![](bg4.png)
3.3 Feature Extraction for Parkinson’s
Disease Diagnosis
After intravenous injection, 123I-FP-CIT SPECT
binds to the dopamine transporters in the striatum. It
has been found that patients with Parkinson’s disease
exhibit a decreased uptake of the tracer (Bhidayasiri,
2006; Hauser and Grosset, 2012). An accurate di-
agnosis of Parkinson’s disease is important because
it enables us to monitor disease progression and the
therapeutic effects of the treatment. This motivates
the development of automated techniques for quan-
tification which neither depend on time consuming
operator-intensive work, nor expert skills in manu-
ally locating the regions of interest in the brain (Pa-
pathanasiou et al., 2006). We believe the α-stable
distribution and its properties can be used to develop
a computer aided diagnosis system as a decision-
making aid for Parkinson’s syndrome diagnosis for
automatic classification.
The discriminative area to perform diagnosis of
Parkinson’s syndrome in FP-CIT SPECT brain im-
ages is located in a specific region of the brain, the
striatum. According to the histogram of the intensity
values, this area also controls the degree of impulsive-
ness in the histogram. The degree of impulsiveness
is related to the characteristic exponent, which also
models the Paretian behaviour of the tails in alpha-
stable distributions. This property can be exploited
to assess differences between images belonging to
normal controls and patients with Parkinsonian syn-
drome, measuring the Paretian behaviour in the tail
of the distribution to extract the discriminant features.
Then, these features could be used for statistical clas-
sification using support vector machines or classifica-
tion trees.
4 NOVELTY, ADVANTAGES AND
DISADVANTAGES
The main advantage of the α-stable methods is that
they are generalizations of the Gaussian distribution
which is widely used in neuroimagingmethods, there-
fore they are expected to perform better than Gaus-
sian, or equally when the Gaussian assumption holds.
The main disadvantage of the α-stable distribution
is, mainly, the non existence of a closed form for
its probability density function, therefore, numerical
methods needs to be used to evaluate it.
The originality of the goals and methods envis-
aged in this paper is demonstrated by the fact that the
α-stable distribution has not been previously used in
neuroimaging apart from two very recent works pub-
lished in 2013 (Salas-Gonzalez et al., 2013a; Salas-
Gonzalez et al., 2013b). We believe is timely to ex-
tend these recently published methods, exploiting ad-
ditional and useful properties of the α-stable distri-
bution in the study of signal processing methods for
brain tomographic applications.
5 CONCLUSION
The Gaussian distribution and mixture of Gaussian
model are ubiquitous in brain imaging literature; nev-
ertheless, the Gaussian distribution, and the mixture
of Gaussians are particular limiting cases of the alpha-
stable distribution, and the mixture of alpha-stable
model. Sometimes, brain-imaging data present a cer-
tain degree of asymmetry and/or impulsiveness and
therefore, it can be modelled more accurately using
alpha-stables. For this reason, the alpha-stable dis-
tribution is expected to work better than those ap-
proaches in the literature assuming Gaussian distri-
bution of the data.
REFERENCES
Aarts, E., Helmich, R. C., Janssen, M. J., Oyen, W. J.,
Bloem, B. R., and Cools, R. (2012). Aberrant re-
ward processing in Parkinsons disease is associated
with dopamine cell loss. Neuroimage, 59(4):3339–
3346.
Balafar, M., Ramli, A., Saripan, M., and Mashohor, S.
(2010). Review of brain MRI image segmentation
methods. Artificial Intelligence Review, 33(3):261–
274.
Bhidayasiri, R. (2006). How useful is (123i) beta-CIT
SPECT in the diagnosis of Parkinson’s disease? Re-
views in Neurological Diseases, 3(1):19–22.
Hauser, R. A. and Grosset, D. G. (2012). [123I]FP-CIT
(DaTSCAN) SPECT brain imaging in patients with
suspected Parkinsonian syndromes. Journal of Neu-
roimaging, 22(3):225–230.
Mandelbrot, B. (1963). Summary of variation of certain
speculative prices. Journal of Business, pages 394–
419.
Papathanasiou, N., Rondogianni, P., Chroni, P., Themis-
tocleous, M., Boviatsis, E., Pedeli, X., Sakas, D.,
and Datseris, I. (2006). Interobserver variability,
and visual and quantitative parameters of 123I-FP-
CIT SPECT (DaTSCAN) studies. Annals of Nuclear
Medicine, 26(3):234–240.
Salas-Gonzalez, D., G´orriz, J., Ram´ırez, J., Schloegl, M.,
Lang, E., and Ortiz, A. (2013a). Parameterization of
the distribution of white and grey matter in MRI using
the α-stable distribution. Computers in Biology and
Medicine, 43(5):559–567.
SIGMAP2014-InternationalConferenceonSignalProcessingandMultimediaApplications
300