Ordinal classification of the affectation level of 3D-images in Parkinson diseases

Abstract

Parkinson’s disease is characterised by a decrease in the density of presynaptic dopamine transporters in the striatum. Frequently, the corresponding diagnosis is performed using a qualitative analysis of the 3D-images obtained after the administration of 123^{123} 123 I-ioflupane, considering a binary classification problem (absence or existence of Parkinson’s disease). In this work, we propose a new methodology for classifying this kind of images in three classes depending on the level of severity of the disease in the image. To tackle this problem, we use an ordinal classifier given the natural order of the class labels. A novel strategy to perform feature selection is developed because of the large number of voxels in the image, and a method for generating synthetic images is proposed to improve the quality of the classifier. The methodology is tested on 434 studies conducted between September 2015 and January 2019, divided into three groups: 271 without alteration of the presynaptic nigrostriatal pathway, 73 with a slight alteration and 90 with severe alteration. Results confirm that the methodology improves the state-of-the-art algorithms, and that it is able to find informative voxels outside the standard regions of interest used for this problem. The differences are assessed by statistical tests which show that the proposed image ordinal classification could be considered as a decision support system in medicine.

Full text



Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports
Ordinal classication
of the aectation level
of 3D‑images in Parkinson diseases
Antonio M. Durán‑Rosal1*, Julio Camacho‑Cañamón2, Pedro Antonio Gutiérrez2,
Maria Victoria Guiote Moreno3, Ester Rodríguez‑Cáceres4, Juan Antonio Vallejo Casas3 &
César Hervás‑Martínez2
Parkinson’s disease is characterised by a decrease in the density of presynaptic dopamine transporters
in the striatum. Frequently, the corresponding diagnosis is performed using a qualitative analysis of
the 3D‑images obtained after the administration of
123
I‑ioupane, considering a binary classication
problem (absence or existence of Parkinson’s disease). In this work, we propose a new methodology
for classifying this kind of images in three classes depending on the level of severity of the disease
in the image. To tackle this problem, we use an ordinal classier given the natural order of the class
labels. A novel strategy to perform feature selection is developed because of the large number of
voxels in the image, and a method for generating synthetic images is proposed to improve the
quality of the classier. The methodology is tested on 434 studies conducted between September
2015 and January 2019, divided into three groups: 271 without alteration of the presynaptic
nigrostriatal pathway, 73 with a slight alteration and 90 with severe alteration. Results conrm that
the methodology improves the state‑of‑the‑art algorithms, and that it is able to nd informative
voxels outside the standard regions of interest used for this problem. The dierences are assessed by
statistical tests which show that the proposed image ordinal classication could be considered as a
decision support system in medicine.
Parkinson’s disease (PD) is a progressive, neurodegenerative disease that causes characteristic motor symptoms
of tremor, bradykinesia, and postural instability1. PD is caused by deterioration of the dopaminergic neurons
in the extrapyramidal tract of the midbrain, that modulates voluntary movements and controls maintenance of
posture and coordination of gait. Degeneration of neurons that release dopamine causes an imbalance of excita-
tory (acetylcholine) and inhibitory (dopamine) neurotransmitters in the region2.
Presynaptic dopamine transporter density can be detected by neuroimaging techniques, which are now stand-
ard practice in the diagnosis of neurodegenerative disorders such as PD. Dopamine deciency in the striatum
can be evaluated using nuclear medicine techniques. e
123
I-ioupane (DaTSCAN, General Electrics Healthcare
Limited, Little Chalfont. Bucks HP79NA U.K.) is a radiopharmaceutical, widely used for this purpose, which
binds to the presynaptic dopamine transporters in the caudate and putamen and allows the density of these to
be evaluated with high sensitivity3,4. is method enables an early diagnosis of neurodegenerative parkinson-
ism. DaTSCAN can be suitable for assessing the presynaptic decit in early stages of disease. Moreover, it can
dierentiate patients with neurodegenerative parkisonism from those with other forms of parkinsonism5.
Visual assessment allows evaluating the normality of dopamine transporter (DAT) binding and the magnitude
of compromised DAT binding, specially focusing on asymmetry and aected structures6. In clinical practice,
DATSCAN images are commonly interpreted with careful visual assessment of the striatal tracer binding. is
approach has high diagnostic accuracy and excellent interobserver agreement7. But this visual classication may
be subjective and strongly depends on the experience and the fatigue of the person in charge of the labelling. Also,
binding quantication is based on a manual delineation of striatal regions of interest (ROIs), but this technique
is still subjective and dependent on one operator8.
Currently, visual diagnoses, with one or several specialised observers, are combined with automatic com-
puter systems that analyse the data and are able to distinguish between two classes, pathological or normal. In
OPEN

            
            
