Interactive Visualization of DICOM Volumetric Datasets in the Web

Providing VR Experiences within the Web Browser

Ander Arbelaiz

, Aitor Moreno

, Luis Kabongo

1,2

, Helen V. Diez

and Alejandro Garc

ıa Alonso

Vicomtech-IK4, Paseo Mikeletegi, 57, 20009 Donostia / San Sebasti

an, Spain

Biodonostia Health Research Institute, Donostia / San Sebasti

an, Spain

University of the Basque Country, Paseo Manuel de Lardizabal 1, 20018 Donostia / San Sebasti

an, Spain

Keywords:

Volume Rendering, Medical Imaging, DICOM, X3DOM, Ubiquitous Platforms, WebGL, WebVR.

Abstract:

Recently the possibility to visualize interactively volumetric datasets in the Web has opened new methods of

exploration and sharing of 3D images coming from different ﬁelds. At the same time, VR technologies are

gaining momentum in the society, where several HMD’s are ready to be bought. This paper presents how

volumetric datasets represented as DICOM images can be loaded and visualized interactively in a WebVR

compatible setup. DICOM images are loaded from local or remote repositories into X3D volume rendering

nodes, which are displayed in the VR devices using WebVR technology. The results show that WebVR

and X3D are compatible web technologies that can be joined together to provide easy and extensible tools

to interact with DICOM datasets. Some enhancements for the interactive VR and non-VR experiences are

presented.

1 INTRODUCTION

DICOM datasets are the de-facto interchange ﬁle for-

mat in the medical ﬁeld. Volumetric datasets are nat-

urally represented as sets of images, covering differ-

ent phases of the medical procedures. Typically, CT

and MRI scans are composed of 2D slices that can

be visualized and analyzed with specialized software.

These slices can be processed and inspected individ-

ually, but they provided more information if they are

considered as a whole volumetric dataset.

The introduction of volume rendering techniques

in desktop platforms was a milestone in the medical

ﬁeld. New methodologies and algorithms were imple-

mented to take into account the 3D information em-

bedded in the set of 2D slices that compose the scans.

In the last years, the web paradigm has in-

troduced new visualization, sharing and interac-

tion schema. Web technologies have been always

a way to democratize the technological advances

to the public, and therefore, new open standards

were deﬁned to harmonize the web ecosystem: The

X3D (Web3DConsortium, 2014), WebGL (Khronos,

2016) and WebVR (WebVR, 2016) standards are rel-

evant players in this standardization process.

X3D standard deﬁnes a set of nodes for volumet-

ric datasets, but they do not materialize the precise

Figure 1: Volume rendering example showing the AGE-

CANONIX volumetric dataset taken from the OsiriX repos-

itory (OsiriX, 2016).

functionality that the users require to interact with the

volumetric information. X3DOM (Fraunhofer IGD,

2014) provides a WebGL based implementation of the

X3D standard, and it can be used to visualize complex

3D scenes in the Web.

The introduction of new VR devices provides new

108

Arbelaiz A., Moreno A., Kabongo L., V. Diez H. and GarcÃ a Alonso A.

Interactive Visualization of DICOM Volumetric Datasets in the Web - Providing VR Experiences within the Web Browser.

DOI: 10.5220/0006154801080115

In Proceedings of the 12th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications (VISIGRAPP 2017), pages 108-115

ISBN: 978-989-758-228-8

Figure 2: Integration mechanisms devised to transfer the DICOM content into the X3DOM framework. Left) direct utilisation

of a canvas. Right) load mechanism using toDataURL function.

means to create virtual experiences for the users. Typ-

ically, state-of-the-art gaming companies ship their

standalone products, fully tested to provide the best

immersive experiences. This kind of developments

are very oriented to the speciﬁc product and very

costly to implement, relying in the commercial suc-

cess to recover the investment.

On the other hand, web technologies provide a

rich multimedia ecosystem to anyone in the world

with access to the Internet: text, audio, video and

interactive 3D thanks to WebGL-based technolo-

gies. WebVR is the next step in this technological

chain, aiming to provide support to the current head-

mounted displays (HMD) in the market through the

Web.

We identify web technologies as a key to give

easy access to volume visualizations for both special-

ists and non-specialists in different scientiﬁc ﬁelds.

