DICOM Bridge Service: Edge Orchestration for Medical Imaging

Executive Summary

Medical imaging data is often siloed within proprietary scanners and PACS servers. I engineered the DICOM Bridge Service, a high-performance edge gateway that breaks these silos.

The service acts as an intelligent intermediary that ingests studies from modalities (CT, MRI, X-Ray), performs real-time image transformation, and orchestrates delivery to multiple destinations—including local PACS storage and Indonesia’s National Health Platform (SATUSEHAT).

Technical Impact

National Interoperability: Engineered the conditional routing logic required to sync clinical imaging metadata with the SATUSEHAT Kemenkes platform.
Operational Speed: Automated the generation of JPEG previews from heavy DICOM files, enabling clinicians to view images instantly via web dashboards without waiting for full DICOM retrieval.
Smart Routing (Fan-out): Implemented a multi-destination engine that allows a single imaging study to be delivered to separate PACS systems and cloud storage simultaneously.
Legacy-to-Modern Bridge: Successfully bridged legacy imaging equipment (some running older DICOM dialects) with modern REST/FHIR-based cloud architectures.
From Research to Production: Successfully evolved the project from a containerized research tool (docker-dcmkit) into a battle-tested production service.

Edge Gateway Architecture

The system sits at the “Edge” of the hospital network, acting as a traffic controller for binary medical data.

flowchart LR
    Modality[Imaging Modalities<br/>CT / MRI / X-Ray]
    
    subgraph "DICOM Bridge (Go Edge Service)"
        Receiver[C-STORE SCP]
        Processor[Metadata & JPEG Pipeline]
        Router[Intelligent Dispatcher]
    end

    PACS_A[PACS System A]
    PACS_B[PACS System B]
    Cloud[National Health Platform<br/>SATUSEHAT]
    Dashboard[Web-based EMR / Viewer]

    Modality -->|DICOM C-STORE| Receiver
    Receiver --> Processor
    Processor --> Router
    
    Router -->|Parallel Routing| PACS_A & PACS_B
    Router -->|Conditional Sync| Cloud
    Processor -->|Store Previews| Dashboard

Key Engineering Features

High-Performance Ingestion: Built on a Go service layer that wraps DCMTK and GDCM for robust handling of the DICOM network protocol.
Image Transformation Pipeline: Automatically extracts pixel data and applies window-leveling to generate web-ready JPEGs, significantly reducing front-end latency for non-diagnostic viewing.
Conditional Dispatcher: A rule-based engine that determines where data should go based on the imaging study’s metadata (Modality type, Patient ID, or Referring Physician).

The Challenge: Clinical Silos

In most hospitals, if you buy a new PACS system, you have to reconfigure every single scanner (Modality) to “talk” to it. This is a massive operational bottleneck. Furthermore, integrating with the national health platform (SATUSEHAT) adds a layer of complexity that legacy scanners simply cannot handle natively.

The Engineering Solution

1. The “Man-in-the-Middle” Strategy

By placing the DICOM Bridge between the scanners and the PACS, I created a single point of configuration. Scanners send data to the Bridge, and the Bridge handles the rest. Result: Adding a new PACS or cloud backup now takes minutes of configuration instead of days of physical modality tinkering.

2. Research-to-Production Pipeline

This project started as docker-dcmkit, a personal research project to containerize DICOM tools. Recognizing its potential, I expanded it into a full service with:

Standardized logging.
Error-handling for unreliable hospital networks.
Automated cleanup of temporary cache files.

3. SATUSEHAT Integration

I implemented the specific API protocols required to bridge DICOM metadata with the national platform’s REST requirements, ensuring the organization remains at the forefront of national healthcare digitization.

Results

Unified Edge Layer: One service manages all outbound imaging traffic.
Zero Data Loss: High-reliability routing even in bandwidth-constrained environments.
Instant Access: Clinicians gain a “web-view” of images seconds after a scan is completed.
Future-Proof: Ready to integrated with any New API or cloud destination.

Technology Stack

Language: Go (Golang)
Tooling: DCMTK, GDCM (wrapped in Go)
Protocol: DICOM (C-STORE, C-ECHO), REST
Containerization: Docker
Cloud Integration: SATUSEHAT Kemenkes API data transfer