 *
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
the case of PD diagnosis, the problem was initially tackled with a classical approach, i.e. by quantifying the loss
of neuronal dopamine in the striatum9. Posterior attempts to automatise PD diagnosis used semi-quantitative
assessment of images with
123
I-Ioupane10,11. Moreover, computer systems based on machine learning12 has
recently shown promising results based in the striatal size and shape and on dierent image properties. In general,
most advanced computational techniques of examining images could help identify the disease and build eec-
tive decision-support systems for the diagnosis of Parkinson. e European Association of Nuclear Medicine
(EANM) recommends quantitative analysis, associated with this type of support models6.
e majority of works that seek to address the problem of classication of pathological and normal patients,
based on their functional images, apply automatic learning techniques based on ROIs13. Using the semi-quan-
titative methods recommended by the EANM guidelines, independent ROIs of caudate, putamen and occipital
background of the axial cuts that integrate the grooves are performed, and the relationship between the dierent
regions is analysed. However, training binary classication models (pathological and normal) by treating these
ROIs as the units of information ignores voxels outside the ROIs and can limit performance. Note that a random
variable associated with a voxel (minimum volumetric unit of the image) can sometimes be signicantly impor-
tant for the classication task, so it is important to treat them individually. In this way, specic voxels, within the
region, may have greater dopamine transporter loss than others and therefore may be more informative. Direct
relations between the voxels and the classication task can be found and, by ignoring these ROIs, we avoid letting
ourselves be guided by prior medical inuence.
Extra-striatal regions may contain a signicant amount of
123
I-ioupane dopamine targets, and the use of these
regions may improve the statistical reliability of the model. is could be the case, for example, if extra-striatal
regions are combined with the caudate to provide a reference for comparison with the putamen. Including all
voxels in these regions as discriminant characteristics and letting a feature selection algorithm decide which
voxels are more informative (i.e. informative voxels) seems to have a higher potential than assuming a priori
which voxels are important according to some expert knowledge. erefore, our study develops an algorithm
based on voxels rather than based on classical ROIs. In studies with 18F-uorodeoxyglucose PET, increases of
metabolic activity were identied in pallidothalamic and pontine areas14. Recent experimental studies, using a
NMR-based metabolomic approach, show metabolic imbalance among dierent brain regions, especially in the
midbrain and right cortex15.
In this work, SPECT 3D-images obtained aer the administration of
123
I-ioupane are labelled in three
classes depending on the level of aectation of the image, that is, without alteration of the presynaptic nigros-
triatal pathway (class 0), with a slight alteration (class 1) or with severe alteration (class 2)16. In this sense, and
given the natural order between classes (
0<1<2
), we propose to use ordinal classication models. e main
objective is not only a correct classication of the patterns but also a nerror reduction for the misclassied ones.
In other words, if the classier cannot correctly classify a pattern of the class 0, the misclassication in class 1 is
preferred than an error in class 2. is paradigm has been applied in a lot of prediction problems of medicine17,
such as cesarean section rates18, breast cancer19, liver transplantation20, or Alzheimer progression21, among oth-
ers. erefore, the application of ordinal classication to PD diagnosis seems to be appealing.
Grading in PD could be dened as the act of classifying patients according to a global staging or severity
ranking. New formulations of levodopa and novel delivery systems are currently being evaluated and gradually
introduced in clinical practice in an attempt to prevent or treat levodopa-related motor complications. With this
gradation, we could study when is the best moment to introduce levodopa or other treatments in the future22.
Furthermore, this gradation is relevant for several reasons as, for example, allowing analysis of the relationships
between group’s characteristics (grades or levels) and many other factors (duration of disease, eects of therapy,
etc.); intergroup comparison, selection of patients for clinical research...23. Pasquini etal.24 also suggest that cau-
date quantication of DAT availability shortly aer diagnosis may have an important role in identifying patients
at risk of clinical progression to cognitive impairment, depression and gait problems in the near future. We can
use the scales for outcome in Parkinson’s Disease–Motor (SCOPA-Motor)25 or the Non-Motor Symptoms Scale
(NMSS). e NMSS has 30 items in nine domains26. For the purpose of obtaining an objective quantication
technique, more complicated methods need to be proposed, and observer-independent automated quantica-
tion methods are preferable27.
Given the large number of features presented in these 3D-images, and considering all the image voxels
(instead of putamen and caudate) as we stated ahead, it is necessary to use techniques to reduce them. It is one of
the problems tackled in this work. It is true that, in machine learning, there are several techniques for the reduc-
tion of the number of features, for instance, compressive sensing-enhanced feature selection28 or using evolution-
ary computation29. e methods of machine learning for the selection of characteristics can, in general, be divided
into two classes: wrapper30, lter31, and embedded methods32. In this paper, we will use a lter method where
each feature is ranked individually based on specic statistical measures without taking into account the type
of learning algorithm used in the classier. We will use the ReliefF algorithm as a feature selector for a nominal
classier33,34, to then develop for ordinal classication. ReliefF makes a ranking of the features according to their
discriminatory quality, where the user needs to specify the percentage of the best characteristics to be selected.
Another problem that we need to tackle is the low number of patterns of the dataset for the least frequent
class (slight alteration). e most natural and most common method to reduce overtting on image data is to
articially enlarge the dataset using label-preserving transformations35,36. A common practice for augmenting
datasets is padding 4 pixels on each side and then doing random cropping and random ipping on the y during
training37. In the same way, in ImageNet dataset38, it is common to subtract the mean and divide by the standard
deviation for each input channel and follow the data augmentation as described by Krizhevsky etal.39 In the
context of imbalanced classication, the noisy replication method40 has been proven to be an eective approach
in improving accuracy for the minority classes, specially for binary classication problems. It randomly chooses
minority patterns and replicate them adding a small amount of noise. In this work, we develop a new method
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
to generate new synthetic images and voxels based on the adjustment of the statistical distribution of the voxels,
where, given that the same coordinate system is used for all 3D images, we can unequivocally characterise the
position of a given voxel.
ere are some recent approaches that consider machine learning and advanced computational techniques
for the detection of PD. It is common to the use of collected voice measures. For instance, Kaur etal.41 proposed
an ensemble of 25 state-of-the-art regression models to analyse the prediction of the motor Unied Parkinson’s
Disease Rating Score (UPDRS). e number of patients (42) is considerably limited. Canturk42 considers a
combination of the dynamic spiral test (DST) and static spiral test (SST) for early detection of PD. For this, the
fuzzy recurrence plot is used to transform time series into grayscale images, which are then analysed using two
deep networks (AlexNet and GoogleNet). e dataset used is based on 62 PD patients and 15 healthy ones. e
conclusions are that the Y signal provides better results for DST, while a combination of Y and P signals performs
better in SST. In the work proposed by Naseer etal.43, deep transfer learning is applied to the identication of
PD, using as input handwriting images, which are one of the earliest indicators of the aection. e source tasks
are ImageNet and MNIST datasets, and the use of freeze and ne-tuning of transfer learning is investigated in
combination with data augmentation. Based on a dataset of 37 PD patients and 38 healthy subjects, the best
results are nally obtained with ne-tuning-based approach considering the ImageNet and PaHaW dataset.
However, all these works approach the problem as a binary classication problem (considering whether the
disease is present or not) and none of them uses 3D-images of
123
I-ioupane as input data.
e main objective of this work is to obtain a tool which acts as decision support system in PD diagnosis for
assessing the level of aectation of the patients using 3D images. e use of grading system (instead of a binary
