Pattern Classification Methods for Analysis and Visualization of Brain Perfusion CT Maps - Computational Intelligence Paradigms in Advanced Pattern Classification - page 164

The author used NyARTToolkit CS [40] that is the implementation of well
known ARToolkit [15] in C# programming language. The important fact is that
this module uses no native code (only managed).

7.2 Real Time Rendering of 3D Data

There are two most popular methods capable of real time rendering 3D volume
data: texture based volume rendering and volume ray casting. The author has cho-
sen the ray casting GPU accelerated algorithm described in [17]. Algorithm uses
standard front to back blending equations in order to find color and opacity of ren-
dered pixels:

C = C + ( − α ) α C

(7.1)

dst dst dst src src

α = α + ( − α ) α

(7.2)

dst dst dst src

C α

Where and are the color and opacity values of the rendered pixels and

dst dst

C α are the color and opacity values of the incoming fragment.

The approach proposed in [17] includes standard acceleration techniques for vo-
lume ray casting like early ray termination and empty-space skipping. By means of
these acceleration techniques, the framework is capable of efficiently rendering large
volumetric data sets including opaque structures with occlusions effects and empty
regions. For many real world data sets, the proposed method is significantly faster
than previous texture based approaches, yet achieving the same image quality.

and

src src

7.3 Augmented Desktop - System Performance Test

One of possible AR visualization interfaces is the Augmented Desktop (the mod-
ule for AR visualization of clinical data created by author). It is capable to render
not only 2D perfusion images but also all types of visual data that are used during
perfusion treatment (in example 3D volume data form angio CT) in real time
(Fig 9). The image data can be intuitively zoomed in and out just by changing its
distance from the ones eyes. If image is near enough to camera in the left side of
the screen description of image done by DMD algorithm is displayed (Fig 9 (B),
(C), (D)).
The rendering algorithm was implemented in XNA Framework with High Lev-
el Shader Language (HLSL). XNA Framework was originally designed as a set of
tools that facilitates computer game development and management for Windows
an Xbox. The basic window frame of XNA Framework is called “Game” and au-
thor used this notation in the paragraph.