Detection of Alzheimer’s disease using Deep 2024-

learning 25

ii
Detection of Alzheimer’s disease using Deep 2024-
learning 25
CHAPTER-1

INTRODUCTION
A person with Alzheimer's Disease (AD) suffers from a developing neurobiological
disorder that affects brain cells to die & atrophy. AD is the prevalent cause of dementia, which
causes memory loss and impaired reasoning abilities. Around 6 million people in the U.S of age
65 and above are caused with Alzheimer's disease. Of those 80 percent are above 75 years old. In
India it is calculated around 5.3 million people live with dementia of which Alzheimer is the
common cause.

Dementia with Alzheimer's is classified into four categories:

1. Very Mild Dementia: Individuals suffers from memory loss as they age.

2. Mild Dementia: Symptoms which includes lack of memory, Behavioral changes,

inability to perform routine tasks.

3. Moderate Dementia: The day to day life becomes complex for the individuals with
moderate dementia, where the patients require extra care and support.

4. Severe Dementia: The patients in this stage may not able to communicate properly, and
they require medical care. One may lose physical control.

Alzheimer's is not a curable disease, but an early diagnosis can help prevent the patient
from suffering from the later stages. In order to diagnose AD manual detection systems for
example: Positron Emission Tomography (PET) were used to track the progression of the various
stages of AD, MRI scans and genotype sequencing results were taken for diagnosis.

Among the most popular fields of research in recent years has been brain-computer
interface (BCI), thanks to its applications in areas such as brain fingerprinting, detecting
neurological illnesses, tiredness, adaptive e-learning and more. By extracting the most significant
characteristics, BCI creates an effective link to interact between the brain and the device. A
complex brain structure varying with age and pathology makes it very difficult to detect
neurodegenerative diseases in their early stages.
Computer-assisted techniques are more successful in detecting these disorders than

Dept.Of 1
CSE,GMIT,Bharathinagara
Detection of Alzheimer’s disease using Deep 2024-
learning 25
traditional approaches. A timely diagnosis and identifying of Alzheimer's disease is essential to
reducing medical expenses, improving treatment, and preventing brain cell degeneration.

Some of the methods included in an early examination include Positron Emission

Tomography (PET), digital imaging methods and genotyping-by-sequencing. It is difficult to
take decisions by analyzing different methods. Furthermore, the patients wi ll have to undergo
radioactive effects during PET medical procedure. According to our findings, MRIs can provide
valuable information on the brain, because they provide flexible imaging, superior tissue
contrast, and do not expose the brain to ionizing radiation.

It is crucial to develop a model that can take MR images as input and detect whether
patients are normal or not. By utilizing a dataset, machine learning can extract knowledge.
Computer science, artificial intelligence, and statistics combine to make up this field.. The ML is
done through training a computer to produce the output based on its past experience to solve a
given problem. Machine learning can be applied in a variety of fields in order to solve problems
quicker than humans, and therefore be more efficient, and reduce time spent on repetitive tasks.
Nowadays, because of the reduction in the cost of computing power and memory. This allows
processing and analyzing huge amounts of data to generate insights.

Additionally, Deep Learning is a subset of ML and an advanced mode of analyzing and

learning information from raw data that computers are able to replicate, much like how humans
are able to do, with a computer. Deep learning is becoming increasingly popular for diagnosing
diseases. Several Machine learning approaches have been proposed recently to aid in this
diagnosis, providing doctors with more information to make informed decisions.

We use Deep Neural Networks (DNNs) for feature extraction using deep learning
techniques in the proposed model. A solution to underfitting is to use sampling techniques
especially oversampling to resolve class imbalance. DL performs classification on given MR
images using the cortical surface of the brain as input.

Using a dementia-specific(Alzheimer’s) dataset, the models are evaluated by

NonDemeneted, Moderate Dementia, Mild Dementia and Very Mild Dementia obtained from
Kaggle. Utilizing the CNN technique, we extract discriminating features for AD classification by

Dept.Of 2
CSE,GMIT,Bharathinagara
Detection of Alzheimer’s disease using Deep 2024-
learning 25
improving the accuracy. By using this model, we can accurately classify the stages of AD.

1.1 Overview:

Dementia is a term for a decline in mental ability severe enough to interfere with daily life.
Alzheimer's is the one type of dementia. Alzheimer’s is the most serious yet common
neurodegenerative disease that initially destroys cells of the part of the brain. It’s responsible for
language and memory resulting in memory loss of the patient and also the ability to perform
regular tasks.

As there is no cure for Alzheimer’s disease, it’s better to detect as early as possible to
slow down the severity of the disease. Usually to diagnose the disease radiologists use manual
methods such as previous medical history, continuous monitoring of the patient to detect the
various stages of AD, however these manual methods may lead to errors!

1.2 Objective:
Implementing Deep Learning algorithm efficiently to identify the stage of the
Alzheimer’s disease patient. Analyzing the various performance metrics of the deep learning
algorithm.

Dept.Of 3
CSE,GMIT,Bharathinagara
Detection of Alzheimer’s disease using Deep 2024-
learning 25

CHAPTER-2

LITERATURE SURVEY

Researchers have made several attempts to classify Alzheimer's patients using the
brain structure information obtained from MR images. Tested the effectiveness of a method
using brain shape information for classification of healthy subjects and Alzheimer’s disease
patients. Here, the shape information of the brain, particularly that of the lateral ventricles,
was analyzed without considering the septum lucidum.[1].

The neurodegenerative disease is currently the 6th leading source of death in the
US. In 2017 this disease costed the nation $1.1 trillion. This study have shown that deep
CNNs are capable of detecting Alzheimer's disease and dementia based on 3D MRI imaging
and can be used to detect the disease. As of the this papers writing, detecting Alzheimer's is
a difficult and time consuming task, but requires brain imaging report and human expertise.
Needless to say, this conventional approach to detect Alzheimer's is costly and often error
prone. In this paper an alternative approach has been discussed, that is fast, costs less and
more reliable. Deep Learning represents the true bleeding edge of Machine Intelligence.
Convolutional Neural Networks are biologically inspired Multilayer perceptron specially
capable of image processing[2].

