Distributed Edge Streaming

Images stream. They don't download. Start reading in seconds.

Sirona uses the same adaptive bitrate technology that powers Netflix — applied to diagnostic DICOM images. Studies stream progressively from CDN edge nodes worldwide. You see a usable image in seconds, then it refines to full diagnostic resolution in the background. No waiting for downloads. No frozen viewers. No bandwidth requirements.

See Sirona in Action

How Sirona is Different

Radiologists see images in seconds, not minutes

Legacy PACS downloads the entire study before displaying anything. A chest CT with 500 slices means waiting for all 500 slices to transfer before you see the first one. On a slow connection, that's minutes of dead time per study. Sirona streams images progressively — a compressed preview of the first slice appears almost immediately, subsequent slices stream as you scroll, and full diagnostic resolution arrives in the background while you're already reading. This isn't a thumbnail. It's a clinically usable preview that refines to pixel-perfect quality. The perceived performance difference is dramatic: radiologists start reading immediately instead of watching a progress bar.

How the streaming architecture works

CloudFront CDN, multi-resolution encoding, progressive loading, and intelligent prefetching — the infrastructure behind instant image delivery.

CloudFront CDN (400+ Edge Locations)

Images are cached on AWS CloudFront edge servers across 400+ locations worldwide. Studies route to the nearest edge node, minimizing latency regardless of where the radiologist sits.

Adaptive Bitrate Encoding

Every study is encoded at multiple resolution tiers. The streaming engine selects the best tier for your current bandwidth and upgrades in real time as conditions improve.

Progressive Image Loading

A compressed preview loads in seconds. Full resolution arrives in the background while you read. No progress bars, no download waits.

Intelligent Prefetching

The platform predicts which slices and series you'll view next based on hanging protocols and scroll patterns, prefetching them before you navigate.

WebGL GPU Rendering

Images render using your device's GPU via WebGL — hardware-accelerated windowing, zoom, pan, and MPR without any plugins or downloads.

Real-Time Bandwidth Adaptation

Bandwidth drops mid-session? The stream downgrades gracefully — no frozen screen, no disconnection. Bandwidth recovers? Quality ramps back up automatically.

Streaming at every layer of the stack

Edge Caching

400+ CDN nodes between the archive and the viewer

When a study is first accessed, it routes from the origin archive (AWS S3 in US-West-2) through CloudFront's global CDN. Subsequent accesses of the same study — by the same radiologist or a colleague at a different site — serve from the edge cache with very low latency. For high-volume practices, this means most studies are already cached at the nearest edge node before the radiologist opens them. The CDN absorbs the read load, so scaling the number of radiologists reading simultaneously doesn't require scaling origin infrastructure. This is how Netflix serves 250 million users from the same content library — the edge does the work.

400+ CloudFront edge locations across 90+ cities globally

Low-latency delivery from edge cache to viewer

Subsequent accesses serve from cache — no round-trip to origin

CDN absorbs read scaling — more radiologists don't require more servers

Encoding

Multi-resolution encoding for instant first paint

Every DICOM study ingested into Sirona is encoded at multiple resolution tiers — from a highly compressed preview suitable for instant display to full diagnostic-quality pixel data. When a radiologist opens a study, the viewer requests the lowest tier first, displays it immediately, then progressively upgrades to full resolution. This multi-resolution pipeline is the core of the 'start reading in seconds' experience. The preview isn't a thumbnail — it's a clinically usable representation of the image that lets a radiologist begin their assessment while the full-resolution data streams in parallel.

Multiple resolution tiers per study — from compressed preview to full DICOM quality

First usable image within seconds on typical connections

Progressive refinement to full diagnostic resolution in the background

Preview quality sufficient for initial clinical assessment during refinement

Intelligence

Prefetching that anticipates your next move

The streaming engine doesn't wait for you to scroll. Based on the active hanging protocol, the current series, and your navigation pattern, Sirona prefetches the slices and series you're likely to view next. Open a chest CT comparison? The prior study's matching series is already loading before you click on it. Scroll through an abdomen? The next 50 slices are cached in the browser before you reach them. This intelligent prefetching turns a variable-bandwidth streaming experience into something that feels local — images appear instantly because they were predicted and pre-delivered.