classier) would increase the information obtained by the experts to better decide the treatment. Moreover, as the
complexity of the problem increases (three classes instead of two), another objective is to include specic methods
to deal with the imbalance ordinal nature of the dataset. Summarising, the main contributions of this paper are:
• e use of an ordinal classier due to the natural order of the labels in the dataset. e patterns will be clas-
sied into three classes: no alteration of the pathway (class 0), slight alteration (class 1) or severe (class 2)
alteration. For that, we use the ordinal logistic regression model available in the soware mord developed by
Pedregosa-Izquierdo44.
• e development of a new method to reduce the number of characteristics of an ordinal classication dataset.
e method, referred to as ordinal ReliefF, is a modication of the standard state-of-the-art ReliefF34, taking
the ordinal nature of the problem into account.
• e development of a new technique for generating synthetic patterns based on the statistical distribution
of the macrovoxels. is method tries to nd the best statistical distribution for each voxel (selected from a
set of well-known distributions). en, new voxels are generated using this distribution. e synthetic image
will be used as part of the training set to improve the quality of the classier.
• e application of the methodology into a real-world dataset obtained by the UGC Medicina Nuclear of the
Hospital Universitario “Reina Sofía” (Córdoba, Spain), including 434 studies with dierent levels of altera-
tion of the presynaptic nigrostriatal pathway: 271 without alteration, 73 with a slight alteration, and 90 with
severe alteration.
e rest of paper is organised as follows: “Methodology” section presents the details of the proposed method.
“Dataset and experimental design” section describes the data considered and the characteristics of the experi-
ments, while “Results and discussion” section includes the results and the associated discussion. Finally, “Con-
clusion” section concludes the paper.
Methodology
As we stated before, the goal is to create an automatic method that classies the 3D-images into three ordered
classes corresponding to: no alteration of the presynaptic nigrostriatal pathway (both striatum were visually
conserved), with a slight alteration (just one putamen was altered), and with a severe alteration (alteration in
both). As can be seen, it results in an ordinal classication problem, given that, for example, the error when clas-
sifying a severe alternation patient as no alteration is not the same as categorising it in the slight alteration class.
An ordinal classication problem consists in the prediction of the label y of a given input vector
x
, where
x∈
X
⊆
R
S
and
y∈Y={C1,C2,...,CL}
, i.e. the input vector
x
is in a S-dimensional input space and y is in
a label space of L dierent possibilities. Given a training set of N patterns, dened as
D={(xi,y
x
i),i=1, ...,N}
,
the objective is to nd a classication rule or function
f:X→Y
to predict the categories of new patterns. It
is important to mention that a natural order is found in the labels of an ordinal classication problem, that is,
C1≺C2≺ ··· ≺ CL
, where
≺
is an order relation. It makes that two dierent elements of
Y
could always be
compared using the relation
≺
, which is not possible in nominal classication. However, labels in ordinal clas-
sication do not carry metric information (we can not establish the distance between the categories), and the
category serves more as a qualitative indication.
According to the taxonomy proposed by Gutierrez etal.45, there are three groups of ordinal classication
methods. e rst one is called naïve approaches, which derive the model by using other standard proce-
dures (such as regression or nominal classication). e second one is related to ordinal binary decomposition
approaches, where the main idea is to decompose the ordinal problem into several binary ones. And the third
one is the set of methods known as threshold models, which are based on approximating a real value predictor
and then dividing the real line into intervals.
is work is focused on the third group. Methods within this group estimate a function
f(x)
for the prediction
of the values of the output variable, and a set of thresholds
b=(b1,b2,...,bL−1)∈
R
L−1
to represent intervals
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
in the range of
f(x)
which must satisfy the constraints
b1≤b2≤ ··· ≤ bL−1
. In this context, cumulative link
models extend the binary logistic regression to ordinal classication by predicting probabilities of groups of
continuous categories taking the ordinal scale into account, that is, cumulative probabilities
P(y�Cj|x)
. In this
work, we use an interesting alternative called immediate-threshold approach46. It is based on including
L−1
thresholds partitioning the real line to L segments and on penalising the predictors outside the correct segment
or too close to its edges, considering, for each labelled example, only the two thresholds limiting this segment.
For the comparison of methods in ordinal classication we should take into account the following considera-
tions. Ordinal classication problems must be evaluated with specic metrics. At rst sight, various measures
of ordinal association and product–moment correlation and regression appear to be based on very dierent
foundations.
In this work, we have used metrics based on a product-moment system. In this system of metrics, the ones
most commonly considered in machine learning for ordinal classication are (1) the mean absolute error (here
denoted as MAE)47,48, also called classication loss49,50. e MAE is dened as the mean deviation of the predicted
class from the true class (expressed both as integers). (2) e mean zero-one error (MZOE, more oen referred
to as error rate)48, where
MZOE =1−CCR
, where CCR is the correct classication rate or accuracy.
Unlike MAE, MZOE has the disadvantage that all errors are treated equally, so it does not suciently penalise
algorithms that make errors of more than one class in the ordinal scale (e.g. severe alterations classied as no
alteration). is kind of errors are penalised by MAE. However, these measures are not adequate when used to
evaluate the performance of classiers on imbalanced ordinal datasets47.
Given that, in our work, some of the classes have a much lower number of patterns than the others (there
is a lower number of patients with alterations), we propose to use the maximum MAE metric (here denoted as
MMAE)51, which measures the performance in the worst ranked class.
In contrast to most previous works that try to obtain a high global accuracy in the test set, we try to achieve
a high level of classication rate with a good level of classication for each individual class. In this way, the
pair of metrics including CCR and MMAE evaluates two characteristics associated with a classier: the overall
performance and the average deviation of the worst classied class. As theoretically demonstrated in previous
studies51,52, these two objectives, above certain levels, are conicting during an optimisation process, i.e. increas-
ing one of the metrics can be achieved by worsening the other (e.g. we can trivially increase CCR by classifying
almost all patterns in the no-alteration class, which will drastically worsen MMAE, given the imbalance character
of the dataset). Next, we formally dene both metrics.
On the one hand, CCR is the percentage of patterns correctly classied and is dened by:
where
ˆy
x
i
is the target predicted for
xi
, and
I(y
x
i=ˆy
x
i)
is the indicator function.
On the other hand, the MAE is a measure of error that takes into account the ordinality of the target variable:
where
|y
x
i−ˆy
x
i|
is the absolute distance between the actual and predicted labels. MAE ranges from 0 to
L−1
(which is the maximum deviation in number of categories). However, in imbalanced problems, the most frequent
classes can dominate the MAE error, masking poor performance for less common classes. at is the reason why
MMAE is dened as follows:
where
MAEl
is the MAE error taking into account only patterns from class l:
e following subsections present the proposed methodology which is divided into three phases. e rst
one corresponds with the preprocessing of the 3D-images, which includes a spatial normalisation and the trans-
formation from 3D-images into 1D-arrays. e second one consists in the reduction of the number of charac-
teristics. And nally, the last one is the application of a technique of data augmentation using the probabilistic
distribution of the macrovoxels.
Preprocessing of the SPECT 3D‑images. e SPECT images can be obtained in dierent conditions,
for example, due to the inclination or rotation of the patient during the taking of the image, two 3D images may
look dierent, but they are really the same, with dierent orientation. To solve this problem, we use the open
source soware Platform for the Evaluation of Medical Imaging (PETRA)53. PETRA is a toolbox developed to
analyse neuroimaging data using multivariate techniques, specically designed to work with PET, SPECT, etc.
images with the goal of Alzheimer and Parkinson diseases diagnosis/detection. is soware has been used to
perform a spatial normalisation, i.e. to align all brains in the same spatial disposition, using the same reference
number to give the same coordinates to all voxels regardless of the brain.
(1)
CCR
=1
N
N