Neelaveni J et al. [3] used machine-learning algorithms, they used psychological

factors to predict Alzheimer's, such as age, MMSE scores, and educational level. Diagnosis
of the disease is done but that too at the later stage only. If the disease can be predicted
earlier, symptoms or progression can be slowed down.

They demonstrate how whole-brain volume can be automatically calculated and

estimated spinal fluid volume used for determining differences associated with ideal aging
and Alzheimer's disease[4].

It is well-established, widely used, and non-invasive to use structural MR imaging

in Alzheimer's disease studies of downstream effect of the neurodegeneration, or atrophy. In
peer-reviewed journals, the imaging and cohort results have been published in detail, and
the data have been made freely available[5].Zhao Fan et al. [6], help aid auxiliary

Dept.Of 4
CSE,GMIT,Bharathinagara
Detection of Alzheimer’s disease using Deep 2024-
learning 25
diagnosis of

Dept.Of 5
CSE,GMIT,Bharathinagara
Detection of Alzheimer’s disease using Deep 2024-
learning 25
Alzheimer's disease, with the help of SVM model to analyze and classify structural brain
MR imaging data, in order to achieve much greater classification predictions, the extracted
MRI detail is combined with the SVM model. The accuracy of classification and prediction
is the best. According to the predicted results, the data characteristics related to diseases can
be determined, which can provide a basis for clinical and basic research, etiology and
pathological changes.

Acharya et al, used several feature extraction techniques along with MR imaging for
classification, the system uses several algorithms for Computer- Assisted-BrainDiagnosis
(CABD) to assess if the scan image shows indications of Alzheimer's disease. The
paradigm consists of a series of quantitative techniques: filtering, feature extraction,
Student’s t-test based feature selection, and k-Nearest Neighbor (KNN) based classification.
Additionally, a comparAative analysis is done by implementing other feature extraction
procedures that are described in the literature. According to their results, Shearlet Transform
(ST) feature extraction is superior to alternative methods for detecting Alzheimer's
disease[7] Chima Stanley Eke et al. [8] the main basis for their method is machine
learning(ML) techniques (specifically support vector machines), which are capable of
learning patterns from complex data in order to create multi-variable models. The objective
in this study is to develop a method to identify potential blood-based non-amyloid
biomarkers for early AD detection. The use of blood is attractive because it is accessible and
relatively inexpensive.

R. Jain et al. [9] propsed convolutional neural network model (CNNs) were used to
locate gray matter in brain voxels and to segment segmented gray matter for clinical
appraisal. VGG-16 trained on ImageNet dataset is used as a feature extractor for the
classification task. Experimentation is performed on data collected from Alzheimer’s
Disease Neuroimaging Initiative (ADNI) database. The proposed methodology achieved an
accuracy rate of 90.47% for segments of gray matter for clinical evaluation. Chenjie Ge et
al.
[10] the paper showcases, an innovative idea to detect Alzheimer's disease (AD) using 3D
multiscale neural network architecture. They proposed a multiscale deep learning
architecture for learning AD features. The main contributions of the paper include,
proposing a novel 3D multiscale CNN architecture for the dedicated task of AD detection,

Dept.Of 6
CSE,GMIT,Bharathinagara
Detection of Alzheimer’s disease using Deep 2024-
learning 25
secondly proposing a feature fusion and enhancement strategy for multiscale features and at
last empirical study on the impact of several settings, including two dataset partitioning

Dept.Of 7
CSE,GMIT,Bharathinagara
Detection of Alzheimer’s disease using Deep 2024-
learning 25
approaches, and the use of multiscale and feature enhancement. It successfully achieved an
accuracy of 87.24% for detection.

Chenjie Ge et al. [10] the paper showcases, an innovative idea to detect Alzheimer's
disease (AD) using 3D multiscale neural network architecture. They proposed a multiscale
deep learning architecture for learning AD features. The main contributions of the paper
include, proposing a novel 3D multiscale CNN architecture for the dedicated task of AD
detection, secondly proposing a feature fusion and enhancement strategy for multiscale
features and at last empirical study on the impact of several settings, including two dataset
partitioning approaches, and the use of multiscale and feature enhancement. It successfully
achieved an accuracy of 87.24% for detection.

Danail Stoyanov et al. [11] used brain structural MRI scans, they developed a 3D
CNN to identify Alzheimer's disease. They highlighted relevant areas of the network using
four distinct ways to visualize gradients and occlusions. Their neural network model
achieved 77% accuracy.

Dept.Of 8
CSE,GMIT,Bharathinagara
Detection of Alzheimer’s disease using Deep 2024-
learning 25
CHAPTER-3

AIM AND SCOPE OF PRESENT INVESTIGATION

3.1 AIM

This project aims to detect the stages of the Alzheimer’s disease by using deep learning

techniques.

3.2 SCOPE OF THE PRESENT INVESTIGATION:

Alzheimer's disease is a progressive neurological disorder that causes the brain to shrink
(atrophy) and brain cells to die. Alzheimer's disease is the most common cause of dementia.
As the disease progresses, a person with Alzheimer's disease will develop severe memory
impairment and lose the ability to carry out everyday tasks. A number of medicines may be
prescribed for Alzheimer's disease to help temporarily improve some symptoms but there is no
permanent cure. An accurate, timely diagnosis gives you the best chance to adjust, prepare and
plan for the future, as well as access to treatments and support that may help. An early
detection can give a patient a better chance to cure and recover.

It is crucial to develop a model that can take MR images as input and detect whether
patients are normal or not. Computer science, artificial intelligence, and statistics combine to
make up this field. been applied in the healthcare sector. Deep Learning is a subset of ML and
an advanced mode of analyzing and learning information from raw data that computers are
able to replicate, much like how humans are able to do, with a computer. Deep learning is
becoming increasingly popular for diagnosing diseases. We use Deep Neural Networks
(DNNs) for feature extraction using deep learning techniques in the proposed model. A solution
to underfitting is to use sampling techniques especially oversampling to resolve class
imbalance. DL performs classification on given MR images using the cortical surface of the
brain as input.

