Adaptive Streaming technology
Adaptive Streaming is similar to Raw Streaming in that data are not downloaded to clients – only frame-buffer contents are transferred, and all computation occurs on the server. The power of the server is used to render final screen images which are then compressed and transmitted to client devices in accordance with requirements regarding bandwidth usage, image quality, and interaction rates. The streaming method adapts to changing conditions in order to meet these requirements.
Unlike Raw Streaming, the methods and techniques used in Adaptive Streaming are highly optimized for the processing of medical image data. These is demonstrated in the following section.
Adaptive Streaming makes good use of available bandwidth (even if it is limited) and provides full functionality on the client side under most conditions. Servers also scale well with the number of clients.
The EasyViz system uses a proprietary form of Adaptive Streaming.
Adaptive Streaming Technology
A clinician is analyzing a stack of 2D images from an MRI scan of a patient – specifically, the clinician is modifying the window width/level for a particular slice in order to get a better look at a suspicious area. The following diagram illustrates how this scenario is handled by the Adaptive Streaming technology.
Illustration A shows that all computations are done on the server, and specialized formulae are transferred to the client indicating how to update the display on the client to match the server state. In B1, the clinician has a specific slice from a scan on the screen. In B2, the clinician has modified the window width/level of the same slice and only the changed pixels have been streamed to the screen. All other elements of the image remain unchanged and do not take up unnecessary bandwidth.
While the width/level is being modified, the Adaptive Streaming method does not send new frame images from the server to the client each time the image brightness or contrast changes. Rather, a CODEC is used such that a formula is sent that specifies in a predictive manner how pixel values should be modified on the client. This transmission of a formula is vastly more efficient than transmitting frame data or image data because the formula is significantly smaller in size.