Hanging protocol-aware prefetching — priors and comparison series preloaded

Scroll prediction caches upcoming slices before navigation

Series-level prefetching based on reading patterns and protocol sequence

Browser-side cache eliminates re-fetching of recently viewed images

Rendering

GPU-accelerated viewing without plugins

Once images arrive at the browser, WebGL renders them using the device's GPU — hardware-accelerated windowing, zoom, pan, scroll, MPR reconstruction, and MIP generation. No Java plugin. No ActiveX control. No downloaded application. The same rendering pipeline works on a 30-inch diagnostic monitor, a laptop, an iPad, and Apple Vision Pro. Window/level adjustments are instant. Zoom is smooth. MPR reconstruction happens in the browser. This is diagnostic-quality visualization running on commodity hardware through a web browser — something legacy PACS requires a dedicated workstation and proprietary software to achieve.

WebGL GPU rendering — hardware-accelerated on any device

Zero plugins — no Java, no ActiveX, no downloads

Full diagnostic tools: window/level, zoom, MPR, MIP, 3D

Same rendering pipeline across desktop, laptop, tablet, and Vision Pro

Streaming performance

Seconds

to a first usable image — refined in the background

400+

CloudFront CDN edge locations worldwide

Low

latency from edge cache to viewer

Any

connection — adaptive streaming adjusts in real time

Images that appear instantly

Everlight RadiologyRead the partnership announcement

“This partnership marks the beginning of the cloud-native era in radiology software globally. Sirona has delivered an architecture that is fundamentally different from anything else in the market today — Sirona is unified, and cloud-native from the ground up. For teleradiology, that architectural distinction is existential, not incremental.”

Andy Donaldson

CTO, Everlight Radiology

Built Different: Why Sirona's RadOS Architecture is Different Than Legacy PACS

Nothing about Sirona's architecture is the same as the PACS and reporting systems you're used to. Hear from Sirona's technical adviser, Jeff Queisser (Co-Founder & EVP of Engineering at NYSE: BOX), about how Sirona was built from the ground up to change the way medical images are stored, analyzed, streamed, and accessed.

See all customer stories

FAQs

What is adaptive bitrate streaming for radiology?

The same technology Netflix uses for video, applied to DICOM images. Studies are encoded at multiple resolution tiers and streamed progressively from the nearest CDN edge node. The viewer displays a clinically usable preview in seconds, then refines to full diagnostic resolution in the background. Bandwidth changes? The stream adapts in real time.

How fast do images load?

A usable image appears within seconds on typical connections. Full diagnostic resolution follows progressively in the background. On high-bandwidth connections, the refinement is nearly instantaneous. On lower bandwidth, you start reading the preview while full quality arrives.

What bandwidth does Sirona require?

There is no hard minimum. The adaptive streaming engine works on any internet connection — from 5G cellular to satellite to fiber. Higher bandwidth means faster refinement to full resolution, but the initial preview loads quickly on any connection that can stream video.

How does edge caching improve performance?

CloudFront caches studies on 400+ edge servers worldwide. When a study is accessed, it routes to the nearest edge node, keeping delivery fast. Subsequent accesses by any radiologist serve from the same cache. For active practices, most studies are already at the edge before they're opened.

Is diagnostic quality maintained with streaming?

Yes. The streaming approach affects how fast images appear, not their final quality. The progressive loading pipeline starts with a compressed preview and refines to full DICOM pixel data. The final rendered image is identical to what you'd see with a traditional full-download approach — it just gets there faster.

Does this work with all modalities?

Yes. The multi-resolution encoding and streaming pipeline handles all DICOM modalities — CT, MR, CR, US, MG, PT, NM, and more. Multi-frame studies (fluoroscopy, cine MR) stream with frame-level progressive loading.

How does prefetching work?

Sirona analyzes the active hanging protocol, the current study, and your navigation pattern to predict which images you'll view next. Prior comparison series, upcoming slices, and protocol-defined series are prefetched before you navigate to them — making the experience feel local even though images are streaming from the cloud.