PICNIC
Portal-based Platform for MRI Processing of Neurodegenerative Diseases
J. Delgado-Mengual, Y. Vives-Gilabert, A. Sainz-Ruiz, M. Delfino-Reznicek
Port d'Informació Científica, UAB Universitat Autònoma de Barcelona, Edifici D Campus Bellaterra, Barcelona, Spain
B. Gómez-Ansón
Hospital Sant Pau i de la Santa Creu, Barcelona, Spain
Keywords: Web-applications, Web-portals, Neuroimaging, Telemedicine.
Abstract: The use of medical image processing techniques is increasing, especially those applied to the early
diagnosis of diseases, like neurodegenerative diseases. The software tools involved in it, are sometimes hard
to use for medical researchers and hospitals don't have neither the hardware resources nor the personnel
expertise, to accomplish the requirements. With these necessities born PICNIC, a technological hardware-
software platform with a web portal interface that integrates MRI processing tools with a user-friendly
interface, a database to manage clinical data and other services like the de-identification, visualization
images and job’s monitoring.
1 INTRODUCTION
With the increase of the magnetic field generated by
the scanners and the consequent increase in the
resolution of MRIs (Magnetic Resonance Images),
imaging techniques are becoming an important
reference in the early diagnosis of a lot of diseases.
For this reason, clinical researchers need
technological platforms dedicated to the storage and
post-processing of data for the analysis of medical
images.
Thinking in a scientific community that travels
around the world, web-portal technologies offer a lot
of possibilities to work remotely, and it also reduces
the required hardware resources and the informatics
knowledge of the users: only an actual Internet
browser and a Java Plug-in for the applets involved
in the portal are required. The high portability is also
an important issue taking into account that medical
researchers have to be able to access with different
browsers, regardless their operating system.
Moreover, some of these services have a strong
graphical interaction, so a platform needs a
mechanism to integrate the interfaces of the software
packages on the web portal with a user-friendly
interaction. Furthermore, the web portal has to
personalize the menus according to the user's profile,
to offer a highly safe and controlled environment.
The current state-of-the-art regarding web-
portals and technological platforms in neuroimage
processing is nowadays centred in complex data
pipelines. However, some of these advanced
platforms become complicated for medical
researchers due to the huge amount of specialized
software packages that they offer, which require a
high degree of understanding before using them.
GRID environments have been introduced in similar
projects with a most global scope.
PICNIC (PIC NeuroImaging Center) platform is
aimed to support research of neuroradiologist
communities whose data is centralised, so they don’t
require the GRID technology to share data but a
local cluster that allows to post-process MRIs in
parallel.
2 REQUIREMENTS AND GOALS
PICNIC is designed with the main objective of
unifying the access to the data and to the different
post-processing tools, until now installed in different
hardware systems and accessible via different
operating systems. The idea is to make the
239
Delgado-Mengual J., Vives-Gilabert Y., Sainz-Ruiz A., Delfino-Reznicek M. and Gómez-Ansón B. (2010).
PICNIC - Portal-based Platform for MRI Processing of Neurodegenerative Diseases.
In Proceedings of the First International Conference on Bioinformatics, pages 239-244
DOI: 10.5220/0002745202390244
Copyright
c
 SciTePress

interaction with the data and the software as easy as
possible for the medical researchers. In this section,
the requirements in terms of management of the
data, software, services and security are described.
2.1 Data Requirements
The users need to work with many types of images
and formats (DICOM, ANALYZE, NIFTI or
PFILES), depending on the project. Therefore,
researchers need a database of images' metadata,
where other information like the age or the gender
can be also associated to the images (very useful to
perform clinical studies). This database has to link
with other applications of PICNIC like the post-
processing tools or an image viewer, in order to
allow the users to interact with this data.
2.2 Post-processing Tools and Services
Requirements
The requirements associated to the software
packages concern the integration of different
hardware platforms and environments (operating
systems) to a unique interface. It is also required that
other tools can be implemented to PICNIC in the
future. The interface between the user, the software
and the hardware implies different issues: the design
of a GUI (Graphical User Interface) for the tools that
don’t have any, the development of scripts to run the
processes in parallel (when possible), to the export
of the results to excel files, etc. This interfaces and
the rest of web services have to be implemented
following the last HTML and CSS standards,
validated by the W3C to guarantee the portability
between the actual browsers, like Internet Explorer,
Mozilla Firefox, Safari, Opera, Google Chrome.
Regarding the services’ requirements, it is
necessary to implement an application to transfer the
data to the centre remotely via internet and also to
install a viewer to visualize MRIs. It is also
necessary to ensure that the data stored in the system
is correctly de-identified.