The incorporation of the latest generation of Web

APIs like WebVR has opened new technological chal-

lenges. This work presents how volumetric datasets,

represented as DICOM ﬁles and stored locally or re-

motely, can be interactively visualized in web appli-

cations composed of the latest web technologies (see

Fig. 1). The preliminary research ﬁndings show that

this combination enhances the user experience in the

interaction with volumetric datasets, hiding the devel-

opment complexity under a clean and clear declara-

tive X3D scene.

This paper is structured as follows. The next sec-

tion provides a summary of the related work. Sect. 3

presents the work carried out to provide a combina-

tion of volume rendering on the web with WebVR us-

ing the X3DOM framework. Finally, this paper con-

cludes with the conclusions and future work.

2 RELATED WORK

This section is divided into two parts: the ﬁrst one

describes previous work regarding the volume ren-

dering raycasting algorithm and the second one de-

scribes how the DICOM datasets can be loaded using

JavaScript libraries.

2.1 Volume Rendering Algorithm

Volumetric visualization has been extensively studied

over the years. Nowadays it is used on a variety of

ﬁelds, but especially in medicine. Different volume

visualization techniques can be found in the literature;

our work focuses in ray-casting. Originally intro-

duced by Kajiya and Herzen (Kajiya and Von Herzen,

1984), ray-casting method was later translated to the

GPU by Kruger and Westermann (Kruger and West-

ermann, 2003).

Our work is integrated in the X3DOM frame-

work under the volume rendering component (Arbe-

laiz et al., 2016). Volumetric rendering is performed

with a single-pass ray-casting algorithm. In a nut-

shell, in the scene a 3D unit cube is rendered and

it will be used to place the actual volume data dur-

ing the ray-casting. Using the programmable pipeline

provided by WebGL, a single pair of vertex and frag-

ment shaders is used. In the vertex shader, each vertex

position of the cube is multiplied by the ModelView

matrix and the Projection matrix, transforming the

vertices into clip space.

In the fragment shader, the ray traversal is actually

computed. When the triangles of the cube are raster-

ized into fragments, the interpolated vertices of the

unitary cube represent 3D texture coordinates. From

the inverse ModelView matrix the camera position can

be obtained. By subtracting the interpolated vertex

position with the camera position the ray direction can

be determined. Taking both the interpolated position

as the ray origin and the ray direction, using a ﬁxed

length loop statement in the fragment shader the ray

traversal is created. For an in depth description of the

single-pass approach, we refer the reader to Mobeen

et al. (Mobeen and Feng, 2012).

The ray traversal is discretized into a series of

steps. At each step the position of the ray is used

as a 3D texture coordinates to fetch the volume data.

However, WebGL does not support 3D textures, as

ﬁrst stated by Congote et al. (Congote et al., 2011)

this can be overcome using a texture atlas (ImageTex-

tureAtlas): the cross sectional slices that compose the

volume are tiled into a matrix conﬁguration to com-

pose a single 2D texture.

Interactive Visualization of DICOM Volumetric Datasets in the Web - Providing VR Experiences within the Web Browser

109

Figure 3: Architecture and modules of the web application. DICOM images are stored in a local or remote repository. The

web application is loaded from Internet and presented to the user in non-VR mode. The web application process the DICOM

dataset to create the ImageTextureAtlas and prepare the X3D scene. WebVR is used to display the VR mode to the users.

2.2 DICOM Datasets Visualization

Volumetric data is often used in the medical ﬁeld in a

large variety of situations: from research and diagno-

sis to educational purposes. In terms of visualization

interactivity and usability, mobile platforms should

provide the same tools as their desktop counterpart.

In pursuit of a ubiquitous medical volumetric vi-

sualization, the support of DICOM ﬁle format is nec-

essary as DICOM is the standard de-facto that the

software of the medical imaging devices uses to store

their scans.

Medical imaging devices do not only produce the

actual set of 2D slices. They are linked with a large

amount of metadata related to the patient health infor-

mation and other medical procedures. DICOM is the

medical image standard to store and transfer all this

information from and between imaging devices and

medical image storage repositories (Fernandez-Bay

et al., 2000). The wide utilization of DICOM by all

manufacturers had a major impact on usability of the

ﬁle format. Resulting sometimes in a variable set of

tags to be read, interpreted and combined in order to

achieve coherent restitution of the images for the ﬁnal

user.

Cornerstone JavaScript library (Cornerstone,

2016) provides a set of functions to read and interact

with the 2D set of slices stored in DICOM ﬁles and it

relies on dicomParser to load DICOM tags, including

pixel data.