3.3 EXISTING SYSTEM:

Early detection of this disorder is being researched to slow down the abnormal
degeneration of the brain, reduce medical care cost reduction, and ensure improved treatment.

Dept.Of 9
CSE,GMIT,Bharathinagara
Detection of Alzheimer’s disease using Deep 2024-
learning 25
In Existing system machine learning algorithms are used to predict the Alzheimer disease using

Dept.Of 1
CSE,GMIT,Bharathinagara 0
Detection of Alzheimer’s disease using Deep 2024-
learning 25
psychological parameters like age, number of visit, MMSE and education.Different modalities
are used for AD study include MRI, Positron Emission Tomography (PET), and genotype
sequencing results. It is time-consuming to analyze different modalities to take a decision.

Furthermore, the patients can encounter radioactive effects in the modalities like PET.
Previously researchers performed 3D tissue segmentation of white matter, gray matter, and
cerebrospinal fluid from MR images after skull striping using FSL tool, calculate the surface
fractal dimension from segmented brain tissue.

From the survey , Numerous techniques exist for AD classification using machine and
deep learning. However, the high model parameter and class imbalance in the multiclass AD
classification is still an issue.

3.3.1 DISADVANTAGES:

 The existing model shows significant accuracy only when MMSE score, education, etc.,
is given.

 The patients can encounter radioactive effects in the modalities like PET.

 3D MRI scan is hard to train and time consuming process.

3.4 PROPOSED SYSTEM:

It is considered important to develop a better computer-aided diagnostic system that can

interpret MRI imaging and determine whether patients are healthy or have Alzheimer’s
disease. Conventional deep learning systems use the cortical surface to input the CNN to
perform AD classification on raw MRI images. In this proposed work, We believe that the MRI
modality benefits from its greater imaging flexibility, excellent tissue contrast, lack of ionizing
radiation, and ability to provide useful information on human brain anatomy.

This paper proposes a model that uses the convolutional neural network to extract the
discriminative features. Class imbalance is addressed using the Synthetic Minority
Oversampling Technique (SMOTE) technique. The model is developed from scratch to classify
the stages of AD more accurately by reducing its parameters and computation cost. The models

Dept.Of 1
CSE,GMIT,Bharathinagara 1
Detection of Alzheimer’s disease using Deep 2024-
learning 25
are evaluated by training them over the MRI dataset from the Kaggle. The dataset comprises
four types of dementia such as Mild Dementia (MID), Moderate Dementia (MOD), Non-
Demented (ND) and Very Mild Dementia (VMD).

A new convolutional neural network architecture is proposed with relatively small

parameters to detect the types of dementia which is suitable for training a smaller
datasetSMOTE technique is used to address the class imbalance problem in the dataset is by
randomly duplicating the minority class of images in the dataset to minimize the overfitting
problem.We created the generalized model that learns from the smaller dataset with reduced
parameters and computation cost, which still performs better for AD diagnosis.We also
compared the proposed model with deep features and hand-crafted features to detect AD stages
in terms of Accuracy, AUC and Cohen’s kappa score.

3.4.1 ADVANTAGES:

 A neural network architecture is proposed with relatively small parameters to detect the
types of dementia which is suitable for training a smaller dataset.

 SMOTE technique is used to address the class imbalance problem.

 We created the generalized model that learns from the smaller dataset with reduced
parameters and computation cost.

Dept.Of 1
CSE,GMIT,Bharathinagara 2
 Detection of Alzheimer’s disease using Deep 2024-
learning 25
CHAPTER 4

EXPERIMENTAL OR MATERIALS AND METHODS; ALGORITH USED

4.1 Deep Learning Overview:

Deep Learning is a subfield of machine learning concerned with algorithms inspired by the
structure and function of the brain called artificial neural networks. It is essentially a neural
network with three or more layers, these neural networks attempts to mimic the human brain
through a combination of data inputs, weights, and bias, allowing it to learn from large amounts
of data. These elements work together to accurately recognize, classify, and describe objects
within the data.

There are different types of neural networks to address specific problems or datasets. For
example, Convolutional neural networks (CNNs), used primarily in computer vision and image
classification applications, can detect features and patterns within an image, enabling tasks,
like object detection or recognition. Recurrent neural network (RNNs) are typically used in
natural language and speech recognition applications as it leverages sequential or times series
data.

4.2 STEPS TO DOWNLOAD AND INSTALL PYTHON:

Download the latest version of python(https://www.python.org/downloads/).Watch

the PIP list where pip is the package installer for python. Now upgrade the pip and setup
tools using the command

pip install --upgrade pip and pip install --upgrade setuptools

ENVIRONMENT INSTALLATION
Jupyter Notebook is an open-sourced web-based application which allows you to
create and share documents containing live code, equations, visualisations, and narrative
text. It is a server-client application that allows editing and running notebook documents
via a web browser. The Jupyter Notebook App can be executed on a local desktop
requiring no internet access.

Installation- (https://jupyter.org/install/)

Dept.Of 1
CSE,GMIT,Bharathinagara 3
 Detection of Alzheimer’s disease using Deep 2024-
learning 25

pip install notebook

4.3 PYTHON LIBRARIES :

There are many libraries in python.In those, we only use a few main libraries needed

4.3.1 NUMPY LIBRARIES :

NumPy is an open-source numerical Python library. NumPy contains a multi-

https://pandas.pydata.org/getting_started.html dimensional array and matrix data structures.
It can be utilized to perform a number of mathematical operations on arrays such as
trigonometric, statistical, and algebraic routines like mean, mode, standard deviation, etc.,

Installation- (https://numpy.org/install/)
Here we mainly use an array, to find the mean and standard deviation.