2.3 Security Requirements
It is crucial to keep personal data of the users and to
perform the authentication process as safe as
possible. Some restrictions have to be applied from
the network point of view, in order to control the
connections to the server.
3 NEUROIMAGING PLATFORM
The platform is designed with a modular structure to
satisfy the needs of the medical community in the
present as well as in the immediate future. As
represented in Figure 1, there are three basic areas in
the architecture scheme: User's area, Data area and
Post-Processing area. Every area is interconnected to
the others through a machine where the following
packages are installed: a CentOS, an Apache server,
mySQL, NX server and different post-processing
tools.
The whole system is strictly controlled with a white
list policy of security, where any action or access is
not permitted by default and the user has to contact
the Administrator to ask for permissions.
So, the “Administrator's” role is very important
to mediate the interaction between the final user and
the portal, in terms of permissions, job monitoring,
and other aspects related to the database and the
interaction with the hardware and the applications.
Figure 1: Architecture Scheme.
3.1 User's Area
The different user’s profiles that will interact with
the platform are: radiologists, neurologists,
psychologists and the administrators of the portal.
Administrators have a full access to all
functionalities, but the rest of the users have to ask
for permissions to use the software, services and also
to access to the data of research projects. The
combination of all of these permissions makes the
platform personal, in terms of access to the
information stored in disk, sections, software tools,
functionalities.
The advantages that PICNIC offers to the users are:
 The software packages and the system are
correctly installed and updated in adequate
machines with a properly operating system.
BIOINFORMATICS 2010 - International Conference on Bioinformatics
240

 The data (images, post-processing results...) is
properly organized in high available hardware
resources.
 The data is maintained in a secure and
consistent manner, with an adequate backup
policy.
 The software assistance, parametrization and
configuration, and analysis of the jobs logs are
covered by personnel formed with the software.
There is an opened communication line between
the users and the administrators to send suggestions,
issues and problems, to improve the platform and the
services offered.
3.2 Data Area
The storage is one of the main problems of clinical
researchers that need a large amount of data to
obtain significant statistical results from their
research. PICNIC offers different possibilities of
storage systems according to the importance of the
data (FAS2020 Cabin, Sun Thumpers...) and a
backup policy, previously negotiated with the
administrators.
In order to facilitate the access to the data, the
platform has a relational database with a query
interface, as a data bank of images for the different
research projects, as well as information to
administrate the platform. The connection between
the database and the storage system is performed
saving a link to the route of the files mounted with a
NFS system in the database entries.
3.3 Post-processing Software Area
The software tools integrated in the portal at the
moment (it will surely be extended in the future) are:
FreeSurfer, FSL, LCModel and SPM8. Some of
these tools allow running jobs in parallel in a cluster
which accelerate the process.
So, according to the characteristics of the
applications, the software tools are executed in the
different hardware systems:
 Cluster of machines, with 115 worker nodes of
two types: 1) Intel Xeon QuadCore 5355 @
2.66MHz (8 cores) with 16GB of RAM and 2)
Xeon Dual Core 5160 @ 3MHz (4 cores) with
8GB of RAM.
 High performance Work Stations Intel Core 2
Extreme Processor QX9650 (3.67 GHz OC,
1333MHZ FSB, 12 MB Cache) with 4GB of
RAM.
Basically there are two types of post-processing
tools included in PICNIC: tools with a graphical
interface to which the researchers are already
familiarized and “command line” tools, who require
a high degree of knowledge in operating systems,
usually Linux.
For the first case, a solution to virtualize
applications is required. PICNIC uses a Java applet
called NX Web Companion of the Nomachine
Company with a Free License for a non commercial
use. This applet allows virtualizing desktops to load
applications remotely, becoming a proper tool we
need to load remotely applications with a strong
graphical interaction. In this case, the application
runs in the machine at which it is installed, but
virtualized with the NX applet. An advantage
comparable to other techniques is that, with NX the
user's sessions in the OS are independent. This
means that the display interaction is completely
independent. The applet is inserted into the required
web pages and using a session file, the applet loads
the preferences of the session, to start with the
correct software and parameters.
Regarding the software tools that don’t have a
GUI, it is the case of some modules of FreeSurfer.
While the modules Tk-medit, Tk-surfer and Qdec
run with the NX plug-in, to post-process images in
the cluster the “command-line” mode is required.
For this purpose, a web interface designed in close
collaboration with clinical researchers (Figure 2),
has been implemented, avoiding the command line
functionality.
Figure 2: FreeSurfer interface in PICNIC.
4 SERVICES
Apart from the storage and the post-processing
PICNIC - Portal-based Platform for MRI Processing of Neurodegenerative Diseases
241

requirements, researchers need other services related
to the management of the data. These other services
implemented in PICNIC are described in this
section.
