Zero-footprint delivery does not force one rendering model
Many teams use “zero-footprint” as shorthand for a browser viewer, but the real architecture question is where image decoding, rendering, and interaction logic run. Different workloads push those responsibilities to different places.
Common delivery and rendering choices
| Model | What moves over the network | Best fit |
|---|---|---|
| Browser-native web viewer | Metadata, objects, frames, or rendered derivatives | Modern DICOMweb-first viewing where the client can render interactively |
| Server-side rendering | Rendered pixels and interaction events | Heavy 3D or centralized GPU workflows where endpoints should stay thin |
| Application streaming | A remoted desktop or application session | Preserving legacy workstation software on managed remote endpoints |
Architecture implication
Do not treat zero-footprint as a synonym for server-side rendering. Browser-native viewers and remoted sessions can both be delivered with low endpoint setup, but their operational tradeoffs are different.
Cornerstone RenderingEngine
Official Cornerstone3D documentation for the client-side rendering engine used in browser-native imaging experiences.
Read the RenderingEngine documentationMedical imaging system architecture
AWS architecture guidance that shows how viewer, metadata, retrieval, and render components can be separated in a cloud design.
Read the cloud imaging architecture guidanceBrowser-native universal viewers depend on a clear stack from archive contract to viewport behavior
A browser viewer is more than a single UI bundle. It needs archive access, data-source configuration, rendering services, and workflow modules that can make sense of study context once data arrives.
Reference flow for a browser-native viewer
Loading diagram...
Server-side rendering and application streaming remain valid options when the compute burden should stay away from the client. The key is to choose them because of workload needs, not because a browser viewer was assumed to be impossible.
OHIF DICOMweb configuration
Official OHIF data-source documentation showing how a browser viewer connects to DICOMweb backends.
Read the OHIF data-source guidePushing pixels with NICE DCV
AWS HPC guidance describing remote visualization and why server-side pixel streaming remains relevant for GPU-heavy workloads.
Read the remote visualization guidanceChoose the rendering model by failure mode, not by what sounded modern in the procurement deck
The cleanest architecture is the one that fails in the least dangerous way for your users. Browser-native viewers fail when endpoints cannot keep up or archive contracts are incomplete. Server-side rendering fails when render capacity, session brokering, or network paths are underdesigned. Application streaming fails when legacy-desktop assumptions become a long-term operational tax.
Decision guide for choosing a rendering path
| Question | If yes | If no |
|---|---|---|
| Can the browser client render the target studies with acceptable latency? | Prefer browser-native delivery first | Evaluate server-side rendering or application streaming |
| Do you need to preserve a mature legacy workstation workflow immediately? | Application streaming may be the lower-risk bridge | A web-native viewer can avoid remoted-desktop debt |
| Are 3D and GPU-heavy workflows common enough to justify centralized compute? | Server-side rendering may be operationally cleaner | Keep the architecture simpler and closer to browser-native retrieval |
Failure-mode-oriented rendering decision path
Loading diagram...
Pushing pixels with NICE DCV
AWS remote-visualization guidance that helps explain when centralized pixel streaming is operationally justified.
Read the remote visualization tradeoff guidanceCornerstone RenderingEngine
Official browser-rendering engine documentation that grounds the client-side rendering option in a concrete implementation model.
Read the client-side rendering modelKnowledge Check
Test your understanding with this quiz. You need to answer all questions correctly to mark this section as complete.