The combination of dicomParser with the volume

rendering nodes deﬁned in a X3D scene provides a

general and ubiquitous solution to the problem: the

slices pixel data is extracted from the DICOM ﬁle

and then, a texture atlas is created and linked to the

required texture ﬁeld of the VolumeData X3D node.

Fig. 2 shows two ways to accomplish the informa-

tion transformation from DICOM ﬁles to the X3DOM

framework. The ﬁrst one is to use a <canvas> node

deﬁned inside the ImageTextureAtlas node and then,

using JavaScript, ﬁll the canvas with a texture atlas.

The second integration method is to use the same

auxiliary <canvas> node but deﬁned out of the Im-

ageTextureAtlas node. The VolumeData node uses an

empty ImageTextureAtlas (no real URL is given). The

JavaScript loader draws the atlas texture in this canvas

as before, and then, the correct URL is provided as a

DataURL, which means that the whole volumetric in-

formation is encoded and passed in the URL.

3 WEB TECHNOLOGIES

ARCHITECTURE

This work presents how volumetric DICOM datasets

composed of 2D slices can be interactively visualized

in modern desktop browsers using exclusively open

web technologies. The preliminary results provide an

interactive VR user experience by presenting an ad-

equate web interface that interacts dynamically with

the volume rendering process.

In this section, we describe the architecture (see

Fig. 3), modules and their relationships among the se-

lected open web technologies. In our work, the cur-

IVAPP 2017 - International Conference on Information Visualization Theory and Applications

110

Figure 4: Asynchronous loading process from DICOM images to ImageTextureAtlas. Image shows an intermediate state

where the dataset is not fully loaded and hence, some subtiles are not ﬁlled into the ﬁnal HTML5 canvas.

rent stack of Web technologies are WebGL, WebVR,

X3DOM, JavaScript, SVG and CSS. These tools pro-

vide us with necessary base technology to develop

web VR and interactive applications to the users.

Next Subsection 3.1 introduces the technological

approach to load DICOM datasets into data structures

ready to be used with WebGL and X3DOM. Subsec-

tion 3.2 presents some interactive modiﬁcations that

can be performed to the original DICOM datasets

in order to select the Window Level. Subsection 3.3

shows our interactive Transfer Function editor, ca-

pable of ﬁne tuning the rendering output of the vol-

ume rendering algorithms. Finally, Subsection 3.4

presents how volumetric datasets can be visualized

from the inside of the volume, instead of the typi-

cal visualization from outside the bounding box of the

dataset.

3.1 Asynchronous Atlas Generation

The construction of the texture atlas is a technical

constraint of the Web based volume rendering algo-

rithm. An integration of the native Drag and Drop

HTML5 API into our web application provides an

easy and transparent approach for the generation of

the required ImageTextureAtlas (see Fig. 4). With-

out this functionality, the atlas creation must be per-

formed at the server in a pre-processing step. Then,

it can be transfered to the client device as an Image-

TextureAtlas image. With the proposed drag and drop

functionality we can avoid this step .

The DICOM data ﬁles can contain the volume

data in a single ﬁle, whereas, usually each DICOM

ﬁle represents a single slice with the additional meta-

data information. As stated in Sect. 2.2 using the di-

comParser open source library we can parse the vol-

ume data and compose the atlas in the background. In

our prototype, the rendering canvas is equipped with

drag and drop functionality. The user can drag a set

of DICOM ﬁles hosted in their own device ﬁlesystem

and drop them in the rendering canvas. The received

ﬁles will be ordered by ﬁlename and composed into

an atlas which is rendered in a hidden HTML5 2D

canvas.

This feature is specially interesting, because it

does not enforce users to adapt the volume data ﬁles

to a certain ﬁle format. As a consequence, it makes

it compatible with other application that output their

results as DICOM or other browser compatible com-

mon image formats, such as: JPG or PNG.

3.2 DICOM Metadata

DICOM ﬁles contain rich information that can be

used in the UI presentation. Some metadata attributes

can be added to the UI (capture date, acquisition ma-

chine, software version...) and others give informa-

tion about the dataset itself, i.e., the number of slices

and their resolution go normally with the pixel resolu-

tion (8, 10, 12 or more bits per channel). Taking into

account that the display devices we are targeting can

only display RGBA (8 bits per channel), a region of

interest is deﬁned with the aid of window width and

window center levels from the DICOM data.