4.1 Upload Images
The portal includes “Jupload” and “Jdownload”
applications to allow users to transfer files (images
mainly) to the server. The images have to be in a
well-defined directory tree, due the fact that the
“Jupload” script verifies the tree directory to save
the data correctly in a temporal destination. Taking
into account that the images travel via Internet, even
if HTTPS protocol is used, for security reasons the
platform requires that they are correctly de-
identified at the moment of the upload. However, to
ensure that all the images in the platform are
anonymous, a de-identification process is executed.
Afterwards the images are added to the database.
4.2 Image De-Identification
After the upload process, the images are saved in a
temporal directory and the de-identification process
is automatically executed. The objective is to extract
personal data of the subject from the header of the
image. Some of the fields of the header that have to
be erased are: subject's name, acquisition's date,
gender, date of birth, hospital information...
When the de-identification process of a subject
finishes, the de-identifier sends an e-mail to the
administrator with the log of the process, indicating
if upload has finished properly and if not, the
problems that arose in the process. It also sends
another e-mail to the user indicating if the process
finished successfully. If everything was fine,
PICNIC adds a new entry to the database with the
image and the subject information.
The de-identification process is transparent to the
user and it runs in background after the upload
process has finished, independently if the user closes
the browser or the session.
4.3 Image Visualization
For radiologists, it is very important to have the
possibility to visualize the images. Therefore,
PICNIC includes an image viewer, MANGO, with
the possibility to visualize the images in three
formats (DICOM, NIFTI and ANALYZE). Like
other software commented before, the viewer uses
NX to load the images, opened in a new window.
If multiple images are opened at the same time,
the platform loads the viewer only once and the
viewer will open a different window for every image
to be loaded.
4.4 Jobs Monitoring
Some of the software tools used in PICNIC send the
jobs to the cluster of machines of PIC. The job
scheduler mechanism used at PIC is PBS (Portable
Batch System). PICNIC offers the possibility to
monitor the state of each job that has been sent to the
cluster at every moment (queued, running, out).
Moreover, it gives information like the number of
the job and the number of action at which the job
belongs to, useful information in case of problems.
Finally, it is also possible to stop the job with only a
one click and the system will automatically kill the
job in the cluster and update the status of the job in
the database.
5 SECURITY
The platform has to satisfy the requirements exposed
previously to keep it as safe as possible. With this
objective, some considerations have been taken into
account.
 A module to encrypt the data transmission with
SSL (Secure Socket Layer) protocol has been
installed in Apache.
 Other options of the Apache have been
configured to deny the access to paths or URLs
of the server and also to put as default mode of
access the HTTPS protocol.
 At the network level, only a minimum number
of ports have been opened (22 for NX and 80
for the web connections).
 Finally, at the implementation level, two
considerations have been taking into account:
1. The database queries are protected against
SQL Injection escaping characters (like ',
“”, \, /, |), using techniques that PHP offers.
2. The access to the hardware and data is very
restrictive to ensure that the user can only
reach the contents at which he/she has
privileges, noted in a white list policy.
6 VALIDATION
Some validation tests have been performed in the
way to get indicative results of the operation of the
BIOINFORMATICS 2010 - International Conference on Bioinformatics
242

platform. These tests have been centred in five basic
points: database, post-processing tools, transfer files,
image visualization and more in general some
attacks to the security of the system.
6.1 Database
The consistency of the data is very important, so it
has been necessary to verify in detail that all the
procedures that interact with the database work
properly. To do it, the test has been focused with a
double strategy: a preview test of the design of the
database using a database manager as
“phpmyadmin” and then in the implementation, a
validation test to ensure that the web interfaces has
been modified according to the actions sent to the
database.
6.2 Post-processing Tools
To test all the post-processing tools included in the
platform, the test was oriented in two ways:
 All the software tools that run with NX work
correctly in terms of NX session files for every
user, software and the parameters.
 For the post-processing tools where a web
interface was developed, an exhaustive test has
been performed to prove that all the features
implemented, work correctly in the cluster and
the results obtained are correctly added to the
database and imported to a spreadsheet.
6.3 Transfer Files and Images
Regarding the transfer of data, firstly it has been
tested that all the directory paths were correctly
processed by the applet in the case of uploading
images. Secondly, the de-identification process of
every slice of DICOM images was tested in order to
extract personal data of the subjects from the slices.
Some of this information is used as input of some
fields of the subject or image in the database, like for
example the age of the subject calculated from the
date of birth.
6.4 Image Visualization
In this case, Mango runs correctly and taking into
account that this service is related to the database
queries, it was also tested that the selected
subjects/images of the query were correctly opened
by the viewer. This was a triple test: database, NX
and viewer.
