Project no.: 2022/347151
Programme name: SMEs Growth Romania – Boost to Romanian Business sector: Business Growth in Start-ups – 2nd call
Project name: CorFlow API Exposure – AI based Computer Aided Medical System Capabilities Exposure with API Service
Project Promoter: SC AUTOSYMED SRL
CorFlow API Exposure
April 2023
We are pleased to inform you that we are actively developing the advanced research and development project CorFlow API Exposure.
Our goal is to create reference databases of medical systems using APIServices supported by artificial intelligence. The project’s vision is to improve cardiac care to achieve its highest efficiency continuously.
It is our priority to achieve the following:
– diagnostic accuracy,
– personalized treatment,
– the highest quality tools based on artificial intelligence.
All this is in cooperation with medical experts and based on constant research.
The CorFlow API Exposure system provides scalable and integrated AI tools to support the most complex cardiac use cases, ensuring efficiency in diagnosing and treating patients with different needs.
The artificial intelligence-based system automatically analyzes the coronarographic image data, detects and characterizes the stenosis, and automatically calculates the SyntaxScore, saving the time and effort of medical staff.
Function APIs enable the integration of Corflow AI services into client applications and many other uses for R&D institutions, universities, and researchers looking for robust datasets for training machine learning models and developing algorithms.
Development phase in the project - CorFlow API Exposure
September 2023
Our team is currently finalizing the development phase of the project with some key deliverables:
– Capabilities Open API services to integrate with “client” applications which will realize more complex cardiological use cases based on Computer-Aided Diagnosis.
– Service API definition to support REST-based operations on DICOM repositories to automatically detect coronary artery stenosis and its quality monitoring and control.
– Service API for refinement of frame selection algorithm, stenosis detection and algorithm characterization, and 3D reconstruction algorithm.
– Acquire advanced storage and firewall solutions to deliver on-premise cloud solutions for hosting DICOM images required for AI algorithm.
– Anonymized DICOM images with cardiac treatment data to enable AI algorithm quality improvements for automated SyntaxScore calculation and reference database services APIs.
Development phase in the project - CorFlow API Exposure
December 2023
Our team is currently finalizing the development phase of the project with some key deliverables:
- Capabilities Open API services to integrate with “client” applications to realize more complex cardiological use cases based on Computer-Aided Diagnosis.
- Service API for different queries and interconnections to enable database interaction for client applications based on specified metamodel schemes
- Integrated service API with flexible parametrization of feedback and review by MD with interface supporting study, series, frame selections, and justifications.
- Platform cloud-native components for Kubernetes to integrate storage drivers to provide storage vendor integration through the K8S operator framework. Monitoring dashboards for storage services configured. The solution is ready for verification.
- Developed API services can be used by research and development institutions, universities, and researchers seeking robust datasets for ML model training and algorithm development, who are encouraged to collaborate with CorFlow.
Hospitals, both public and private, and other medical units have the opportunity to integrate CorFlow into their workflow for real-time coronarography analysis. Our cloud-based teleconsultation platform facilitates collaboration among heart teams, enabling the secure sharing of medical data for consultations. Autosymed CorFlow’s commitment to data sharing contributes to advancing science and cardiac care. By collaborating and sharing knowledge, the medical community can collectively elevate the standard of cardiac medicine, leading to better patient outcomes and improved healthcare practices.
- CorFlow can be used to create ground truth datasets. In other medical applications by automation of data acquisition, anonymization, curation and validation, data storage, monitoring, metadata management, and facilitation of manual and semi-automatic data labeling.
- AI/ML/DL facilitation. We offer the opportunity to validate algorithms on our reference database and infrastructure. Additionally, CorFlow can be leveraged on external datasets, enabling comparisons of algorithm performance.
We are constantly developing and improving the Corflow application with capabilities APO.
Finalization of development in the project - CorFlow API Exposure
March 2024
The Autosymed team is happy to announce the finalization of the CorFlow API Exposure project. The goal of R&D activities was to enhance the Corflow solution with the Capabilities of Open API services to integrate with “client” applications to realize more complex cardiological use cases based on Computer-Aided Diagnosis. Our team developed Service APIs to be used by research and development institutions, universities, and researchers seeking robust datasets for ML model training and algorithm development, who are encouraged to collaborate with CorFlow.
Service endpoints for DICOM artifacts
Service API for different queries and interconnections on DICOM repositories, based on a specified metamodel scheme, to enable CRUD database interaction for client applications.
Service endpoints for feedback operations
Service API to facilitate CRUD operations on data related to automatic corrections provided to AI algorithms deployed on specific images. Feedback for selected series and frames regarding segmentation, branch identification, stenosis detection and characterization, SyntaxScore calculation, and justifications for any changes. Apart from corrections provided by MD in a dedicated format, feedback data is based on metadata such as study details, primary/secondary angle, date of acquisition, data of feedback generation, author, etc.
Service endpoints for AI algorithms
Dedicated API facilitates CRUD operations on data related to AI algorithms’ results. Service API with dedicated user interface for Syntax Scores calculation with integrated AI algorithms for vessel segmentation, branch identification, stenosis detection and characterization, template-based reporting, and 3D data visualization. Database failover system to enable reliable hosting and system deployment.
Service endpoints for the Feedback loop to support automatic integration of corrections provided to AI algorithms’ results
MD provides a feedback loop for manual corrections through a dedicated user interface for AI algorithms’ results (frame and series selection, segmentation, branch identification, stenosis detection, and characterization). This loop enables automatic model feeding with provided corrections and training results enhancements so that AI algorithms’ results successively improve.
Diagnosis and Treatment decision support system with Corflow Rich-UI
Service API to enable diagnosis and treatment decision support based on SyntaxScore through an advanced and flexible customized interface using a shareable database or integrated customers’ database. Rich UI features cover data selection and integration, 3d data visualization, SyntaxScore calculation, and customized diagnostic report generation.
Hospitals, both public and private, and other medical units have the opportunity to integrate CorFlow into their workflow for real-time coronography analysis. Our cloud-based teleconsultation platform facilitates collaboration among heart teams, enabling the secure sharing of medical data for consultations. Autosymed CorFlow’s commitment to data sharing contributes to advancing science and cardiac care. By collaborating and sharing knowledge, the medical community can collectively elevate the standard of cardiac medicine, leading to better patient outcomes and improved healthcare practices.
