Physica Medica: European Journal of Medical Physics - Volume 32 - Supplement 1 - February 2016

133 / 146

in the recent publication of the ISS ISTISAN 15/9 report, they are still of
impediment to performing exams in patients with medical devices. In this
work, some results regarding the use of medical devices in children un-
dergoing to MRI at 1.5 T and 3.0 T are presented.
Materials and Methods: Different methodologies to discriminate the safety
in MRI have been compared and in particular experimental measure-
ments on device, literature review, use of web resources and manufacture
information have been considered. This work shows also the specific role
of the medical physicist in the process of evaluating patient risk, and de-
scribes the technical consultative opinion/report structure and its area of
application in reference to the current international and national techni-
cal regulations. Furthermore, a method to measure spatial gradient of static
field is described.
Results: The results show then a summary of the reports produced for over
4 years and for over 40 pediatric patients with implantable active and passive
devices (neurostimulators, cochlear implants, baclofen pump, coronary
stents, pulmonary, orthopedic plates and fastening systems, catheters, valves
etc) and describe specific cases of interest and/or great complexity and de-
lineate the procedure recently adopted in our hospital.
Conclusion: The proposed methodology and procedure adopted by our hos-
pital allow safe MRI exams in patients with medical devices.
E.433
MRI FERROMAGNETIC DETECTOR SYSTEM FOR PATIENTS’ AND
OPERATORS’ SAFETY: EXPERIENCE IN OPBG
E. Genovese * A. Napolitano S. Donatiello C. Orlandi P. Toma’
V. Cannata’ a Enterprise Risk Management/Medical Physics, Bambino Gesù
Children’s Hospital, IRCCS, Rome, Italy; b Department of Imaging, Bambino Gesù
Children’s Hospital, IRCCS, Rome, Italy
Introduction: Legislative Decree 81/08 defines the indirect harmful effects
on health, those which are due to the risk from ferromagnetic objects in
static magnetic field or interference with electronic equipment or medical
devices. The metal detector is the most commonly spread system for de-
tecting metal objects. However, in the field of patient and operator MRI
safety, such device presents strong limitations since it is unable to dis-
criminate between ferrous objects and metal objects that are not affected
by the magnetic field. A possible alternative is constituted by ferromag-
netic detector (FD) present on the market in three different configurations:
portal, column or portable.
Materials and Methods: The 3 T scanner site in our hospital is equipped
with a portal system. User regulation has been determined based on the
Italian legislation (DM2/8/91) aswell as on international and scientific reports.
The sensitivity and specificity of FD detection have been experimentally
evaluated by recording the number of false positive and false negative events.
To this end, devices and objects commonly present in the MRI sites were
exploited: in particular devices and equipment used for children.
Results: The results showed that the use of detection systems in complex
environments, such as pediatric, should be carefully evaluated, in partic-
ular by identifying the best compromise between sensitivity detection and
false negative / positive alarms.
Conclusion: The FD use requires careful evaluation and a shared method-
ology among Magnetic Resonance Safety Officer, Magnetic Resonance
Medical Director/Physician and top hospital managers. Also, staff training
as well as site-specific safety regulations are needed.
E.434
MACHINE LEARNING TECHNIQUES IMPLEMENTED ON STRUCTURAL MRI
FEATURES AT DIFFERENT SPATIAL SCALES FOR PRESCHOOLERS WITH
AUTISM SPECTRUM DISORDERS
A. Giuliano I. Gor F. Murator I. Saviozzi P. Oliva R. Tancredi
A. Cosenza M. Tosett S. Calderon A. Retic a Istituto Nazionale di Fisica
Nucleare, Sezione di Pisa, Pisa, Italy; b Dipartimento di Chimica e Farmacia,
Università di Sassari, Sassari, Italy; c IRCCS Fondazione Stella Maris, Pisa, Italy;
d Dipartimento di Medicina Clinica e Sperimentale, Università di Pisa, Pisa, Italy;
e Istituto Nazionale di Fisica Nucleare, Sezione di Cagliari, Cagliari, Italy
Introduction: In the present study we aim to explore the predictive power
of structural brain magnetic resonance imaging (sMRI) data of autism spec-
trum disorders (ASD) on preschoolers. We highlighted the neuroanatomical
alterations that account for two-class separation in a case-control study
through a support vector machine (SVM) classification.