i=1
I(yxi=ˆyxi)
,
(2)
MAE
=1
N
N

i=1
|yxi−ˆyxi|
,
(3)
MMAE =max {MAE1,MAE2,...,MAEl,...,MAEL},
(4)
MAEl=1
Nl
N
l

i=1
|yxi−ˆyxi|,l=1, ...,L
,.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
Once the spatial normalisation is done, the 3D-images are converted into a one-dimensional array to work
with them for the next operations. Figure1 shows the procedure for a toy 3D-image of size
4×2×3
(rows
×
columns
×
layers).
Feature selection: ReliefF and proposed ordinal ReliefF. Given the large amount of voxels present in
SPECT images (see “Dataset and experimental design” section for more details on the size of the images used in
this paper), and considering that we use all image voxels instead of the ROIs, we need to reduce the number of
features using automatic machine learning techniques.
For that, as we stated in the introduction, we initially use a lter feature selector in such a way that this selec-
tion does not take into account the learning algorithm used by the classier. is lter, which is called ReliefF34,
is based on ranking the features using the quality of them according to how well their values distinguish between
instances that are near to each other. For that, the quality, Q, of each attribute, A, is initialised to
Q[A]=0
, an d
then, the following procedure is repeated m times (usually, m is set to the number of instances N):
1. Randomly select an instance, a 1D-array of the 3D-image in our case,
xi
.
2. Find the k nearest neighbours of the same class, called nearest hits:
x∗={x∗
1,
...
,x∗
k}
, where
∀j∈{1, ...,k}
,
x∗
j∈
Cx
i
and
x
i
= x∗
j
.
C
x
i
is the class which includes
xi
.
3. For each class
Cl = C
x
i
, nd the k nearest neighbours, called nearest misses:
x∗∗ ={x∗∗
1l,
...
,x∗∗
kl }
, where
∀j∈{1, ...,k}
,
x∗∗
jl ∈
C
l
.
4. For each attribute A, we update its quality Q[A] taking into account that dierent values of the attribute A
for instances in the same class will decrease the quality estimation Q[A], while dierent values for instances
in dierent classes, which is desirable, will increase the quality Q[A]. e following expression updates the
Q[A] values:
where
P(Cl)
is the prior probability of class
Cl
which is estimated from the frequency of class
Cl
in the train-
ing set, and di
(A,x1,x2)
is dened as:
where
value(A,xi)
is the value of the attribute A for the instance
xi
. is function is also used to calculate
the distance between two instances (to determine the k nearest neighbours), where the total distance is the
sum of distances over all attributes (Manhattan distance).
In this work, we propose a novel variation of the ReliefF, called ordinal ReliefF (OReliefF), which takes the
natural order of the labels of the dataset into account. e modication consists in the inclusion of an ordinal
penalty (
ρ
C
l
) in the second term of the equation:
where
ρCl
is dened as:
where
y
x
i
is the class label of the pattern
xi
,
y
C
l
is the label of the class
Cl
, and
y
C
′
l
is the class label of
C′
l
.
As can be seen, with
ρ
C
l
, OReliefF penalises less the dierences of those patterns which are in a class near to
the class of pattern
xi
(
C
x
i
), giving more importance to dierent values for instances in farthest classes.
(5)
Q[
A
]=
SQ
[
A
]− 
k
j=1diff(A,xi,x∗
j)
k+
Cl�=Cxi
P
(Cl)
1−P(Cxi)

k
j=1diff(A,xi,x∗∗
jl )

k,
(6)
diff(A,x1,x2)=|value(A,x1)−value(A,x2)|,
Q
[A]=Q[A]− 
k
j=1diff(A,xi,x∗
j)
k
+
Cl�=Cxi

P(Cl)
1−P(Cxi)ρCl

k
j=1diff(A,xi,x∗∗
jl )

k,
(7)
ρ
Cl=
|y
xi
−y
Cl
|

C′
l�=
Cx
i
|
yxi
−
yC′
l
|,
Figure1. Example of a 3D-image of size
4×2×3
reshaped to a 1D-array.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
Data augmentation. Given the diculty of obtaining SPECT 3D-images for dierent reasons such as
patient privacy, the high cost of taking images or the necessity to carry out inter-centre studies due to the dif-
ferent hardware and soware used in dierent nuclear medicine units, we develop new methods to generate
synthetic images for the training set in order to improve the quality of the classier. For this, we propose a meth-
odology based on the generation of new 3D-images using the probabilistic distribution of the macrovoxels. e
algorithm1 summarises this procedure.
As can be seen, for each informative voxel determined by the OReliefF algorithm, we select its macro-
voxel which is dened as the cube of voxels of width w where the selected voxel is in the middle. For example,
Fig.2 shows an image of dimensionality
5×5×5
, where the macrovoxel of width 3 based on voxel (2,3,2)
(
row =2, col =3, layer =2
) is the one which is coloured in blue.
Aer that, the same macrovoxel is extracted for each 3D-image i and, for each class, all the macrovoxels of
the dierent 3D-images are grouped. In this way, for each voxel, we have L groups of macrovoxels corresponding
to each class label, so each group is formed by all the voxels inside all these selected macrovoxels.
ese groups (
z
) are use to calculate the probabilistic distribution of the voxel. To do so, we consider a set of
possible probabilistic distributions. ese distributions has been selected due to the fact that they are the best-
tted ones in almost all voxel groups considering the distributions presented in the scipy library (see https://
docs. scipy. org/ doc/ scipy/ refer ence/ stats. html). ese are:
• Alpha distribution, whose probability density function (pdf) is:
where

is the normal cumulative distribution function (CDF),
µ
and
σ
are the location and scale parameters,
respectively, which are considered in all distributions, and
α
is the shape parameter.
• Generalised extreme value distribution, whose pdf is:
where
ξ
is the shape parameter.
• t-Student distribution, whose pdf is:
where
ν
is the number of freedom degrees, and
Ŵ(a)
is the gamma function dened as:
where if a is a positive integer, then
Ŵ(a)=(a−1)!
.
• Beta distribution,whose pdf is dened as:
(8)
f(z,α,µ,σ) =
1
((z−µ)/σ )
2
�(α)
√
2π
·exp(−
1
2(α −
1
((z
−
µ)/σ ) )2)
,
(9)
f(z,ξ,µ,σ) =

exp(−exp(−
z−µ
σ)) exp(−
z−µ
σ), for ξ=
0,
exp(
−
(1
−
ξz−µ
σ
)1/c)(1
−
ξz−µ
σ
)1/c
−
1, for ξ
�= 0,
(10)
f(z|ν,µ,σ) =Ŵ(ν+1
2)
Ŵ(ν
2)√πνσ