6.5 Security
In the way to prove the robustness and security of
the implementation, the following features were
tested:
 For a different URL of the platform, the access
to the derivative paths of the root path is not
allowed by the Apache options.
 The PHP implementations have been attacked
with SQL injection applications like “sqlier-
0.8b.sh”, obtaining negative results that confirm
the strength of the SQL implementation.
6.6 Analysis Result
To make balance of the results obtained by the
implementation and testing of the platform, some
contributions and weaknesses are exposed.
6.6.1 Contributions
 It is a scalable solution that can be completed
with more services.
 It is in accordance with the clinical research
mechanisms, with the advantage of being
accessible from everywhere.
 It is easy to use.
 The web-portal is dynamically loaded using
PHP and AJAX techniques.
6.6.2 Weaknesses
 The service is at mercy of the Internet
connection and traffic, so the transfer time can be
limiting.
 The volume of data that can be transferred with
the applets, is limited to 1GB.
 The authentication process in the platform is
performed with a password and a user-name. In
order to improve it, other solutions will be
envisaged in the future.
7 COPYRIGHT FORM
The authors of this work let to reference the content
of this paper, but not to make a derivative work of
this or to copy it.
8 CONCLUSIONS
PICNIC represents a portable solution for
PICNIC - Portal-based Platform for MRI Processing of Neurodegenerative Diseases
243

neuroimaging researchers to manage their data and
post-process images easily without any previous
knowledge of scripting. Based in a web technology,
it offers a high availability of the services from
everywhere.
PICNIC offers a robust infrastructure in terms of
hardware, with the adapted resources according to
the post-processing needs, running jobs in parallel in
the cluster of machines or running others
sequentially in powerful workstations. Periodical
backups of the original and post-processed data are
also performed.
Due to the modular structure of PICNIC, in the
future new software tools will be implemented in the
platform, according to the demands of the medical
community. Moreover, the authentication of the
platform will be improved by implementing a more
robust authentication system through certificates. In
a long term period and if the requirements of the
researchers are oriented to the data sharing, PICNIC
could be slightly modified to be used as interface for
a GRID infrastructure.
REFERENCES
UCLA – University of California Los Angeles (2009).
LONI - Laboratory of Neuro Imaging.
http://ww.loni.ucla.edu
UTAH Scientífic Computing and Imaging Institute (2009).
UCNIA - Utah Center for Neuroimage Analysis
(UCNIA) http://www.ucnia.org/
Martinos Center for Biomedical Imaging (2009) –
FreeSurfer Wiki
http://surfer.nmr.mgh.harvard.edu/fswiki/FreeSurferW
iki
Shawn N. Murphy, MD Ph.D. et al. (2006) - A Web Portal
that Enables Collaborative Use of Advanced Medical
Image Processing and Informatics Tools through the
Biomedical Informatics Research Network (BIRN)
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid
=1839506
SPM: Statistical Parametric Mapping (2009) -
http://www.fil.ion.ucl.ac.uk/spm/
LCModel (2009) –
http://s-provencher.com/pages/lcmodel.shtml
W3Schools (2009)- Online Tutorials
http://www.w3schools.com/
PHP: Hypertext Preprocessor (2009) http://www.php.net
Apache (2008) – HTTP Server Project
http://httpd.apache.org/
University of Florida (2008) – Installing Apache, PHP and
Mysql on Fedora core.
http://www.flmnh.ufl.edu/linux/install_apache.htm
MySQL (2009) – The world's most popular open source
database http://www.mysql.com/
MySQL Hispano (2009) – La Comunidad de usuarios de
MySQL http://www.mysql-hispano.org/
Andrew Crabb (2009) – Imaging Software
http://www.idoimaging.com/index.shtml
NIH Blueprint for Neuroscience Research - NITRC
The Source for Neuroimaging Tools and Resources -
http://www.nitrc.org/
ImageJ (2009) – Image Processing and Analysis in Java
http://rsbweb.nih.gov/ij/
Jack L. Lancaster, Ph.D., Michael J. Martinez - Research
Imaging Center, University of Texas Health Science
Center San Antonio (2009). Multi-Image Analysis GUI
Mango http://ric.uthscsa.edu/mango/index.html
Jupload (2009) – Jupload the multi-files uploader
http://jupload.biz/
Jdownload (2009) – Download Manager multiple files and
folders http://jdownload.jupload.biz/
Dr Allen D. Malony – Computer & Information Science
Department, Computational Science Institute CIBER
University of Oregon (2008). Distributed
Computational Architectures for Integrated Time-
Dynamic Neuroimaging.
Nomachine (2008) – Desktop Virtualization and Remote
Access Management Software.
http://www.nomachine.com/
BIOINFORMATICS 2010 - International Conference on Bioinformatics
244