These values specify a linear conversion from

stored pixel values in the DICOM ﬁle to values to

be displayed on the screen. To change these lev-

els, we provide dynamic sliders to the users. The

Interactive Visualization of DICOM Volumetric Datasets in the Web - Providing VR Experiences within the Web Browser

111

Figure 5: INCISIX dataset from the OsiriX reposi-

tory (OsiriX, 2016) with different window levels. On the

top window center= 395 and window width= 2068. On the

bottom, window center= 1056 and window width= 1489.

user can manually discard or extend the desired im-

age data range by modifying the window width and

window center levels. As a result, the contrast be-

tween structures within the volume can be adjusted.

Fig. 5 shows the INCISIX dataset from OsiriX li-

brary (OsiriX, 2016) with different window levels.

3.3 Interactive Transfer Function

In volume rendering, applying a transfer function

(TF) is one of the most common techniques to illus-

trate a volume. A TF is a lookup function that maps

each scalar value of the dataset [0-255] with a given

color and opacity. Typically, a set of predeﬁned and

general TF can be used (from red to green to blue,

rainbow schema...) but it is very common to ﬁnd

domain-speciﬁc TF’s (like in the weather radar infor-

mation) that are widely accepted by the experts of that

domain. Additionally, it is quite common to offer the

possibility to load a predeﬁned TF, but with the pos-

sibility of customizing it interactively.

In our web application, we have developed a web

TF editor based on the scalable vector graphics (SVG)

technology. In Fig. 6 the TF editor can be seen at the

bottom of the captures. In this interactive chart, an

histogram of the volume data values is plotted. Then

the user can add control points and assign a color to

each of them. Colors will be interpolated between

control points and the opacity will be calculated in

function of the control points height. As a result of

the user interaction, a 255 pixel width 1D RGBA im-

age is generated. This image is directly passed as a

texture input to the GPU and therefore, any changes

in the TF editor have an immediate effect in the vol-

ume rendering visual output.

3.4 Inside Exploration of Volume Data

Our work aims to the interactive visualization un-

der VR technologies, where the user can perceive the

stereoscopic effect of the 3D scene. In other words,

the user can perceive distances and what is near and

far in the volumetric information. During the inspec-

tion of the dataset, the user can move freely around

the dataset, and therefore, the dataset can be located

further or closer to the user. When the user gets closer

to the volumetric object, it would be very intrusive

for the virtual experience if the object suddenly dis-

appears. As the volume rendering algorithms render

the output on the surface of a 3D cube, if the user en-

ters inside this cube, nothing would be seen.

Some modiﬁcations of the raycasting algorithms

have been developed to allow correct visualizations of

the volumetric dataset even if the user enters the 3D

cube. An inside exploration allows the user to easily

discern the internal composition of the volume rather

than looking at the cross-sectional 2D images.

In our implementation we have allowed the ex-

ploration of the volume by dynamically changing

the initial position of the ray origin (see Fig. 7).

First the camera position is obtained from the inverse

ModelView matrix. Then using the maximum and

minimum boundaries of the cube, it can be deter-

mined whether or not the camera is inside the vol-

ume. If the camera is outside the cube, the ray origin

is assigned as the interpolated vertex position from

the output of the vertex shader (varying vertex po-

sition), if not, the ray origin is the camera position.

Without changing the ray origin, the camera face

of the cube will be clipped, making it impossible to

examine the inside. Thus, back face culling must be

disabled when internal exploration is required. When

the camera is moved into the cube, the ray-casting di-

rection for each ray is obtained by subtracting the in-

terpolated back face vertex position with the camera

position.

IVAPP 2017 - International Conference on Information Visualization Theory and Applications

112

Figure 6: Volume rendering web application interface pro-

totype in non-VR mode.

Figure 7: The visualization from the inside of the dataset

requires that the user moves the virtual camera location into

the cube that holds the volumetric dataset. Zoom function-

ality can be triggered with the wheel mouse.

3.5 WebVR Interactive Application

The creation of new VR oriented content with the

WebVR (WebVR, 2016) framework is simpliﬁed to

the extent that only knowledge of HTML, JavaScript,

CSS and related web technologies are required to de-

velop VR experiences. But even with this simplic-

ity, teachers, researchers and other non-technological-

aware target groups are not prepared for such a task.

To solve this issue, X3DOM provides a We-

bGL implementation of the X3D representation. The

declarative nature of X3D scene suits better for most

of the people who want to deploy VR experiences

but lack the technological capabilities of doing every-

thing by themselves. We have combined our proposed

medical functionalities implemented in the X3DOM

framework and test them along with WebVR.