4.3.2 PANDAS LIBRARY :

Pandas is a high-level data manipulation tool developed by Wes McKinney. It is

built on the Numpy package and its key data structure is called the Data Frame. Data
Frames allow you to store and manipulate tabular data in rows of observations and columns
of variables. There are several ways to create aDataFrame.

Installation- (https://pandas.pydata.org/getting_started.html/)

pip install pandas

4.3.3 MATPLOTLIB :

Matplotlib is a comprehensive library for creating static, animated, and interactive

visualizations in Python. Matplotlib makes easy things easy and hard things
possible. Use interactive figures that can zoom, pan, update, visualize etc..,

Installation- (https://matplotlib.org/users/installing.html/)

pip install matplotlib

Dept.Of 1
CSE,GMIT,Bharathinagara 4
Detection of Alzheimer’s disease using Deep 2024-
learning 25
4.3.4 PILLOW LIBRARY:

The Python Imaging Library adds image processing capabilities to your

Python interpreter. The core image library is designed for fast access to data stored in a few
basic pixel formats.

Installation-(https://pillow.readthedocs.io/en/stable/installation.html/)

pip install pillow

4.3.5 TENSORFLOW LIBRARY:

TensorFlow is a framework created by Google for creating Deep Learning models.

Deep Learning is a category of machine learning models (=algorithms) that use multi-layer
neural networks. TensorFlow is a great tool which, if used properly has innumerable
benefits.

Installation-(https://www.tensorflow.org/install/)

pip install scikit-learn

4.3.6 KERAS LIBRARY :

Keras is a powerful and easy-to-use free open source Python library for developing
and evaluating deep learning models. It is an open-source software library that provides a
Python interface for artificial neural networks. Keras acts as an interface for the TensorFlow
library. Keras follows best practices for reducing cognitive load: it offers consistent &
simple APIs, it minimizes the number of user actions required for common use cases, and it
provides clear and actionable feedback upon user error.

Installation-(https://keras.io/getting_started/)

pip install keras

Dept.Of 1
CSE,GMIT,Bharathinagara 5
Detection of Alzheimer’s disease using Deep 2024-
learning 25
4.3.7 SCIKIT-LEARN LIBRARY :

Scikit-learn is a free machine learning library for the Python. It features various
algorithms like support vector machine, random forests, regression and k- neighbours, and it
also supports Py(thon numerical and scientific libraries like NumPy and SciPy.

Installation-(https://scikit-learn.org/stable/install.html/)

pip install scikit-learn

4.4 MODULE IMPLEMENTATION :

A modular design reduces complexity, facilities change (a critical aspect of software

maintainability), and results in easier implementation by encouraging parallel development of
different part of system. Software with effective modularity is easier to develop because
function may be compartmentalized and interfaces are simplified.Software architecture
embodies modularity that is software is divided into separately named and addressable
components called modules that are integrated to satisfy problem requirements.Modularity is
the single attribute of software that allows a program to be intellectually manageable.

The five important criteria that enable us to evaluate a design method with respect to its
ability to define an effective modular design are: Modular decomposability, Modular Comps
ability, Modular Understand ability, Modular continuity, Modular Protection.

The following are the modules of the project, which is planned in aid to complete the
project with respect to the proposed system, while overcoming existing system and also
providing the support for the future enhancement.

4.4.1 DATA PRE-PROCESSING :

The dataset is derived from the Kaggle an open source platform, which contains around
6500 images comprising of four classes namely moderately demented, very mildly demented
and non-demented, mildly demented. The dimension of images used in the dataset is of form
176x208. The images are reshaped into 180x180. The SMOTE technique is applied to the
dataset to solve the class imbalance problem in the dataset by randomly duplicating minority
classes in the dataset to match the majority classes [33].

Dept.Of 1
CSE,GMIT,Bharathinagara 6
Detection of Alzheimer’s disease using Deep 2024-
learning 25
With the random seed of 42, the minority classes oversampled using SMOTE
technique. The benefits of using SMOTE include the ability to reduce knowledge loss and
minimize over-fitting. Table 2 shows the dataset distribution after SMOTE technique
increased to 12800 images, with each class contains 3200 images. It as machine learning
engineers use this data to fine- tune the model hyper parameters. Data collection, data analysis,
and the process of addressing data content, quality, and structure can add up to a time-
consuming to-do list. During the process of data identification, it helps to understand your data
and its properties; this knowledge will help you choose which algorithm to use to build your
model.

A number of different data cleaning tasks using Python Pandas library and specifically,
it focus on probably the biggest data cleaning task, missing values and it ableto more quickly
clean data. It wants to spend lesstime cleaning data, and more time exploring and modeling.

Some of these sources are just simple random mistakes. Other times, there can be a
deeper reason why data is missing. It’s important to understand these different types of missing
data from a statistics point of view. The type of missing data will influence how to deal with
filling in the missing values and to detect missing values, and do some basic imputation and
detailed statistical approach for dealing with missing data. Before, joint into code, it’s important
to understand the sources of missing data.

Here are some typical reasons why data is missing:

1. User forgot to fill in a field.
2. Data was lost while transferring manually from a legacy database.
3. Due to Non-availability of data.
4. Users chose not to fill out a field tied to their beliefs about how the results would be used
or interpreted.