1+1
ν

z−µ
σ

2

−
ν+1
2
,
(11)
Ŵ(
a)
=∞
0
ta−1e−tdt
,
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
where a and b are the shape parameters.
en, the best-tted distribution for each voxel is determined by the minimisation of the sum of squares
errors calculated as the dierences between the theoretical distribution and the empirical one (histogram of the
values of the groups of voxels,
z
).
Finally, to create new instances, each pattern will be randomly generating using the estimated theoretical
distribution for each of the voxels of the image.
Dataset and experimental design
Dataset. As we stated previously, our work is focused on SPECT 3D-images obtained aer the administra-
tion of the
123
I-ioupane radiopharmaceutical, which is commonly used for binding to the presynaptic dopa-
mine transporters in the caudate nucleus and the putamen. e substantial decrease of dopamine in the nigros-
triatal dopaminergic pathway is one of the neuropathological characteristics of PD.
In this work, we have generated a dataset in collaboration with the UGC Medicina Nuclear of the Hospital
Universitario ”Reina Sofía” (Córdoba, Spain). e access to the images used in this study has been done aer
previous anonymization by personnel authorised by the hospital. is access (previous authentication) has been
registered and audited. e anonymization procedure has been carried out by means of DICOM Anonimyzer,
obtaining image les that do not contain any data that could identify the patient. Each series of images was
assigned a number that was related in a protected table to the origin. e entire procedure was approved and
authorized by the Center’s Healthcare Administration and the UGC Medicina Nuclear. We also arm that all
methods were carried out in accordance with relevant guidelines and regulations. Finally, all subjects provided
their informed consent for this study.
e data set consists of 434 studies divided according to the level of alteration of the presynaptic nigrostriatal
pathway in which PD is likely to be found: 271 without alteration (class 0), 73 with a slight alteration (class 1),
and 90 with severe alteration e(class 2).
Each 3D-image has dimensionality
79 ×95 ×69
, which results in 517,845 voxels. e representation of these
images can be done using dierent cuts in the cerebrum, which are axial, sagittal and coronal. Figure3 shows a
graphical explanation of this kind of cuts and their corresponding SPECT view. Following these representations,
and considering only the axial view, Fig.4 shows an example of a patient from each class.
Experimental design. In “Methodology” section, we presented a methodology based on the following
steps: pre-processing of 3D images, reduction of the number of features with ReliefF or OReliefF methodologies,
and generation of new patterns with statistical distributions of macrovoxels. For the rst step, it is not necessary
to congure any parameter since the PETRA soware automatically reorients the image in the
79 ×95 ×69
dimensional space. Our main goal is to determine whether the two methods developed for feature reduction and
data augmentation make sense in this type of problem using a real-world data set.
In this sense, for the second step, initially we randomly divide the database into a 70–30% stratied hold-out,
that is, 70% of the images are selected for training, and the rest for test (generalisation results). e resulting
training set is formed by 189 patterns of class 0, 51 patterns of class 1, and 63 of class 3. To determine if OReliefF
is better than ReliefF or the original set, we perform a 5-fold cross validation over the training set (70% of the
patterns as we mentioned before). We set the k parameter (k-nearest neighbours) of ReliefF and OReliefF to
5, the number of iterations m equal to the number of features N, and we explore dierent alternatives for the
percentage of selected characteristics in the range
{1%,
2%,
5%,
10%,
15%,
20%,
25%,
50%,
75%}
, considering it as
a hyper-parameter to be validated. Note that the experimental validation is done using the logistic regression
proposed by Rennie and Srebro46.
Once the best feature selector is determined, we use the reduced dataset to continue for the third step, where
we test the proposed method for data augmentation, based on the use of dierent statistical distributions. Firstly,
(12)
f(z
|
a,b,µ,σ)
=
Ŵ(a+b)((z−µ)/σ )
a−1
(1−((z−µ)/σ ))
b−1
Ŵ(a)Ŵ(b),
Figure2. Macrovoxel of width 3 of the voxel (2,3,2) in a 3D-image of size
5
×
5
×
5
.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
we estimated the best statistical distribution for each voxel (considering all voxels of its macrovoxel). en, we
consider dierent congurations to generate new brains for the training set:
• CONF0: Without applying data augmentation.
• CONF1: Duplicate the patterns of each class.
• CONF2: Duplicate the patterns of the minority classes 1 and 2 (subjects with slight or severe alteration).
• CONF3: Duplicate the patterns of the minority class 1 (subjects with slight alteration).
• CONF4: Duplicate the patterns of the second minority class 2 (subjects with severe alteration).
• CONF5: Triplicate the patterns of the minority class 1 (subjects with slight alteration).
ese ve congurations are compared against the simple method of noisy replication40 (referrered to as RAND
in this paper), where we randomly choose patterns from the class to be oversampled and replicate them including
a small amount of noise. For this RAND conguration, we have considered a normal distribution with mean zero
and a standard deviation of 0.01 (N(0,0.01)). Moreover, the conguration used is triplicating class 1, subjects
with slight alteration, because, as above discussed, this is the one leading to the best results.
Finally, we executed the procedure of data augmentation 30 times given the stochasticity in the procedure of
the generation, and we compared the results obtained in the test set (30% of the initial hold-out) to determine
Figure3. How are the cuts performed? (a) axial, (b) sagittal, and (c) coronal.
Figure4. Example of a patient of each class: (a) without alteration (class 0), (b) with a slight alteration (class 1),
and (c) with a severe alteration (class 2).
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
which conguration is the best in our problem. To check the results obtained, we plot informative voxels found
in the brain by our algorithm, showing that they are not limited to the classical ROIs.
Results and discussion
Results analysis. To analyse the validity of the proposed OReliefF algorithm, Table1 shows the results in
CCR and MMAE with the dierent congurations proposed in the previous section, that is, using 5-fold cross-
validation over the training set to determine the best parameter setting. ese validation results help establish
the most appropriate percentage of voxels to be considered during classication, together with the best alterna-
tive for ReliefF (nominal or ordinal), based on which, we will evaluate the test results.
As can be seen, the best feature selector in terms of CCR is the ReliefF when we maintain 10% of the features
for the classier, but its MMAE is not good. us, the second best algorithm in terms of CCR is the proposed
OReliefF when 15% of the characteristics are used for the classier, and, in addition, it is the best conguration