Our preliminary demonstrations have been tested

with the Oculus Rift DK2, under Windows 10, with

the 1.5 and 1.6 runtimes. A developer build of

the Chrome browser with WebVR support (Chrome,

2016) has been used to test the VR developments.

The technical limitations regarding this setup will

be fully overcome when the WebVR 1.0 implemen-

tations of the standard reaches the release version of

the most common browsers: Firefox, Chrome, Safari

and Edge. Additionally, it is expected that mobile ver-

sions (for Android and iOS) would be available with

WebVR support.

Once the setup is running and a X3D volume ren-

dering scene is declared, loading a VR experience is

as easy as loading a URL. Then the VR experience is

started by clicking in a Enter VR button or the white

goggles icon, see Fig. 8.

Interactive Visualization of DICOM Volumetric Datasets in the Web - Providing VR Experiences within the Web Browser

113

Figure 8: Web based volume rendering in a VR scene with WebVR. Using the INCISIX dataset from the OsiriX repos-

itory (OsiriX, 2016). Left: inspection from a far point of view. Right: Visualization from a closer point of view. The

wireframe cube and the solid colored spheres help the users to know their location at any time.

3.6 VR Experience Enhancements

One of the key elements in the success of the VR is

to control how the VR experiences are perceived by

the users. In this regard, the transitions from non-VR

to VR and backwards should be controlled. In addi-

tion to this, any sudden change in the scene should

be carefully treated to avoid intrusive pop-up effect in

the user ﬁeld of view.

In our prototypes, DICOM ﬁles (through Corner-

stone JavaScript library) are loaded asynchronously,

i.e, each slice of the volumetric dataset is retrieved

as soon as it is loaded by the library. Therefore, ini-

tially the 3D scene could be empty, which is some-

thing to avoid in order to reduce stress to the users:

being ﬂoating in empty space is against ergonomic

rules regarding VR (Rebenitsch, 2015).

To solve this potential situation, the empty 3D

cube where the volume will be loaded is surrounded

with a 3D wireframe with colored solid spheres in

the corners of the cube (see Fig. 8). This solution

offers two beneﬁts: i) the scene is not empty while

the dataset is being loaded and ii) the colored spheres

give visual clues of the orientation of the scene that

can be used in case the users got lost. In our tests, this

3D visual clue has been proved very helpful in VR

environment, but also in non-VR setups.

The HMD’s like Oculus Rift are more focused in

immersive environments where the user is transported

to a virtual world that can be navigated and explored

in ﬁrst person. In our case, the typical navigation and

exploration of the volumetric dataset is more compat-

ible with the orbit navigation style, which conﬂicts

somehow with the nature of the VR. We refer the

reader to this collection of publications (Christie and

Olivier, 2009) to get an insight of the different camera

styles in virtual environments.

Our solution to cope with this situation was to

map the VR interaction capabilities to the actions that

should be carried out when the volumetric datasets are

being displayed:

• Mouse: Typically in VR, the mouse rotates the

user in the world. In our case, the mouse rotates

the 3D volumetric dataset around its center, like

in the orbit navigation style.

• Arrow Keys: Typically in VR, the keys allow to

move freely in the virtual scene (in walking or

ﬂying mode). In our case, the keys have been dis-

abled. In order to get closer to the volume or to

inspect it from the inside, the wheel mouse is used

to zoom into the scene.

• Head tracking: The modern HMD’s provide in-

formation about where the user is looking at. We

kept this functionality as it would be very intru-

sive to remove this functionality.

• Information display: In VR mode we remove all

the UI elements (TF editor, window level man-

agement...) in order to provide a clean VR expe-

rience. Only the Exit VR button is present. In pre-

liminary tests we added another button to reset the

viewpoint in case the user got lost, but ultimately,

it was discarded. It was easier for the user to get

out VR mode in that extreme case.

The resulting VR experience allows to inspect the

volumetric dataset from any point in an easy and

straightforward manner.

4 CONCLUSIONS

This work has introduced our preliminary efforts to

bring VR and non-VR volumetric visualizations on

IVAPP 2017 - International Conference on Information Visualization Theory and Applications

114

Figure 9: Leap Motion can be attached to the Oculus Rift in

order to provide gesture recognition of the user’s hands.

the Web by using a combination of open web tech-

nologies: WebGL, WebVR and X3D. DICOM ﬁles

are supported through the Cornerstone JavaScript li-

brary, providing a custom piece of code to construct

a texture atlas in the client. A X3DOM implementa-

tion of the volume rendering nodes has been used to

render the texture atlas.

