Clinical Imaging Consortium

Improving AI solutions based on clinical images without compromising privacy and security
CLIC Press Release August 16 2022
Download (.pdf)

About the CLIC-Project

The shared vision of the CLIC Project is to accelerate the development and implementation of image-based AI solutions in the Danish health care sector to the benefit of patients.

The purpose of CLIC is to develop a model for a One Stop Shop data platform with an agile infrastructure.

The One Stop Shop will facilitate collaboration between private companies offering AI solutions and the public health care system that wants to integrate the AI solutions into the clinical workflow.

This platform ensures secure access to data and services which supports development, validation and implementation of image-based AI solutions.

This way CLIC will help demonstrating a much smoother and transparent way to validate and develop image-based AI-solutions in clinical practices.


Data Lake
Developing an infrastructure for data lakes that can be accessed by public and private companies when developing and validating AI solutions.
Compliance
Building a platform that documents compliance with all relevant standards and regulatory requirements for imaging based AI-Solutions.
Support Services
Developing support services related to AI-development and usage of the data lakes- This includes framework agreements for legal access to the data lakes as well as for the usages of the CLIC support services.

Benefits to patients

CLIC facilitates increased usage of image based AI solutions that can enhance the quality and precision of diagnostics for the patients. Furthermore with the use of decision support the clinical processes and the patient journey can be optimized and valuable highly skilled professionals can be released to more complex tasks. AI solutions thus can help to increase the efficiency in health care provision.

Benefits to health care system

There are approximately 600 radiologists in Denmark. If the workload of radiologists dedicated to processing clinical images can be reduced by five hours every week this would be the equivalent of gaining 71 extra full-time radiologists in Denmark.

What is image based AI?

Using artificial intelligens for analyzing image data, can bring us from vast amounts of data to valid conclusions faster than ever before. To get started, the data needs to be collected and annotated, and an AI model needs to be set up, trained and deployed.
Swipe sideways to see all
Illustration explaining AI

Use case
To ensure that the data platform in CLIC is driven by actual needs 4 use cases are included in the project. Each use case represents a public-private AI-collaboration at different stages. This way, the use cases reflects the primary functionalities, services and administrative processes that the project aspire to develop, streamline and implement with the CLIC-platform.
‍

Detection of Fractures based on X-rays

Project partner Radiobotics is developing proprietary algorithms for automated analysis of plain X-rays / radiographs.
‍
Radiobotics brings in their CE-marked fracture detection product ‘RBfracture', that serves as an immediate second opinion for clinicians assessing trauma X-rays. The aim of the project is to set up a monitoring process for RBfracture after clinical implementation and ensure continuous improvements to the algorithm through quality feedback on local data.

Detection of brain strokes, brain tumors, and hemorrhages through MRI diagnostics

Apollo is a fully automated AI software solution developed by Cerebriu. It detects significant brain conditions incl. strokes, brain tumors, and hemorrhages in real-time with 95%+ sensitivity during MRI diagnostic examination.

Apollo reduces the burden on radiologists by automating MRI workflows by cutting short examinations and providing notification of acute findings for patient triage and visual localization of detected findings.

Decision support for treatment of a Macula Degeneration

Age related macular degeneration (AMD) is a chronic retinal disease. It affects the central vision and is the most common cause of blindness in Denmark.

Enversion has been part of developing an AI algorithm which can determine which patients should be treated and which patients should be observed in the clinic. The algorithm can enable non-medical staff to diagnose AMD and thus free up important ophthalmologist resources.

AI for Decision Support in Alzheimer Clinical Trials

Patients enrolled in clinical trials for medicine to treat Alzheimer's are scanned up to three times a year. The scan determines reduction in brain volume - is an indicator of drug efficacy.

The project will explore how the CE marked AI-Rad Companion Brain MR solution which is developed by Siemens Healthineers can be used in clinical trials for a new Alzheimer’s treatment developed by Roche.

The solutions can potentially reduce the need for manual analysis by neurologists.

Objectives

Map the data landscape for image data and other health data in Denmark that potentially will create a basis for the data lakes required for developing, validating and implementing AI solutions.
Clarify regulatory requirements for building and accessing such data lakes hosted by Danish regions and develop and implement an action plan for streamlining the process for obtaining regulatory approval.
Build a model for a One Stop Shop offering access to data-lakes and to services that will enable shorter time to market for image based AI solutions through more effective development and validation of AI solutions in a public private partnership.
Deploy an AI data model by building individual data lakes in collaboration with Region H.
Integrate the CLIC One Stop Shop with the One Stop Shop in the OSCAR organization, thus realising important synergies and ensuring the political, financial and data-ethical sustainability of the CLIC project upon completion.
Test the value creation potential of the One Stop Shop against a range of different image based AI solutions currently under development and/or implementation.
Share the results of the CLIC project in order to encourage the implementation of similar One Stop Shops in all Danish Regions.

Services

During the project the partners involved in the CLIC project will develop and test solutions and services that supports AI development. The services will be developed throughout the project but aims at making a number of the following solutions available: 
Data access, sharing and processing
Access to framework agreements for server hosting and data processing
Access to data in individual data lake
Up-load of own data to individual data lakes
Support to development of algorithms
Access to data lake for algorithm development
Manual and automated annotation and labelling services
Validation and integration
Access to clinicians for validation of algorithms
Access to hospitals for integration with clinical workflow
Collaboration with  clinicians for publication
Health Technology Assessment
Regulatory compliance
Documentation of compliance with GDPR
Documentation of compliance with European Medical Device Directive
Documentation of compliance with European Tender Regulation
Documentation of compliance with Data Ethical Rules and Standards

Organization

Radiology AI Test Center
Project Administrator and responsible for securing data processing and hosting agreements in collaboration with DataFair and for building data lake in collaboration with ENV, TEAL, and DF.
DataFair
Overall project managementer of the CLIC project and resposible for securing data processing and hosting agreements in close collaboration with RAIT; for developing annotation service, and data lake access user interface and for developing and implementing Communication Plan.
Enversion
Development, deployment and operation of algorithms for structuring and pseudonymisation of health data; building AI data lake in collaboration with TEAL and RAIT and specification for, and validation of, One Stop Shop for late stage AI development case.
Teal Medical
Pseudonymisation and retrieval of image data and for building data lake in collaboration with ENV, RAIT, and DF.
Siemens Healthineers
Use case on AI Decision Support. Specification for and test and validation of data lake and services.
Radiobotics
Specification for and test and validation of data lake and services.
Cerebriu
Specification for and test and validation of data lake and services 2021.
2021.ai
Adaptation of GRACE solution as documentation and compliance tool for Health AI development.
Roche
Input to the development of algorithms for detecting alzheimer’s based on brain scans. Specification for and test and validation of data lake and services.

Contact

PROJECT MANAGER
Troels Bierman Mortensen
tbm@datafair.org+45 31 55 10 15

‍

DataFair ApS