Low-latency retrieval is the center of the imaging scenario
The AWS healthcare lens describes the medical imaging scenario as a cloud PACS or VNA problem with one overriding operational expectation: studies have to stay durably available while still being retrievable with low latency. That expectation shapes the whole topology, from viewer front ends to application tiers, metadata databases, image caches, and archive storage.
AWS recommends multi-AZ deployment, independent scaling of tiers, and protocol-aware access paths. The page explicitly notes that DIMSE and DICOMweb both matter, and that a Network Load Balancer can route DIMSE traffic on the appropriate ports.
Hot-path study retrieval in a cloud imaging platform
Loading diagram...
Medical imaging system reference architecture
Official AWS healthcare lens page covering multi-AZ design, retrieval protocols, tiered storage, hybrid locality, and AI-adjacent data lakes.
Read the imaging architecture guidanceAWS HealthImaging DICOMweb guide
Specific AWS documentation for DICOMweb-compatible access in AWS HealthImaging, useful when mapping the reference architecture to a managed imaging service boundary.
Review DICOMweb accessStorage tiering balances latency, durability, and cost
The lens is explicit that medical images may need immediate retrieval even years after acquisition, but not every object needs the same storage characteristics all the time. Newly ingested studies may sit in SAN, EBS, or FSx-backed hot storage, then land durably in Amazon S3 Standard and later move into lower-cost tiers such as Glacier Instant Retrieval as access declines.
How the reference architecture separates image storage roles
| Layer | Primary purpose | Typical fit |
|---|---|---|
| Hot cache | Low-latency display of recent or likely-to-open studies | SAN, EBS, FSx, or other high-performance storage |
| Durable archive | Authoritative long-term study retention with immediate access | Amazon S3 Standard or S3-based archive path |
| Lower-cost retention tier | Reduce cost as access frequency declines while preserving retrieval expectations | S3 Intelligent-Tiering or Glacier Instant Retrieval, depending on access goals |
| Metadata and cache | Resolve study, series, and object locations quickly | Managed database plus in-memory cache |
Study lifecycle across storage tiers
Loading diagram...
S3 Intelligent-Tiering overview
Specific S3 documentation for automatic movement between storage access tiers, relevant to the reference architecture recommendation on age-based cost optimization.
Review S3 Intelligent-TieringHybrid locality and AI-adjacent paths stay part of the architecture
The imaging scenario is not purely cloud-native in the abstract. AWS calls out Local Zones and Outposts for latency, hybrid architecture, and data-sovereignty concerns, along with redundant connections and Direct Connect for customer sites with high study volume. The same page also notes that data lakes can support both operations and R and D, including AI feature development and labeling.
Hybrid retrieval path for high-volume imaging sites
Loading diagram...
- Keep direct clinical retrieval and AI dataset creation as separate operating paths, even when they share the same source archive.
- Design hybrid connectivity around actual site volume and tolerated disruption, not just general cloud preference.
- Preserve standards-aware access for imaging workflows while making research exports deliberately governed and secondary.
AWS Direct Connect user guide
Specific AWS networking documentation relevant to the lens recommendation that high-volume imaging sites should use Direct Connect.
Review Direct Connect guidanceKnowledge Check
Test your understanding with this quiz. You need to answer all questions correctly to mark this section as complete.