MTA Portal

Connecting different Picture Archiving and Communication Systems (PACS) remains one of the biggest challenges in medical informatics. In academic hospitals, where research and clinical care converge, seamless data exchange is essential for validating new imaging techniques.

The MTA Portal Central standard defines a layered architecture, based on HL7 FHIR and DICOMweb, which creates an abstraction layer above existing PACS infrastructures. This layer translates vendor-specific protocols into a uniform data model, enabling quantitative analyses of MRI and CT images regardless of the source system.

Figure 1: Conceptual model of the interoperability layer.

Core Components of the Standard

Our approach rests on three pillars:

Metadata Harmonization: Automatic mapping of DICOM headers to a standardized schema, including patient anonymization for research.
Query Federation: A uniform API for searching multiple PACS archives simultaneously, based on clinical parameters such as scan protocol or contrast agent.
Performance Metrics Pipeline: An automated workflow for extracting, transforming, and loading (ETL) image quality indicators, such as signal-to-noise ratio and spatial resolution.

A practical case at the AMC showed that this architecture reduced the time for data extraction for a multicenter MRI study from weeks to hours. The quantitative validation of a new diffusion tensor imaging (DTI) sequence could thereby be accelerated.

Future Direction: AI-Ready Datasets

The next evolution focuses on generating AI-ready datasets. By combining interoperability with standardized pre-processing steps (such as reorientation and intensity normalization), we create a foundation for machine learning models that can directly predict clinical effectiveness.

The sterile, data-driven approach of MTA Portal Central ensures that validation studies are conducted in a reproducible, scalable, and vendor lock-in free manner.