Variable identification with Uni-variate, Bi-variate and Multi-variate analysis:

1. import libraries for access and functional purpose and read the given dataset
2. General Properties of Analyzing the given dataset
3. Display the given dataset in the form of data frame
4. show columns

Dept.Of 1
CSE,GMIT,Bharathinagara 7
 Detection of Alzheimer’s disease using Deep 2024-
learning 25
5. shape of the data frame
6. To describe the data frame
7. Checking data type and information about dataset
8. Checking for duplicate data
9. Checking Missing values of data frame
10. Checking unique values of data frame
11. Checking count values of data frame
12. Rename and drop the given data frame
13. To specify the type of values
14. To create extra columns

Table 4.1 Dataset Distribution in Obtained Dataset

Class Number of Images

Non-Demented 2240
Moderate Demented 3200
Mild Demented 896
Very Mild Demented 64

Table 4.2 Dataset Distribution after applying SMOTE

Class Number of Images

Non-Demented 3200
Moderate Demented 3200
Mild Demented 3200
Very Mild Demented 3200

4.4.2 DATA VALIDATION/CLEANING/PREPARING PROCESS :

Importing the library packages with loading the given dataset. To analyzing the variable
identification by data shape, data type and evaluating the missing values, duplicate values. A
validation dataset is a sample of data held back from training your model that is used to give an
estimate of model skill while tuning model's and procedures that you can use to make the best
use of validation and test datasets when evaluating your models. Data cleaning / preparing by

Dept.Of 1
CSE,GMIT,Bharathinagara 8
Detection of Alzheimer’s disease using Deep 2024-
learning 25
rename the given dataset and drop the column etc. to analyze the uni-variate, bi-variate and
multi-variate process.
The steps and techniques for data cleaning will vary from dataset to dataset. The
primary goal of data cleaning is to detect and remove errors and anomalies to increase the value
of data in analytics and decision making

Fig 4.1: Pre-processing Diagram

GIVEN INPUT EXPECTED OUTPUT

Input: data.
Output: removing noisy data, Solve class imbalance problem.

4.4.3 EXPLORATION DATA ANALYSIS OF VISUALIZATION :

Data visualization is an important skill in applied statistics and machine learning.

Statistics does indeed focus on quantitative descriptions and estimations of data. Data
visualization provides an important suite of tools for gaining a qualitative understanding. This
can be helpful when exploring and getting to know a dataset and can help with identifying
patterns, corrupt data, outliers, and much more. With a little domain knowledge, data
visualizations can be used to express and demonstrate key relationships in plots and charts that
are more visceral and stakeholders than measures of association or significance. Data
visualization and exploratory data analysis are whole fields themselves and it will recommend a
deeper dive into some the books mentioned at the end.

Sometimes data does not make sense until it can look at in a visual form, such as with
charts and plots. Being able to quickly visualize of data samples and others is an important skill
both in applied statistics and in applied machine learning. It will discover the many types of
plots that you will need to know when visualizing data in Python and how to use them to better
understand your own data.
1. How to chart time series data with line plots and categorical quantities with
bar charts.

Dept.Of 1
CSE,GMIT,Bharathinagara 9
 Detection of Alzheimer’s disease using Deep 2024-
learning 25
2. How to summarize data distributions with histograms and box plots.

Fig. 4.2 :Data Visualization Diagram

GIVEN INPUT EXPECT OUTPUT

Input: data
Output: visualized data

4.44 COMPARING ALGORITHM WITH PREDICTION IN THE FORM OF

BEST ACCURACY RESULTS

It is important to compare the performance of multiple different deep learning

algorithms consistently and it will discover to create a test harness to compare multiple
different deep learning algorithms in Python with scikit-learn. It can use this test harness as a
template on your own problems and add more and different algorithms to compare. Each
models will have different performance characteristics.

Using resampling methods like cross validation, you can get an estimate for how
accurate each model may be on unseen data. It needs to be able to use these estimates to
choose one or two best models from the suite of models that you have created. When have a
new dataset, it is a good idea to visualize the data using different techniques in order to look at
the data from different perspectives. The same idea applies to model selection. You should
use a number of different ways of looking at the estimated accuracy of your algorithms in
order to choose the one or two to finalize. A way to do this is to use different visualization
methods to show the average accuracy, variance, and other properties of the distribution of
model accuracies.

In the next section, you will discover exactly how you can do that in Python with scikit-
learn. The key to a fair comparison of algorithms is ensuring that each algorithm is evaluated
in the same way on the same data and it can achieve this by forcing each algorithm to be

Dept.Of 2
CSE,GMIT,Bharathinagara 0
Detection of Alzheimer’s disease using Deep 2024-
learning 25
evaluated on a consistent test harness.

Pre-processing refers to the transformations applied to our data before feeding it to the
algorithm. Data Preprocessing is a technique that is used to convert the raw data into a clean
data set. In other words, whenever the data is gathered from different sources it is collected in
raw format which is not feasible for the analysis. To achieve better results from the applied
model the data has to be in a proper manner. Different deep learning algorithms are executed
in a given dataset. In the example below these 2 different algorithms are compared with our
algorithm:

 ResNet50
 Inception V3

False Positives (FP): A person who will pay predicted as defaulter. When actual class is no
and predicted class is yes. E.g. if actual class says this passenger did not survive but predicted
class tells you that this passenger will survive.

False Negatives (FN): A person who default predicted as payer. When actual class is yes but
predicted class in no. E.g. if actual class value indicates that this passenger survived and
predicted class tells you that passenger will die.

True Positives (TP): A person who will not pay predicted as defaulter. These are the correctly
predicted positive values which means that the value of actual class is yes and the value of
predicted class is also yes. E.g. if actual class value indicates that this passenger survived and
predicted class tells you the same thing.

True Negatives (TN): A person who default predicted as payer. These are the correctly
predicted negative values which means that the value of actual class is no and value of
predicted class is also no. E.g. if actual class says this passenger did not survive and predicted
class tells you the same thing.

True Positive Rate(TPR) = TP / (TP + FN)

False Positive Rate(FPR) = FP / (FP + TN)

Dept.Of 2
CSE,GMIT,Bharathinagara 1
Detection of Alzheimer’s disease using Deep 2024-
learning 25
Accuracy: The Proportion of the total number of predictions that is correct otherwise overall
how often the model predicts correctly defaulters and non-defaulters.
Accuracy calculation:
Accuracy = (TP + TN) / (TP + TN + FP + FN)
Accuracy is the most intuitive performance measure and it is simply a ratio of correctly
predicted observation to the total observations. One may think that, if we have high accuracy
then our model is best. Yes, accuracy is a great measure but only when you have symmetric
datasets where values of false positive and false negatives are almost same.