Some UI elements and visual clues have been

added to help the users to interact with the volumet-

ric information (Window Level modiﬁcation, interac-

tive Transfer Function editor, 3D wireframe, visual-

ization from inside the volume and VR compatible

orbit navigation style). Our preliminary research ac-

tivities have dealt with situations regarding the VR

and non-VR environments and the transitions among

them. Our results show that there is room to improve

the Human-Computer Interaction within the web en-

vironment and the current and forthcoming collection

of HMD devices: Oculus Rift, HTC Vive, Microsoft

Hololens, Samsung GearVR, PlayStation VR, Google

Cardboard and Daydream...

The presented mixture of web technologies pro-

vide a real ecosystem that can facilitate the deploy-

ment of VR experiences of volumetric datasets for

experts in their corresponding ﬁelds, but unaware of

the technological advances under the hood. This sit-

uation will democratize the access to the information

and the distribution and sharing of volumetric visual-

ization among different user groups.

Our next steps will be oriented to the further test-

ing of UI elements with real users through subjective

surveys. Additionally, we will explore the possibility

to add 3D interaction widgets into VR mode through

the combination of gesture based recognition (using

devices like Leap Motion (Leap Motion, 2016), see

Fig. 9) and novel navigation modes.

REFERENCES

Arbelaiz, A., Moreno, A., Kabongo, L., and Garc

ıa-Alonso,

A. (2016). X3DOM volume rendering component for

web content developers. Multimedia Tools and Appli-

cations, pages 1–30.

Christie, M. and Olivier, P. (2009). Camera Control in

Computer Graphics: Models, Techniques and Appli-

cations. In ACM SIGGRAPH ASIA 2009 Courses,

SIGGRAPH ASIA ’09, pages 3:1–3:197, New York,

NY, USA. ACM.

Chrome (2016). Chrome developer builds with WebVR

support. https://webvr.info/get-chrome/.

Congote, J., Segura, A., Kabongo, L., Moreno, A., Posada,

J., and Ruiz, O. (2011). Interactive Visualization of

Volumetric Data with WebGL in Real-time. In Pro-

ceedings of the 16th International Conference on 3D

Web Technology, Web3D ’11, pages 137–146, New

York, NY, USA. ACM.

Cornerstone (2016). JavaScript library to display interactive

medical images including but not limited to DICOM.

https://github.com/chafey/cornerstone.

Fernandez-Bay

o, J., Barbero, O., Rubies, C., Sent

ıs, M.,

and Donoso, L. (2000). Distributing Medical Im-

ages with Internet Technologies: A DICOM Web

Server and a DICOM Java Viewer. RadioGraphics,

20(2):581–590. PMID: 10715352.

Fraunhofer IGD (2014). X3DOM: Open-source frame-

work and runtime for 3D graphics on the Web.

http://www.x3dom.org.

Kajiya, J. T. and Von Herzen, B. P. (1984). Ray Trac-

ing Volume Densities. SIGGRAPH Comput. Graph.,

18(3):165–174.

Khronos (2016). WebGL: OpenGL ES 2.0 for the Web.

https://www.khronos.org/webgl/.

Kruger, J. and Westermann, R. (2003). Acceleration

Techniques for GPU-based Volume Rendering. In

Proceedings of the 14th IEEE Visualization 2003

(VIS’03), VIS ’03, pages 38–, Washington, DC, USA.

IEEE Computer Society.

Leap Motion (2016). Truly immersive VR technology.

https:// www.leapmotion.com/.

Mobeen, M. M. and Feng, L. (2012). High-performance

volume rendering on the ubiquitous webgl platform.

In Proceedings of the 2012 IEEE 14th International

Conference on High Performance Computing and

Communication & 2012 IEEE 9th International Con-

ference on Embedded Software and Systems, HPCC

’12, pages 381–388, Washington, DC, USA. IEEE

Computer Society.

OsiriX (2016). DICOM image sample sets. http://

www.osirix-viewer.com/resources/dicom-image-

library/.

Rebenitsch, L. (2015). Managing cybersickness in virtual

reality. XRDS, 22(1):46–51.

Web3DConsortium (2014). Extensible 3D (X3D)

basic example archives. http://www.web3d.org/x3d-

resources/content/examples/Basic/VolumeRendering/.

WebVR (2016). Bringing virtual reality to the web. http://

webvr.info/.

Interactive Visualization of DICOM Volumetric Datasets in the Web - Providing VR Experiences within the Web Browser

115