AI Model Portability & DEP Reusability

This use case addresses the challenge of transferring and storing large medical image files in Data Exploitation Platforms (DEPs). It explores the use of advanced compression methods—specifically JPEG 2000 Part 2 in DICOM format—to reduce file size without losing critical diagnostic detail. The goal is to improve storage efficiency, transfer speed, and integration with AI workflows. The work is led by MUG and COM, and benefits image-heavy Research Infrastructures like BBMRI and Euro-BioImaging.

Challenge

As imaging technologies advance, research infrastructures (RIs) are generating increasingly large and complex image datasets. Transferring these large files from data holdings to Data Exploitation Platforms (DEPs), and storing them even temporarily, presents significant logistical and financial challenges.

Approach

Assess JPEG 2000 Part 2 compression embedded in DICOM files
Measure encoding/decoding times and compatibility with AI pipelines (training & inference)
Evaluate image quality through expert reviews by pathologists
Analyse the potential for integration into clinical/research workflows and long-term archiving systems

Goal

To evaluate advanced image compression methods that reduce file size without sacrificing detail, enabling more efficient storage, faster data transfer, and improved accessibility in DEPs.

Relevance

This use case is particularly relevant for Research Infrastructures that manage or support image-intensive research—such as those in biomedical and clinical domains. For example, BBMRI and Euro-BioImaging generate high-resolution images that are critical for diagnostics, research, and AI training but are often too large for efficient processing and sharing. These RIs need scalable solutions to reduce storage and transfer burdens while preserving data quality.

Expected Impact

This use case will demonstrate how advanced compression can improve performance and reduce costs across the image data lifecycle—from scanner to analysis—especially for Research Infrastructures aiming to scale up AI-driven research while maintaining data quality and accessibility.

Partners

Other Technological Use Cases

DEP Scalability on EuroHPC with DestinE

This use case tests the scalability and performance of RI-SCALE’s Data Exploitation Platform (DEP) on...

Green Computing Improvement

This use case evaluates the potential of GROQ’s high-performance, energy-efficient hardware to reduce the environmental...

Credit Management System

This use case develops a Credit Management System for Data Exploitation Platforms (DEPs) that tracks...