Precision: The proportion of positive predictions that are actually correct. Precision = TP
/ (TP + FP)
Precision is the ratio of correctly predicted positive observations to the total predicted
positive observations. The question that this metric answer is of all passengers that labeled as
survived, how many actually survived? High precision relates to the low false positive rate. We
have got 0.788 precision which is pretty good.

Recall: The proportion of positive observed values correctly predicted. (The proportion of
actual defaulters that the model will correctly predict)
Recall = TP / (TP + FN)
Recall(Sensitivity) - Recall is the ratio of correctly predicted positive observations to the all
observations in actual class - yes.

F1 Score is the weighted average of Precision and Recall. Therefore, this score takes both false
positives and false negatives into account. Intuitively it is not as easy to understand as
accuracy, but F1 is usually more useful than accuracy, especially if you have an uneven class
distribution. Accuracy works best if false positives and false negatives have similar cost. If the
cost of false positives and false negatives are very different, it’s better to look at both Precision
and Recall.

General Formula:
F- Measure = 2TP / (2TP + FP + FN)
F1-Score Formula:

Dept.Of 2
CSE,GMIT,Bharathinagara 2
Detection of Alzheimer’s disease using Deep 2024-
learning 25
F1 Score = 2*(Recall * Precision) / (Recall + Precision)

4.5 ALGORITHM AND TECHNIQUES:

4.5.1 ALGORITHM EXPLANATION:

Deep learning algorithms play a crucial role in determining the features and can handle
the large number of processes for the data that might be structured or unstructured. Although,
deep learning algorithms can overkill some tasks that might involve complex problems because
they need access to huge amounts of data so that they can function effectively. Deep learning
algorithms are highly progressive algorithms that learn about the image that we discussed
previously by passing it through each neural network layer. The layers are highly sensitive to
detect low-level features of the image like edges and pixels and henceforth the combined layers
take this information and form holistic representations by comparing it with previous data.
Convolutional neural network (CNN) popularly known as ConvNets majorly consists of
several layers and are specifically used for image processing and detection of objects.

CNNs process the data by passing it through multiple layers and extracting features to
exhibit convolutional operations. The Convolutional Layer consists of Rectified Linear Unit
(ReLU) that outlasts to rectify the feature map. The Pooling layer is used to rectify these
feature maps into the next feed. Pooling is generally a sampling algorithm that is down-
sampled and it reduces the dimensions of the feature map. Later, the result generated consists of
2-D arrays consisting of single, long, continuous, and linear vector flattened in the map. The
next layer i.e.,called Fully Connected Layer which forms the flattened matrix or 2-D array
fetched from the Pooling Layer as input and identifies the image by classifying it.

4.5.2 USED PYTHON PACKAGES :

Sklearn:
 In python, sklearn is a machine learning package which include a lot of ML
algorithms.
 Here, we are using some of its modules like train_test_split,
DecisionTreeClassifier or Logistic Regression and accuracy_score.
NumPy:

Dept.Of 2
CSE,GMIT,Bharathinagara 3
Detection of Alzheimer’s disease using Deep 2024-
learning 25
▪ It is a numeric python module which provides fast maths functions for
calculations.
 It is used to read data in numpy arrays and for manipulation purpose.
Pandas:
 Used to read and write different files.
 Data manipulation can be done easily with data frames.
Matplotlib:
 Data visualization is a useful way to help with identify the patterns from given
dataset.
 Data manipulation can be done easily with data frames.

4.5.3 NEURAL NETWORKS:

Neural networks are artificial systems that were inspired by biological neural
networks. These systems learn to perform tasks by being exposed to various datasets and
examples without any task-specific rules. The idea is that the system generates identifying
characteristics from the data they have been passed without being programmed with a pre-
programmed understanding of these datasets.

Components of a typical neural network involve neurons, connections, weights,

biases, propagation function, and a learning rule. Neurons will receive an input ( )
from

predecessor neurons that have an activation ( ), threshold j , an activation function f, and

an output function .

Connections consist of connections, weights and biases which rules how neuron
transfers output to neuron . Propagation computes the input and outputs the output and sums
the predecessor neurons function with the weight. The learning rule modifies the weights
and thresholds of the variables in the network.
The most well-known and simplest-to-understand neural network is the feedforward
multilayer neural network. It has an input layer, one or many hidden layers, and a single
output layer. Each layer can have a different number of neurons and each layer is fully
connected to the adjacent layer.
Types of Neural Networks

Dept.Of 2
CSE,GMIT,Bharathinagara 4
Detection of Alzheimer’s disease using Deep 2024-
learning 25
There are seven types of neural networks that can be used.
1. The first is a multilayer perceptron which has three or more layers and uses a
nonlinear activation function.

2. The second is the convolutional neural network that uses a variation of the
multilayer perceptrons.

3. The third is the recursive neural network that uses weights to make structured
predictions.

4. The fourth is a recurrent neural network that makes connections between the
neurons in a directed cycle.

5. The long short-term memory neural network uses the recurrent neural
network architecture and does not use activation function.

4.5.3 CONVOLUTIONAL NEURAL NETWORK :

A convolutional neural network, or CNN, is a deep learning neural network

designed for processing structured arrays of data such as images. The CNN architectures
are the most popular deep learning framework. CNNs are used for a variety of applications,
ranging from computer vision to natural language processing. Convolutional neural
networks can operate directly on a raw image and do not need any preprocessing.A
convolutional neural network is a feed-forward neural network, often with up to 20 or 30
layers. The power of a convolutional neural network comes from a special kind of layer
called the convolutional layer.