in terms of MMAE. We can arm that OReliefF results in a better feature selection, because the CCR only drops
from 0.7756 to 0.7723 which is practically the same value, while the MMAE improves from 0.5490 to 0.5098. If
we do not apply any feature selection the results are 0.5397 in MMAE and 0.7657 in CCR , which are worse than
the obtained by the proposed methodology.
Given the above results, we consider the reduction of the dataset with a 15% of the characteristics (informative
voxels) returned by the proposed OReliefF algorithm. In this sense, for each 3D-image, a total of 517,845 voxels
are reduced to 77,676 informative voxels. e selected voxels have been graphically represented in red in Fig.5,
together with the ROIs corresponding to caudate and putamen in blue, which are the common areas considered
for PD diagnosis. A video has also been uploaded in the website of the research group (http:// www. uco. es/ ayrna/
parki nson/ infor mativ eVoxe ls. mp4), where the image is rotated to ease visualisation. As can be checked, there are
distinct informative voxels outside the classical ROIs. ey correspond to small cortical areas, predominantly
temporal and in the medial region of the parietal lobe. ese extrastriatal areas follow a recent tendency of other
studies (e.g. magnetic resonance imaging) which show alterations in brain areas outside the striatum. is opens
up a much deeper area of investigation for future work, as these zones are dicult to identify anatomically.
e next step is the generation of new patterns according to the data augmentation procedure described in
“Data augmentation” section with the congurations shown in “Experimental design” section. We compare the
results against the noisy replication strategy (RAND strategy also described in “Experimental design” section).
It is important to mention that the generation is done using only the selected informative voxels so that the
Table 1. OReliefF versus ReliefF comparison in terms of CCR and MMAE obtained by the ordinal classier
without the application of data augmentation (5-fold cross validation results). e best results are shown in
bold. e second best results are shown in italics
ReliefF MMAE (
↓
)CCR (
↑
)
1% 0.6470 0.7426
2% 0.7255 0.7129
5% 0.6471 0.7393
10% 0.5490 0.7756
15% 0.5686 0.7591
20% 0.5238 0.7624
25% 0.5556 0.7525
50% 0.5238 0.7525
75% 0.5238 0.7690
OReliefF MMAE (
↓
)CCR (
↑
)
1% 0.5882 0.7723
2% 0.7059 0.7261
5% 0.6471 0.7558
10% 0.5882 0.7624
15% 0.5098 0.7723
20% 0.5238 0.7525
25% 0.5556 0.7558
50% 0.5556 0.7558
75% 0.5397 0.7624
No ReliefF MMAE (
↓
)CCR (
↑
)
100% 0.5397 0.7657
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
macrovoxels can have empty voxels, but it is not a problem because they provide enough information to estimate
the theoretical probability distribution.
e results in Table2 show the performance of the classier when we use the proposed data augmentation and
compare it against the noisy replication method. For the reasons presented in the “Methodology” section, CCR
and MMAE are conicting objectives. In this way, for an imbalanced ordinal classication problem, optimising
the MMAE metric, without signicantly reducing the value of CCR , should be considered as a priority. e con-
guration number 0, which corresponds to the reduced data set without the application of data augmentation,
has the best CCR but its MMAE is the worst, showing that the algorithm tend to classify all the patterns into
the majority class (subjects without alteration) to the detriment of the minority classes (subjects with slight or
severe alteration). In almost all cases, when data augmentation is applied, the classier improves the results in
MMAE showing the ability of the proposed method to give more importance to subjects with alteration without
losing CCR quality.
Looking at the dierent congurations, it seems that the generation of patterns for the classes corresponding
to subjects with alteration involves a high decrease of the MMAE from 0.6364 to 0.6212 when we double the pat-
terns and up to 0.5561 when we triple them. In addition, the CCR only decreases from 0.7710 to 0.7685, which
is negligible considering the large improvement in the MMAE metric. us, CONF5 is the best conguration
given that a much lower MMAE value,
0.5561 ±0.0192
, is obtained when compared to the other congurations,
without losing too much accuracy (CCR value is
0.7685 ±0.0036
, the second best average value). It makes sense
because class 1 (the oversampled class in this conguration, individuals with slight alteration) is the most dif-
cult one to be correctly classied. Other congurations also improve the quality of the error metric, but we
consider that it is caused by the pattern generation in class 1. Finally, when compared against the noisy replica-
tion method (RAND conguration, where we also triplicate the slight alteration class), the results in CCR are
very similar, but CONF5 is able to signicantly improve the levels of MMAE. In this way, the use of the tted
statistical distribution of the macrovoxel structures help generate new patterns that better respect the ordinal
disposition of the minority class.
In order to check whether the dierences found in Table2 could be attributed to randomness, we apply a set
of statistical tests. First of all, we need to assess the character of the results obtained regarding their distribution:
if they follow a normal distribution, we will apply parametric tests, while non-parametric tests will be considered
Figure5. Example of informative voxels outside (red points) the ROIs areas (blue points) in a patient showing
(a) axial, (b) sagittal and (c) coronal views.
Table 2. Mean and standard deviation results in terms of CCR and MMAE for the 30 executions of the data
augmentation algorithm (test set). e best results are shown in bold. e second best results are shown in
italics
Conguration MMAE (
↓
)CCR (
↑
)
CONF0 0.6364 0.7710
CONF1
0.6303 ±0.0155
0.7588 ±0.0054
CONF2
0.6349 ±0.0082
0.7575 ±0.0051
CONF3
0.6212 ±0.0214
0.7646 ±0.0053
CONF4
0.6364 ±0.0000
0.7557 ±0.0000
CONF5 0.5561 ± 0.0192 0.7685 ± 0.0036
RAND 0.5958 ± 0.0208 0.7613 ± 0.0048
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
in other case. We discard the result for CONF0, given that, as only one test value is available, it does not make
sense to apply statistical tests, the MMAE being clearly worse than those obtained by data augmentation.
First, we analyse the MMAE results. One Kolmogorov–Smirnov (K–S) test54 is used for each set of 30 results,
rejecting all the hypotheses of normality (p values
<0.001
), i.e. non-parametric tests need to be applied. Conse-
quently, we select the Kruskal–Wallis (K–W) test55, which is a non-parametric alternative for the one-way analysis
of variance, when we want to compare more than two independent samples. e results of this test are included
in Table3. is table includes the average ranks obtained for each method, where the lower the rank, the better
the method (as MMAE has to be minimised). It also includes the associated
χ2
and the corresponding p value,
showing that there are signicant dierences in test MMAE, where the two best congurations (according to the
ranks) are CONF5 and RAND. ese two congurations obtain the best results because both are using the option
of triplicating patterns from class 1 (slight alteration), which seems to be the most problematic in this problem
