milestones, including the Data Management Plan & ethics reports. WP1 ensures the build-up of the Bigpicture community with an online working environment, regular project meetings, monthly communications, a newsletter, a website and social media. Active participation at the 2 consortium meetings demonstrated collective engagement. WP1 started to reach out to external stakeholders, including Slide-Contributing Third Parties. Advisory boards are implemented.
WP2 gathered initial infrastructure requirements and established the development platform to support iterative development of the technical infrastructure. A pilot system was set up based on the requirements and basic features are implemented. Finally, the security of the pilot system was reviewed and penetration testing was performed. Pilot operations and iterative development of the infrastructure and services are ongoing. Notably, WP2 coordinated the development of the Enterprise Architecture describing all functions and processes in the platform.
WP3 set up the framework for coordination and interaction between and within the data collector nodes. The first datasets have been selected, extraction pipelines have been established at three different collection sites, and data from these sites was submitted to the repository. The metadata schema was developed, which will be further extended within the coming period. The honest broker requirements are gathered and the system was piloted and integrated in the federated authentication and authorisation system.
WP4 completed the collection of user requirements for the front-end (Cytomine). The Cytomine demo server was implemented, for use and preliminary integration with Bigpicture backend infrastructure through containerization. WP4 developed tools for standardization of WSI formats of varying vendors according to the DICOM standard. Initial metadata validation tools have been developed to validate that metadata comply with Bigpicture requirements. WP4 also developed a weakly-supervised learning methodology (Streaming CLAM), that allows end-to-end training with WSI and slide- or patient-level annotations.
WP5 provided an overview of the existing clinical digital pathology legal frameworks, reporting on the landscape of guidelines, standards and regulatory requirements for non-clinical digital pathology, and creating a pseudonymisation strategy. The Data Privacy Impact Report and the report on the technical and organisational measures are being finalized. Input from the Ethics Advisory Board (EAB) was collected in an EAB meeting and written recommendations were provided. WP5 is leading the GDPR and Quality Coordination Center taskforces.
WP6 started to mobilize and engage key stakeholders through leveraging the diverse networks of Bigpicture partners. A stakeholder analysis was initiated. WP6 initiated exploring organisational and governance models. A first set of indicators have been tabulated in the dynamic KPI reporting document for the Bigpicture platform, leveraging on the ELIXIR Europe expertise and serving as a basis for quality management of the platform.
WP2 gathered initial infrastructure requirements and established the development platform to support iterative development of the technical infrastructure. A pilot system was set up based on the requirements and basic features are implemented. Finally, the security of the pilot system was reviewed and penetration testing was performed. Pilot operations and iterative development of the infrastructure and services are ongoing. Notably, WP2 coordinated the development of the Enterprise Architecture describing all functions and processes in the platform.
WP3 set up the framework for coordination and interaction between and within the data collector nodes. The first datasets have been selected, extraction pipelines have been established at three different collection sites, and data from these sites was submitted to the repository. The metadata schema was developed, which will be further extended within the coming period. The honest broker requirements are gathered and the system was piloted and integrated in the federated authentication and authorisation system.
WP4 completed the collection of user requirements for the front-end (Cytomine). The Cytomine demo server was implemented, for use and preliminary integration with Bigpicture backend infrastructure through containerization. WP4 developed tools for standardization of WSI formats of varying vendors according to the DICOM standard. Initial metadata validation tools have been developed to validate that metadata comply with Bigpicture requirements. WP4 also developed a weakly-supervised learning methodology (Streaming CLAM), that allows end-to-end training with WSI and slide- or patient-level annotations.
WP5 provided an overview of the existing clinical digital pathology legal frameworks, reporting on the landscape of guidelines, standards and regulatory requirements for non-clinical digital pathology, and creating a pseudonymisation strategy. The Data Privacy Impact Report and the report on the technical and organisational measures are being finalized. Input from the Ethics Advisory Board (EAB) was collected in an EAB meeting and written recommendations were provided. WP5 is leading the GDPR and Quality Coordination Center taskforces.