The architecture of a convolutional neural network is a multi-layered feed-forward

neural network, made by stacking many hidden layers on top of each other in sequence. It
is this sequential design that allows convolutional neural networks to learn hierarchical
features.The hidden layers are typically convolutional layers followed by activation layers,
some of them followed by pooling layers.The key building block in a convolutional neural
network is the convolutional layer. We can visualize a convolutional layer as many small
square templates, called convolutional kernels, which slide over the image and look for
patterns. Where that part of the image matches the kernel’s pattern, the kernel returns a
large positive value, and when there is no match, the kernel returns zero or a smaller

Dept.Of 2
CSE,GMIT,Bharathinagara 5
Detection of Alzheimer’s disease using Deep 2024-
learning 25
value.Convolutional networks are used for alternating between convolutional layers and
max- pooling layers with connected layers (fully or sparsely connected) with a final
classification layer. The learning is done without unsupervised pre-training. Each filter is
equivalent to a weights vector that has to be trained.
The diagram below represents CNN architecture. The diagram below represents CNN architecture.

Fig. 4.3 : CNN Architecture Representation

The following are definitions of different layers shown in the above architecture:

1. Input Layer: As input, normalized MRI image datasets are given as the first layer of
the presented neural network model

2. Convolutional layer: Convolutional layers are made up of a set of filters (also

called kernels) that are applied to an input image. The output of the convolutional
layer is a feature map, which is a representation of the input image with the filters
applied. Convolutional layers can be stacked to create more complex models, which
can learn more intricate features from images.

3. Pooling layer: Pooling layers are a type of convolutional layer used in deep
learning. Pooling layers reduce the spatial size of the input, making it easier to
process and requiring less memory. Pooling also helps to reduce the number of
parameters and makes training faster. There are two main types of pooling: max
pooling and average pooling. Max pooling
2 takes the maximum value from each
3

Dept.Of 2
CSE,GMIT,Bharathinagara 6
Detection of Alzheimer’s disease using Deep 2024-
learning 25
feature map,

Dept.Of 2
CSE,GMIT,Bharathinagara 7
 Detection of Alzheimer’s disease using Deep 2024-
learning 25
while average pooling takes the average value. Pooling layers are typically used after
convolutional layers in order to reduce the size of the input before it is fed into a
fully connected layer.

4. Fully connected layer: Fully-connected layers are one of the most basic types of
layers in a convolutional neural network (CNN). As the name suggests, each neuron
in a fully- connected layer is Fully connected- to every other neuron in the previous
layer. Fully connected layers are typically used towards the end of a CNN- when the
goal is to take the features learned by the previous layers and use them to make
predictions. For example, if we were using a CNN to classify images of animals, the
final Fully connected layer might take the features learned by the previous layers
and use them to classify an image as containing a dog, cat, bird, etc.

5. Dense Layer: By using flatten layer, high-dimensional data becomes column

vectors, which follows the convolutional layer, and then there is a dense layer,
which does the same functions as a neural network.

6. Dropout Layer: The dropout is regularization technique for neural networks.

During training, neurons in the hidden layers are randomly dropped out of the
dropout layer. By eliminating some of the neurons, this layer minimizes overfitting
in the model.

The following is a list of different types of CNN architectures:

LeNet:

The LeNet architecture is an excellent “first architecture” for Convolutional Neural

Networks (especially when trained on the MNIST dataset, an image dataset for handwritten
digit recognition). LeNet is small and easy to understand — yet large enough to provide
interesting results. Furthermore, the combination of LeNet + MNIST is able to run on the
CPU, making it easy for beginners to take their first step in Deep Learning and
Convolutional Neural Networks.

The main reason behind the popularity of this model was its simple and

Dept.Of 2
CSE,GMIT,Bharathinagara 8
Detection of Alzheimer’s disease using Deep 2024-
learning 25
straightforward architecture. It is a multi-layer convolution neural network for image
classification.The LeNet5 network has 5 layers with learnable parameters and hence named
Lenet-5. It has three sets of convolution layers with a combination of average pooling. After
the convolution and average pooling layers, we have two fully connected layers. At last, a
Softmax classifier which classifies the images into respective class.

The input to this model is a 32 X 32 grayscale image hence the number of channels
is one. We then apply the first convolution operation with the filter size 5X such filters. As a
result, we get a feature map of size 28X28X6. Here the number of channels is equal to the
number of filters applied. After the first pooling operation, we apply the average pooling and
the size of the feature map is reduced by half.

Then we have a final convolution layer of size 5X5 with 120 filters. As shown in the
above image. Leaving the feature map size 1X1X120. After which flatten result is 120
values. After these convolution layers, we have a fully connected layer with eighty-four
neurons. At last, we have an output layer with ten neurons since the data have ten classes.

Fig. 4.4 : LeNet Architecture Representation

VGGNet:

VGGNet is the CNN architecture that was developed by Karen Simonyan, Andrew
Zisserman et al. at Oxford University. VGGNet is a 16-layer CNN with up to 95 million
parameters and trained on over one billion images (1000 classes). The VGG-16 is one of the
most popular pre-trained models for image classification. It can take large input images of

Dept.Of 2
CSE,GMIT,Bharathinagara 9
Detection of Alzheimer’s disease using Deep 2024-
learning 25
224 x 224-pixel size for which it has 4096 convolutional features. CNNs with such large
filters are expensive to train and require a lot of data, which is the main reason why CNN
architectures like GoogLeNet (AlexNet architecture) work better than VGGNet for most
image classification tasks where input images have a size between 100 x 100-pixel and 350
x 350 pixels. The VGG CNN model is computationally efficient and serves as a strong
baseline for many applications in computer vision due to its applicability for numerous
tasks including object detection.

Fig. 4.5 : VGG-16 Architecture representation

Inception:

Inception network was once considered a state-of-the-art deep learning architecture

(or model) for solving image recognition and detection problems.Inception Layer is a
combination of all those layers (namely, 1×1 Convolutional layer, 3×3 Convolutional layer,
5×5 Convolutional layer) with their output filter banks concatenated into a single output
vector forming the input of the next stage.Along with the above-mentioned layers, there are
two major add-ons in the original inception layer 1×1 Convolutional layer before applying
another layer, which is mainly used for dimensionality reduction. A parallel Max Pooling
layer, which provides another option to the inception layer, the Inception Module just
performs convolutions with different filter sizes on the input, performs Max Pooling, and
concatenates the result for the next Inception module. The introduction of the 1 * 1
convolution operation reduces the parameters drastically. The Inceptionv2 model was a
major improvement on the Inceptionv1 model which increased the accuracy and further
made the model less complex. The researchers has introduced the Inceptionv3 model with a
few more improvements on v2.