(easily confused with neighbouring classes). In order to establish the signicance of the dierences between
CONF5 and RAND for test MMAE results, we nally apply the Mann–Whitney (M–W) test56, a post-hoc non-
parametric test for independent samples, which results in a p value
<0.001
, i.e. there are signicant dierences
favouring CONF5 with respect to RAND.
If we repeat the same statistical treatment with the test CCR results, the K–S test again rejects the hypotheses
of normality for all the results (p values
<0.001
). e results of the M–W test for the test CCR values are also
included in Table3, showing that there are signicant dierences and that the two best congurations are, in
this case, CONF5 and CONF3. e M–W establishes again signicant dierences favouring CONF5, with a p
value of 0.004.
Summarising, CONF5 applied to the data augmentation process proposed in this paper stands out as a new
way to generate synthetic brain patterns that help the classication of the most confusing class (slight alteration).
It can be a decision support system for medical purposes where, not only the global correct classication of
patients is important, but also a good accuracy in the confusing class corresponding to slight level of alteration
of the presynaptic pathway. In this way, this work suggests a new option to approach at PD, with good values for
diagnosis accuracy and classication in the dierent disease stages.
Conclusions
In this paper, we propose a new methodology for classifying dierent stages of PD using an ordinal classier
from 3D-images obtained aer the administration of 123I-ioupane in three classes depending on the level of
aectation of the image. e classes corresponds to patients without alteration of the presynaptic nigrostriatal
pathway, with a slight alteration, or with a severe alteration. is methodology is based on an ordinal logistic
regression model where we have previously executed an ordinal lter algorithm to select the most discriminant
voxels and a novel data augmentation technique based on probability distributions associated with samples
formed by macrovoxels to classify better the classes corresponding to subjects with alteration. In this dicult
problem, where it is sometimes necessary the concurrence of several experts to determine the value of the label
associated with the 3D-image, we have managed to decrease the metric related to the good classication of the
intermediate class, the most dicult to diagnose, by 12.62%, still obtaining a very acceptable overall classica-
tion, an average of 76.85% and a standard deviation of 0.36%. It is important to note that both the ordinal feature
selection method and the data augmentation method could be applied to other datasets and could be coupled
with other classiers. Finally, we would like to point out that the new database on which we have carried out the
validation is very interesting for this type of study, and we will make it available to the scientic community in
order to carry out experiments with it and make comparisons with our methodology.
Regarding the limitations of the present study, the size of the dataset could still be increased, by including
more subjects, what would better assess the generalisation capability of the proposals. Another problem associ-
ated with the method proposed lies in the fact that the computational cost of feature selection process is high,
given that all the areas of the brain are considered for classication (although this enables the classier to nd
new informative voxels, not previously considered in the literature). Given that a denitive diagnosis is not always
available, the models depend on a clinical evaluation (from the labels in the dataset), which unavoidably includes
certain degree of subjectivity and can inuence the training process (i.e. classication accuracy may be aected
Table 3. Kruskal–Wallis statistical test results for CCR and MMAE (test set). e best results are shown in
bold. e second best results are shown in italics
Conguration
Average ranks
MMAE(
↓
)CCR (
↑
)
CONF1 115.17 71.60
CONF2 122.92 59.93
CONF3 99.67
120.25
CONF4 125.50 47.00
CONF5
20.78
146.50
RAND
58.97
97.72
χ2
129.33 95.38
p value
<0.001
<0.001
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
by variations in clinical condence). Another limitation may be derived from the main clinical objective: the
automatic separation between control subjects and parkinsonian patients with dierent degrees of aectation.
However, the most challenging objective in PD diagnosis is obtaining a method able to dierentiate between
the dierent parkinsonian syndromes.
Received: 15 September 2020; Accepted: 16 March 2021
References
1. Armstrong, M. J. & Okun, M. S. Diagnosis and treatment of Parkinson disease: A review. JAMA 323, 548–560 (2020).
2. Olanow, C. W., Stern, M. B. & Sethi, K. e scientic and clinical basis for the treatment of Parkinson disease (2009). Neurology
72, S1–S136 (2009).
3. Booij, J. et al. Imaging of dopamine transporters with iodine-123-fp-cit spect in healthy controls and patients with Parkinson’s
disease. J. Nucl. Med. 39, 1879 (1998).
4. Son, S.-J., Kim, M. & Park, H. Imaging analysis of Parkinson’s disease patients using spect and tractography. Sci. Rep. 6, 38070
(2016).
5. Booij, J. et al. [123i] fp-cit spect shows a pronounced decline of striatal dopamine transporter labelling in early and advanced
Parkinson’s disease. J. Neurol. Neurosurg. Psychiatry 62, 133–140 (1997).
6. Darcourt, J. et al. Eanm procedure guidelines for brain neurotransmission spect using 123i-labelled dopamine transporter ligands,
version 2. Eur. J. Nucl. Med. Mol. Imaging 37, 443–450 (2010).
7. Benamer, H. T. et al. Accurate dierentiation of parkinsonism and essential tremor using visual assessment of [123i]-fp-cit spect
imaging: e [123i]-fp-cit study group. Mov. Disord. 15, 503–510 (2000).
8. Kas, A. et al. Validation of a standardized normalization template for statistical parametric mapping analysis of 123i-fp-cit images.
J. Nucl. Med. 48, 1459–1467 (2007).
9. Antonini, A. et al. Complementary pet studies of striatal neuronal function in the dierential diagnosis between multiple system
atrophy and Parkinson’s disease. Brain J. Neurol. 120, 2187–2195 (1997).
10. Hamilton, D., List, A., Butler, T., Hogg, S. & Cawley, M. Discrimination between parkinsonian syndrome and essential tremor
using articial neural network classication of quantied datscan data. Nucl. Med. Commun. 27, 939–944 (2006).
11. Zhang, Y. C. & Kagen, A. C. Machine learning interface for medical image analysis. J. Digit. Imaging 30, 615–621 (2017).
12. Segovia, F., Górriz, J. M., Ramírez, J., Martínez-Murcia, F. J. & Castillo-Barnes, D. Assisted diagnosis of parkinsonism based on
the striatal morphology. Int. J. Neural Syst. 29, 1950011 (2019).
13. Prashanth, R., Roy, S. D., Mandal, P. K. & Ghosh, S. Automatic classication and prediction models for early Parkinson’s disease
diagnosis from spect imaging. Expert Syst. Appl. 41, 3333–3342 (2014).
14. Strafella, A. P. et al. Molecular imaging to track Parkinson’s disease and atypical Parkinsonisms: New imaging frontiers. Mov.
Disord. 32, 181–192 (2017).
15. Yang, C. et al. Brain-region specic metabolic abnormalities in Parkinson’s disease and levodopa-induced dyskinesia. Front. Aging
Neurosci. 12, 75 (2020).
16. Perlaki, G. et al. Validation of an automated morphological mri-based 123 i-fp-cit spect evaluation method. Parkinsonism Relat.
