From scanner output to structured pathology data.
DPARS is a modular orchestration platform designed to connect heterogeneous whole-slide imaging workflows with metadata intelligence, quality control, standards-based conversion, delivery and retention-aware archival.
A modular journey for every whole-slide image.
DPARS separates responsibilities into observable services so each step can evolve independently while preserving provenance across the workflow.
The complete DPARS orchestration layer.
Each service owns a focused responsibility. Dagster coordinates execution while PostgreSQL provides the persistent lineage, status and audit backbone across the system.
A workflow you can reason about.
↳ identity established
↳ metadata resolved
↳ QC gate evaluated
↳ derivative lineage recorded
↳ downstream state persisted
Specialized services. Shared lineage.
Metadata intelligence
Multi-route metadata extraction combines fast paths with deeper barcode and OCR strategies when needed.
BABELSHARK →Quality gates
Image-quality metrics create explicit checkpoints before downstream conversion and delivery.
QC →Interoperability
Format-aware normalization and DICOM conversion reduce dependence on a single scanner ecosystem.
WSI-FORGE + DICOMIZER →Controlled delivery
Transfer state, retries, durations and failures remain observable rather than disappearing into a black box.
UPLOADER →Archival lifecycle
Long-term storage policy is separated from the clinical processing path and governed by validation-first rules.
WALL-E →Orchestration & audit
Database-backed lineage and orchestration provide a coherent operational view across independently evolving modules.
DPARS CORE →Processing and archival are intentionally separated.
Full-resolution derivatives support active processing. Reduced-resolution archival derivatives belong to a separate lifecycle and never re-enter the processing pipeline.
Scanner → metadata/routing → quality gate → normalization when required → DICOM conversion → controlled delivery.
Database-driven retention eligibility → reduced-resolution OME-TIFF derivative → independent validation → future cleanup eligibility under fail-closed safety rules.
Scientific oversight. Clear engineering ownership.
DPARS is developed within the Computational Pathology environment at the Institute of Pathology, Heidelberg University Hospital. Project-specific responsibilities below are separated from the group's formal institutional titles.
Cleo-Aron Weis
Scientific supervision and strategic oversight. Head of the Computational Pathology group.
Shahram Aliyari
Overall DPARS coordination, platform architecture, cross-service integration, orchestration and database/lineage design; lead responsibility for BabelShark, WSI-Forge, Uploader and Wall-E.
Balamurugan Thirukonda S. B.
DPARS Quality Control service responsibility. Formally listed by the group as Research Scientist, System Administrator and Operation Manager.
Maximilian Legnar
DPARS Dicomizer responsibility. Formally listed by the group as Research Scientist and System Administrator.
Christoph Blattgerste
Research Scientist in Computational Pathology Heidelberg.
Nils Englert
Research Scientist in Computational Pathology Heidelberg.
Ayk Jessen
Research Scientist and System Administrator in Computational Pathology Heidelberg.
Team membership and institutional titles are based on the public Computational Pathology Heidelberg page. DPARS service ownership is project-specific information supplied by the project coordinator.
One WSI. One traceable journey.
A conceptual walkthrough of how DPARS turns heterogeneous scanner output into an observable, standards-oriented digital pathology workflow.
Computational Pathology Heidelberg
The group publicly describes responsibility for developing digital pathology pipelines since 2023, addressing routine diagnostic and research applications separately. Its scientific portfolio includes BabelFish/BabelShark and optimization of HistoQC for routine digital pathology workflows.
Digital pathology pipeline development within Computational Pathology Heidelberg.
BabelFish established a framework for slide naming and metadata extraction.
BabelShark extends the concept toward improved robustness and scalability.
A modular platform view connects metadata, QC, conversion, delivery and lifecycle management.
Built for heterogeneous digital pathology.
DPARS focuses on the operational gap between scanner output and downstream clinical or research systems: identity, metadata, quality, interoperability, traceability and lifecycle management. The architecture is modular so individual services can be tested, replaced and extended without collapsing the workflow into one monolithic application.