Speaker
Livia Torterolo
(University of Genoa)
Description
The SPM software package, based on the comparison of the
candidate case to normal
cases through a Statistical Parametric Mapping (SPM)
algorithm, is largely used by
the neurological research community to quantify ipometabolic
patterns in brain
PET/SPECT studies for the early diagnosis of Alzheimer’s
Disease (AD). Since the
accuracy of ipoperfusion maps is strictly related to the
number of normal studies
compared to the test image, a large set of images of normal
patients is required for
an accurate statistical analysis. However, due to ethical
issues and to high costs of
neuroimaging technologies, PET/SPECT studies for normal
patients are very rare.
Moreover, because of privacy and security issues, the images
of normal subjects
cannot be freely moved between sites or published by the
centre that collected them.
As a consequence, only doctors working at very large
institutions, locally owning
large databases of normal images, can usually carry out
SPM-based analyses. With our
application, doctors from small peripheral hospitals can
remotely access large sets
of normal PET/SPECT images provided by different medical
research institutes and
extract the information needed for the statistical analysis
without moving the
original image files. Grid technologies allow easy access to
distributed data as well
as to distributed computational resources. In order to
provide a user friendly
interface, remote access to SPM is being made available
through the Italian Portal of
Neuroinformatics. The portal contains a section entirely
dedicated to the statistical
analysis of PET/SPECT images, accessible by authorized users
only. Directly from the
portal, any authorised user can upload the suspect AD image,
select normal cases for
statistical calculation, and eventually collect the results
of analysis.
The basic steps of the Grid implementation of the SPM portal
service are listed below:
1. Acquisition of the test image on the user node
2. Transfer of the test image to the management node
3. Query on DB catalogue of normal images on the management node
4. Transfer of a small software executable for information
extraction to the
repository nodes
5. Extraction from normal images of the information needed
for the statistical analysis
6. Transfer of the extracted information to the management node
7. SPM statistical analysis on the management node
8. Transfer of SPM results to the user node
In many eHealth Grid applications, the crucial point is a
reliable and efficient data
and metadata management. In our application this is needed
for identifying suitable
normal images across hospitals. From a data management point
of view, the mandatory
requirements for the Grid implementation of the above
described SPM application are
the following:
- resources and services for the storage of PET/SPECT images
on different Grid sites;
- data management services for the registration in a
catalogue and the association of
metadata to images.
Two different environments (gLite and AliEn) have been
experimented, especially as
regards their functionalities concerning the management of
metadata, enabling
institutions to choose the deployment that better fits their
needs and resources. In
both environments, integration with the portal allows
seamless GUI interaction with
the available functionalities from the catalogue, the
analysis software and portal
services. Through the portal users can transparently exploit
both systems. Both
environments have provided satisfactory results.
Primary author
Livia Torterolo
(University of Genoa)
Co-authors
Andrea Schenone
(University of Genoa)
Barbara Canesi
(University of Genoa)
Elisa Molinari
(University of Genoa)
Marco Fato
(University of Genoa)
Piergiorgio Cerello
(INFN Turin)
Stefano Bagnasco
(INFN Turin)