WP6 started to mobilize and engage key stakeholders through leveraging the diverse networks of Bigpicture partners. A stakeholder analysis was initiated. WP6 initiated exploring organisational and governance models. A first set of indicators have been tabulated in the dynamic KPI reporting document for the Bigpicture platform, leveraging on the ELIXIR Europe expertise and serving as a basis for quality management of the platform.
Setting up a resource for AI development in Pathology of the size of Bigpicture is unprecedented, and already attracts large international attention. Both the extent of the project (numbers of partners, disease areas, images, etc) and the fact that this project combines data from the clinical and preclinical field, offer new perspectives. Despite the significant challenges foreseen at project's initiation, Bigpicture is on track, meeting project plans and exceeding expectations beyond the state-of-the-art in a least two areas, addressed below.
A limiting factor to progress in computational pathology is the lack of standards, e.g. image standards. As a result, many different digital image formats exist, which limit interoperability and, eventually, integration of AI tools in the pathologists’ workflow. Bigpicture developed and released software tools for easy conversion of a range of existing file formats to the DICOM standard (https://github.com/imi-bigpicture). Bigpicture promotes the adoption and use of DICOM by storing and releasing all collected images in this format. This will set a standard to the field that will benefit both producers and users of AI in pathology; it will markedly increase the opportunities for SMEs as vendor lock-ins are diminished.
Next, the EU landscape of data sharing regulations is variable and complex. This leads to an overly protective attitude, limiting availability of data. We produced a public report on GDPR interpretation and law for 8 EU countries (https://cordis.europa.eu/project/id/945358/results). This will be at the basis of setting up local procedures for GDPR-compliant data sharing, resulting in an increased availability of data for research, while still adhering to all applicable laws and regulations. This, in turn, will contribute to trust in the Bigpicture platform.
The prospect of getting access to large, high-quality datasets has created the interest of commercial parties not yet formally connected to Bigpicture. With the Bigpicture beneficiaries, they will be instrumental in translating the data and models into products/services creating tangible value in research and health care. We are setting up ways to keep these parties connected via open webinars and conferences. Pharmaceutical companies (beneficiaries as others) are also studying unforeseen usage of Bigpicture data, further explored by interviewing stakeholders to refine business models. Last, Bigpicture opened discussions with regulators (FDA/EMA) on the use of digitized slides in the regulated nonclinical environment, leading to possible future reduction of control animals in preclinical toxicity studies.
A limiting factor to progress in computational pathology is the lack of standards, e.g. image standards. As a result, many different digital image formats exist, which limit interoperability and, eventually, integration of AI tools in the pathologists’ workflow. Bigpicture developed and released software tools for easy conversion of a range of existing file formats to the DICOM standard (https://github.com/imi-bigpicture). Bigpicture promotes the adoption and use of DICOM by storing and releasing all collected images in this format. This will set a standard to the field that will benefit both producers and users of AI in pathology; it will markedly increase the opportunities for SMEs as vendor lock-ins are diminished.
Next, the EU landscape of data sharing regulations is variable and complex. This leads to an overly protective attitude, limiting availability of data. We produced a public report on GDPR interpretation and law for 8 EU countries (https://cordis.europa.eu/project/id/945358/results). This will be at the basis of setting up local procedures for GDPR-compliant data sharing, resulting in an increased availability of data for research, while still adhering to all applicable laws and regulations. This, in turn, will contribute to trust in the Bigpicture platform.
The prospect of getting access to large, high-quality datasets has created the interest of commercial parties not yet formally connected to Bigpicture. With the Bigpicture beneficiaries, they will be instrumental in translating the data and models into products/services creating tangible value in research and health care. We are setting up ways to keep these parties connected via open webinars and conferences. Pharmaceutical companies (beneficiaries as others) are also studying unforeseen usage of Bigpicture data, further explored by interviewing stakeholders to refine business models. Last, Bigpicture opened discussions with regulators (FDA/EMA) on the use of digitized slides in the regulated nonclinical environment, leading to possible future reduction of control animals in preclinical toxicity studies.