A Novel Functional MRI-based Immersive Tool for
Dementia Disease Severity Prediction via Spectral
Clustering and Incremental Learning
Wei Huang
School of Information Engineering
Nanchang University
Nanchang, China 330031
Email: huangwei@ncu.edu.cn
Peng Zhang
School of Computer Science
Northwestern Polytechnical University
Xi’an, China 710129
Email: zh0036ng@nwpu.edu.cn
Guang Chen
Xian Communications Institute
Xi’an, China 710072
Email: chenguang322@gmail.com
Abstract—An image-based immersive tool is proposed to diag-
nose the severity of the dementia disease based on arterial spin
labeling images for the first time in this paper. A Gaussian form
of mahalanobis distance is incorporated to measure the pair-wise
similarity of images, and its with-in full matrix is automatically
determined via spectral clustering and incremental learning tech-
niques. There are totally 350 demented patients belonging to three
groups, i.e., Alzheimer’s disease, mild cognitive impairment, and
non cognitive impairment, utilized for experimental evaluation.
Several other methods are also implemented and compared with
the newly proposed method from the statistical perspective. It
is concluded that the new method outperforms others in this
dementia diagnosis study.
Keywords– immersive tool, spectral clustering, incremental
learning, dementia.
I. INTRODUCTION
Dementia disease is often considered as a wide-range type
of brain diseases, which may result in gradual decrease
or even long-term impairment of cognitive capabilities (i.e.,
short-memory, long-memory, logic analysis, etc) of human
beings. According to the statistics conducted by the World
Health Organization, the dementia disease is often diagnosed
worldwide in patients who are over 60 years old, and is now
one of the five most severe non-communicable diseases in the
whole world (i.e. others include cardiovascular disease, cancer,
diabetes and chronic lung disease) [1]. According to another
population study conducted by the United Nations, there are
more than 26.6 million demented patients diagnosed globally
[2], and 1 in 85 worldwide people is predicted to be suffering
from the dementia disease by the year 2050 [3]. In China,
there are over 10 million demented patients by the year 2012
[4]. It can be summarized from the above statistics that, the
dementia disease becomes an actual threat in many countries
of aging societies nowadays, and accurate diagnosis as well
as timely treatment are often essential to delay the onset and
progression of the dementia disease.
Therefore, identifying the progression of the dementia dis-
ease into various stages precisely is often of great importance
to understand mechanisms of the disease, making correct treat-
ments to corresponding symptoms of the disease possible at a
later stage [3]. In order to accurately perceive the progression
of the dementia disease in clinical diagnosis, a variety of
methods have been proposed and utilized to date, including
pathography analysis, cognitive examination, brain scanning,
etc. Generally speaking, pathography is helpful to predict
curable symptoms of demented patients who may usually
suffer from other types of diseases (e.g., stroke, heart disease,
renal failure, etc) at the same time [5]. Cognitive examination
evaluates the progression of demented patients through a series
of cognition tests based on diverse cognitive capabilities of
patients, and popular cognition examinations include mini-
mental state examination (MMSE) [6], Addenbrooke’s cog-
nitive examination (ACE) [7], etc. Although these cognitive
exams require few trainings for clinicians and are relatively
easy to be carried out by them, outcomes of those exams could
be highly biased by patients specialities. For example, patients
of high-level education suffering from the dementia disease
are more likely to outperform ordinary people of low-level
education without dementia disease in those cognitive exams.
For brain scanning, it is now accepted as an effective and
affordable way in the dementia disease diagnosis. There are
several imaging tools incorporated for the dementia diagno-
sis, including computed tomography (CT), positron emission
tomography (PET), magnetic resonance imaging (MRI), etc.
Among them, MRI receives vast popularity because of its
prominent capability in both generating high-resolution im-
ages of brain tissues and free of ionizing radiation exposure,
compared with other scanning tools such as CT for patients’
safety considerations. Also, most contemporary MRI scanning
techniques can be categorized into anatomic MRI (aMRI) and
functional MRI (fMRI), and both of them have already been
incorporated in various dementia studies nowadays [8], [9].
In this study, a novel fMRI-based immersive tool using
spectral clustering and incremental learning techniques is
introduced for the first time, to handle the dementia disease
diagnosis problem. The structure of this paper is described
as follows. First, a recently popular fMRI scanning tool, i.e.,
arterial spin labeling (ASL), is introduced in Section II. The
basic idea of ASL as well as its generated ASL images are