MIRACLE: An Open-Access API for Evidence-Based Reference Values in Cardiovascular MRI

Presenting Author(s)

• 

  Ankush Ankush, MD

  Cardiothoracic Imaging Fellow
  Atrium Wake Forest Baptist Medical Center
  Winston Salem, North Carolina, United States

Primary Author(s)

• 

  Ankush Ankush, MD

  Cardiothoracic Imaging Fellow
  Atrium Wake Forest Baptist Medical Center
  Winston Salem, North Carolina, United States

Co-Author(s)

• JP

  Janardhana Ponnatapura, MD

  Section Head
  Atrium Health Wake Forest Baptist Medical Center, United States
• MS

  Muthu Sakthivel, MD, FSCMR

  Assistant Professor/Cardiothoracic Imaging Fellowship Program Director
  UNC School of Medicine
  Chapel Hill, North Carolina, United States
• ES

  Eswara Sellamuthu, MD

  Assistant Professor
  Atrium Health Wake Forest Baptist Medical Center, United States

Description:

Introduction: The interpretation of cardiovascular magnetic resonance (CMR) measurements requires comparison to normative data, a process that is often manual, time-consuming, and prone to error. This can be a barrier to efficient and standardized clinical reporting and research. To address this gap, we have developed MIRACLE (MR Imaging Reference API for Cardiovascular Limits from Evidence), an open-access, RESTful API that provides immediate access to evidence-based reference values for a wide range of CMR parameters.

Methods/Features: MIRACLE is an open-access API requiring no authentication, built on a collection of peer-reviewed publications. It provides reference data across multiple clinical domains, including pediatric, adult, and athletic populations. Users can submit a patient's demographic and anthropometric data along with a measured CMR parameter via a simple API call and receive a JSON response containing the calculated z-score, percentile, and corresponding normal range.

Applications and Innovation: MIRACLE is designed for seamless integration into clinical software, research workflows, and Picture Archiving and Communication Systems (PACS). A key innovation is its utility for Large Language Models (LLMs), enabling the development of next-generation, AI-powered diagnostic assistants for CMR. To demonstrate this, we have developed a web-based Pediatric CMR calculator that generates a print-ready report and a prototype LLM-based chatbot that leverages the function-calling capabilities of modern LLMs. Furthermore, MIRACLE provides Model-centric Protocol (MCP) server functionality for AI agents to enable direct API access, documentation search, and code generation assistance.

Conclusion: By providing a standardized, programmatic, and open-source method for accessing CMR reference data, MIRACLE has the potential to enhance the efficiency and accuracy of CMR interpretation. The future development roadmap includes a dedicated Python package (miracleapi-py) to facilitate easy batch processing and statistical analysis. We invite the SCMR community to use and contribute to this open-source project.

Access Instructions:

• API Documentation: https://miracleapi.readme.io/
  Comprehensive documentation for the MIRACLE API, including all endpoints and usage examples.
  
  


• Main GitHub Repository: https://github.com/drankush/MIRACLE-API
  Source code and information for the core MIRACLE API project.
  
  


• Web App Demo: https://miracle-app.surge.sh/
  A live, web-based Pediatric CMR reference calculator that generates a print-ready report using the API.
  
  


• Web App Source Code: https://github.com/drankush/MIRACLE-webapp
  The GitHub repository containing the source code for the demonstrative web application.
  
  


• Chatbot Demo: https://miracle-chat.netlify.app/
  A live, prototype LLM-based chatbot demonstrating the API's utility with AI agents.
  
  


• Chatbot Source Code: https://github.com/drankush/MIRACLE-ChatBot
  The GitHub repository containing the source code for the chatbot application.
  
  

Figure 1: Flow diagram showing complete data to application pipeline with MIRACLE API as the core bridge.
Figure 2: Chatbot built by integrating the functional tool-calling capabilities of LLaMA 3.3 70B with the MIRACLE API. Demo: https://miracle-chat.netlify.app Source Code: https://github.com/drankush/MIRACLE-ChatBot. Inference Provider: Groq
Figure 3: Pediatric CMR Reference Calculator built using MIRACLE API. Demo: https://miracle-app.surge.sh Source Code: https://github.com/drankush/MIRACLE-webapp