Disord. 29, 24–29 (2016).
17. Anderson, J. A. Regression and ordered categorical variables. J. R. Stat. Soc. Ser. B Methodol. 46, 1–30 (1984).
18. Jang, W. M., Eun, S. J., Lee, C. E. & Kim, Y. Eect of repeated public releases on cesarean section rates. J. Prev. Med. Public Health
44, 2–8 (2011).
19. Cardoso, J. S., da Costa, J. F. P. & Cardoso, M. J. Modelling ordinal relations with SVMS: An application to objective aesthetic
evaluation of breast cancer conservative treatment. Neural Netw. 18, 808–817 (2005).
20. Pérez-Ortiz, M. et al. An organ allocation system for liver transplantation based on ordinal regression. Appl. So Comput. 14,
88–98 (2014).
21. Doyle, O. M. et al. Predicting progression of Alzheimer’s disease using ordinal regression. PLoS ONE 9, e105542 (2014).
22. Tarakad, A. & Jankovic, J. Diagnosis and management of Parkinson’s disease. Semin. Neurol. 37, 118–126. https:// doi. org/ 10.
1055/s- 0037- 16018 88 (2017).
23. Martinez-Martin, P. & R ay Chaudhuri, K. Comprehensive grading of Parkinson’s disease using motor and non-motor assessments:
Addressing a key unmet need. Expert Rev. Neurother. 18, 41–50 (2018).
24. Pasquini, J. et al. Clinical implications of early caudate dysfunction in Parkinson’s disease. J. Neurol. Neurosurg. Psychiatry 90,
1098–1104. https:// doi. org/ 10. 1136/ jnnp- 2018- 320157 (2019).
25. Martinez-Martin, P. et al. International study on the psychometric attributes of the non-motor symptoms scale in Parkinson
disease. Neurology 73, 1584–1591 (2009).
26. Chaudhuri, K. R . et al. e metric properties of a novel non-motor symptoms scale for Parkinson’s disease: Results from an inter-
national pilot study. Mov. Disord. 22, 1901–1911 (2007).
27. Perlaki, G. et al. Validation of an automated morphological mri-based 123i-fp-cit spect evaluation method. Parkinsonism Relat.
Disord. 29, 24–29 (2016).
28. Yang, J. & Ma, J. Compressive sensing-enhanced feature selection and its application in travel mode choice prediction. Appl. So
Comput. 75, 537–547 (2019).
29. Hosseini, E. S. & Moattar, M. H. Evolutionary feature subsets selection based on interaction information for high dimensional
imbalanced data classication. Appl. So Comput. 82, 105581 (2019).
30. Kohavi, R. & John, G. H. Wrappers for feature subset selection. Artif. Intell. 97, 273–324 (1997).
31. Duda, R. O., Hart, P. E. & Stork, D. G. Pattern Classication (Wiley, New York, 2012).
32. Argyriou, A., Evgeniou, T. & Pontil, M. Multi-task feature learning. Adv. Neural Inf. Process. Syst. 19, 41–48 (2007).
33. Kononenko, I. Estimating attributes: Analysis and extensions of relief. In European Conference on Machine Learning 171–182
(Springer, 1994).
34. Robnik-Šikonja, M. & Kononenko, I. eoretical and empirical analysis of relie and rrelie. Mach. Learn. 53, 23–69 (2003).
35. Ciresan, D.C., Meier, U. & Schmidhuber, J. Multi-column deep neural networks for image classication. CoRR abs/1202.2745
(2012).
36. Simard, P.Y., Steinkraus, D. & Platt, J.C. Best practices for convolutional neural networks applied to visual document analysis. In
Null, 958 (IEEE, 2003).
37. Jiang, X., Pang, Y., Li, X., Pan, J. & Xie, Y. Deep neural networks with elastic rectied linear units for object recognition. Neuro-
computing 275, 1132–1139 (2018).
38. Deng, J. etal. Imagenet: A large-scale hierarchical image database. In IEEE Conference on Computer Vision and Pattern Recognition.
CVPR 2009 248–255 (IEEE, 2009).
39. Krizhevsky, A., Sutskever, I. & Hinton, G. E. Imagenet classication with deep convolutional neural networks. Adv. Neural Inf.
Process. Syst. 25, 1097–1105 (2012).
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2021) 11:7067 | 
www.nature.com/scientificreports/
40. Lee, S. S. Noisy replication in skewed binary classication. Comput. Stat. Data Anal. 34, 165–191 (2000).
41. Kaur, H., Malhi, A. K. & Pannu, H. S. Machine learning ensemble for neurological disorders. Neural Comput. Appl. 32, 12697–12714
(2020).
42. Canturk, I. Fuzzy recurrence plot-based analysis of dynamic and static spiral tests of Parkinson’s disease patients. Neural Comput.
Appl. 33, 349–360 (2020).
43. Naseer, A. et al. Rening Parkinson’s neurological disorder identication through deep transfer learning. Neural Comput. Appl.
32, 839–854 (2020).
44. Pedregosa-Izquierdo, F. Feature extraction and supervised learning on fMRI: From practice to theory. eses, Université Pierre et
Marie Curie - Paris VI (2015).
45. Gutierrez, P. A., Perez-Ortiz, M., Sanchez-Monedero, J., Fernandez-Navarro, F. & Hervas-Martinez, C. Ordinal regression methods:
Survey and experimental study. IEEE Trans. Knowl. Data Eng. 28, 127–146 (2016).
46. Rennie, J.D. & Srebro, N. Loss functions for preference levels: Regression with discrete ordered labels. In Proceedings of the IJCAI
multidisciplinary workshop on advances in preference handling, 180–186 (Kluwer Norwell, MA, 2005).
47. Baccianella, S., Esuli, A. & Sebastiani, F. Evaluation measures for ordinal regression. In 2009 Ninth international conference on
intelligent systems design and applications, 283–287 (IEEE, 2009).
48. Dembczyński, K., Kotłowski, W. & Słowiński, R. Ordinal classication with decision rules. In International Workshop on Mining
Complex Data, 169–181 (Springer, 2007).
49. Basilico, J. & Hofmann, T. Unifying collaborative and content-based ltering. In Proceedings of the 21st International Conference
on Machine Learning, vol. 9 (2004).
50. Chu, W. & Keerthi, S.S. New approaches to support vector ordinal regression. In Proceedings of the 22nd International Conference
on Machine Learning, 145–152 (2005).
51. Cruz-R amírez, M., Hervás-Martínez, C., Sánchez-Monedero, J. & Gutiérrez, P. A. Metrics to guide a multi-objective evolutionary
algorithm for ordinal classication. Neurocomputing 135, 21–31 (2014).
52. Caballero, J. C. F., Martínez, F. J., Hervás, C. & Gutiérrez, P. A. Sensitivity versus accuracy in multiclass problems using memetic
pareto evolutionary neural networks. IEEE Trans. Neural Netw. 21, 750–770 (2010).
53. S egovia, F. et al. Petra: Multivariate analyses for neuroimaging data. In IWBBIO, 1302–1312 (2014).
54. Massey, F. J. Jr. e Kolmogorov–Smirnov test for goodness of t. J. Am. Stat. Assoc. 46, 68–78 (1951).
55. McKight, P. E. & Najab, J. Kruskal–Wallis test. In e Corsini Encyclopedia of Psychology 1–1 (2010).
56. McKnight, P. E. & Najab, J. Mann–Whitney u test. In e Corsini Encyclopedia of Psychology, 1–1 (2010).
Acknowledgements
is research has been partially supported by the “Ministerio de Economía, Industria y Competitividad” of Spain
(Ref. TIN2017-85887-C2-1-P) and the “Fondo Europeo de Desarrollo Regional (FEDER) y de la Consejería de
Economía, Conocimiento, Empresas y Universidad” of the “Junta de Andalucía” (Spain) (Ref. UCO-1261651).
Author contributions
A.M.D.R and J.C.C. processed the experimental data; P.A., J.A.V.C. and C.H.M. were involved in planning and
supervised the work, A.M.D.R. performed the analysis, wrote the manuscript and designed the gures; V.G.M.
and E.R.C. extracted the 3D-images obtained aer the administration of
123
I-ioupane. All authors reviewed
the manuscript.
Competing interests
e authors declare no competing interests.
Additional information
Correspondence and requests for materials should be addressed to A.M.D.-R.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2021
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com