Materials and Methods: The sMRI data of 41 male children (2–6 years, 21
ASD matched to 20 controls (CTRL) for age and non-verbal intelligence quo-
tient (NVIQ)) were preprocessed with two protocols to finally extract
morphometric brain features at different spatial scales. The SPM8 soft-
ware was applied to extract the global volumes of the gray matter (GM),
white matter and cerebrospinal fluid segmented tissues, and the voxel-
based features of the GM. The Freesurfer image analysis suite 5.1.0 was used
to obtain region-of-interest (ROI)-based features (five surface-based fea-
tures for each of the 62 structures the cortex in parceled into). To investigate
which are the most relevant features to distinguish ASD from CTRL, we clas-
sified the three different sets with linear-kernel SVM, whose performance
is evaluated computing the area (AUC) under the receiver operating char-
acteristic (ROC) curve, estimated in leave-pair-out cross validation.
Results: The relevance of ASD features in a binary discrimination problem
emerges at the intermediate scale of GM subregions, where the best clas-
sification performance is achieved (74%). This value is enhanced to 80%when
considering only subjectswithNVIQ> 70. Themost discriminant brain regions
belong mainly to frontal, temporal and parietal lobes, that are part of cir-
cuits commonly involved in ASD and linked to typical ASD behavioral deficits.
Conclusions: The application of the SVM classifiers to sMRI data led to high-
light the salient features in the ASD-CTRL classification tasks and to identify
morphometric abnormalities of ASD children consistently at voxel and at
ROI dimensional scales.
E.435
A CUSTOM-MADE MR-COMPATIBLE PASSIVE DEVICE FOR FMRI
INVESTIGATIONS IN NEURAL CORRELATES OF ANKLE MOVEMENTS
DURING MOTOR TASKS
F. Levrero A. Bellini R. Iandolo I. Marre G. Bommarito
N. Oesingmann L. Fleyshe G. Mancardi M. Casadio M. Ingles
a IRCSS S.Martino-IST – U.O.Fisica Medica e Sanitaria, Genova, Italy; b Università
di Genova – DIBRIS, Genova, Italy; c Istituto Italiano di Tecnologia – RBCS,
Genova, Italy; d Università di Genova – DINOGMI, Genova, Italy;e Mount Sinai
School of Medicine – Dpt. Neuroscience, New York, USA; f Mount Sinai School
of Medicine – Dpt. Radiology, New York, USA
Introduction: Ankle motions are affected even in the early stages of neu-
rological diseases and a functional evaluation could be useful for an early
diagnosis. Here we present a custom-made MR-compatible device that
allows for independent dorsi-plantar flexion of each ankle, while prevent-
ing the transmission of significant motion to the head, and a preliminary
study involving healthy subjects.
Material and Methods: The prototype device is made of wood consisting
of two feet platforms and a thigh support. The subjects were laid on the
MR table with feet and thighs fixed with Velcro straps. The experimental
protocol includes ipsilateral and contralateral block tasks. Four healthy sub-
jects (2 males, 2 females, 25 ± 2 years, right-handed) performed seven
different active, passive, and proprioceptive tasks, monitored by 3 axes ac-
celerometers. The MR-compatibility was validated by S/N measurements
(EUROSPIN II test phantom). All the subjects underwent MRI at 1.5 T Signa
Excite (8 ch. T/R coil). The MRI protocol included high-resolution (voxel
1 mm3) 3D T1 FSGE and single-shot EPI (TR/TE 3000/60, matrix 64, FOV
240, ST 4 mm). fMRI data analysis was performed using FEAT (fMRI Expert
Analysis Tool) Version 6.0, part of FSL (FMRIB’s Software Library).
Results: The S/N test was performed both on phantom images (ROI
7800 mm2) and on noise images (RF transmitter switched off); no signif-
icant difference was detected with or without the device (min S/N 9.74;
max S/N 10.99). We found differences in the comparison between passive
and active tasks and between active and proprioceptive tasks. Specifical-
ly, in the proprioceptive tasks, more areas were recruited and the activations
were greater.
Conclusion: The experimental set-up resulted suitably and with very good
MR compatibility. The results suggest that, with proprioceptive targets,
passive movements can too induce significant brain activations;
e128 Abstracts/Physica Medica 32 (2016) e124–e134