Dept.Of 3
CSE,GMIT,Bharathinagara 0
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Fig. 4.6 : Inception Architecture Representation

ResNet:

ResNet is the CNN architecture that was developed by Kaiming He et al. The
network has 152 layers and over one million parameters, which is considered deep even for
CNNs because it would have taken more than 40 days on 32 GPUs to train the network on
the ILSVRC 2015 dataset. CNNs are mostly used for image classification tasks with 1000
classes, but ResNet proves that CNNs can also be used successfully to solve natural
language processing problems like sentence completion or machine comprehension, where
it was used by the Microsoft Research Asia team in 2016 and 2017 respectively. ResNet50
is a variant of ResNet model which has 48 Convolution layers along with 1 MaxPool and 1
Average Pool layer. Real-life applications/examples of ResNet CNN architecture include
Microsoft’s machine comprehension system, which has used CNNs to generate the answers
for more than 100k questions in over 20 categories. The CNN architecture ResNet is
computationally efficient and can be scaled up or down to match the computational power of
GPUs.

Fig. 4.7 : ResNet Architecture Representatio

4.5.4 OPTIMIZATION ALGORITHM :

As the optimal algorithm, the proposed model is trained with the Adaptive Moment

Dept.Of 3
CSE,GMIT,Bharathinagara 1
Detection of Alzheimer’s disease using Deep 2024-
learning 25
Estimation (Adam) algorithm. This method is very efficient when working big data set or
parameters. This algorithm is a combination of the root mean square propagation algorithm and
gradient descent algorithm. This algorithm takes the best characteristics of the two methods and
builds on them to produce a more optimized gradient descent algorithm. The rule is given by
equation 1, 2, and 3.

(1)

(2)

(3)

Where, =weights at time, =weights at time t+1, =learning rate, e=small positive
constant, =bias corrected weight parameters, =bias corrected , =sum of square of past
gradients, & =decay rates of average of gradients. Adam optimizer gives high performance
and outperforms other optimization algorithms.

Deep Learning Pipeline and Implementation:

The two main stages of implementing an algorithm are the training and testing phases.

First, the training phase creates the CNN model using a training dataset based on the
chosen models. The three most commonly known CNN models, namely, LeNet, Residual
Network (ResNet), and Inception, are employed. The performance validation of the training stage
guarantees the general performance of the classifier model and is used to avoid the overfitting
issue.

Then, verification is performed on the trained model in the testing stage using the testing
dataset as input to the trained classifier. This testing dataset is the other partitioned data from the
original dataset; therefore, it has identical characteristics to the training dataset. The original
dataset is partitioned into 70% and 30% for training and testing, respectively.Performance
metrics are used to evaluate both stages. By comparing the performance of both training and
testing, any overfitting issue can be determined. It occurs when the training performance is
relatively higher than the testing performance.

Dept.Of 3
CSE,GMIT,Bharathinagara 2
Detection of Alzheimer’s disease using Deep 2024-
learning 25
4.6 MODEL ARCHITECTURE :

Fig. 4.8 : Model Architecture

4.7 WORKFLOW DIAGRAM :

Fig. 4.9 : Workflow Diagram

CHAPTER-5

RESULTS AND DISCUSSION, PERFORMANCE ANALYSIS

5.1 RESULTS:

The model training is performed using parameter 100 epoch, batch size of 6500. The
model attains a validation accuracy of 94% and training accuracy of 98%. In order to
determine whether the model is correctly distinguishing positive and negative categories, at
every epoch, AUC (Area Under Curve) is determined. The confusion matrix is used to
calculate the each and every class metrics and shown in table.

Dept.Of 3
CSE,GMIT,Bharathinagara 3
Detection of Alzheimer’s disease using Deep 2024-
learning 25
Table 5.1 : Evaluation Metrics of Individual Class

Diseas PR REC F1- Support

e Score
ND 1 0.92 0.96 639
VMD 1 1 1 635
MD 0.9 0.94 0.92 662
MOD 0.9 0.92 0.91 624

5.2 PERFORMANCE ANALYSIS:

Comparing with previous works, they got an accuracy of 91%, but in our proposed model we
have achieved an overall accuracy of 94%.

Table 5.2 : Performance Analysis

Metric Our Model ResNet50 InceptionV3

Test Accuracy 94.59% 75.70% 88.83%

Training Accuracy 0.98 0.77 0.94

Precision 0.94 0.5833333134651184 0.89

Recall 0.94 0.09851446747779846 0.89
F1-Score 0.94 0.16603372991085052 0.89

CHAPTER-6
SUMMARY AND CONCLUSIONS
6.1 SUMMARY :

This project objective is to detect the stage of Alzheimer’s Disease using MR images. So
this model will greatly reduce the time taken to diagnose the disease, which is the important
factor.

6.2 Conclusion:

Dept.Of 3
CSE,GMIT,Bharathinagara 4
Detection of Alzheimer’s disease using Deep 2024-
learning 25
In this paper, a custom CNN model is proposed to predict the class of Alzheimer
disease among the given classified images. The proposed model has been tested with testing
data consisting of four classes and accomplish a 93.59% accuracy. The dataset has a major
disadvantage - a class disparity, the SMOTE method is employed to resolve this issue. In this
way, it is well suited to identify brain areas often associated with AD and can facilitate
physicians' decision making by helping them determine each patient's AD severity level
according to the level of dementia. Here we observe and compare the accuracy of three
models namely VGG16, ResNet50, Inception_V3 among these, The ResNet 50 model has the
best overall accuracy and F1 score.

Dept.Of 3
CSE,GMIT,Bharathinagara 5
Detection of Alzheimer’s disease using Deep 2024-
learning 25
REFERENCES

1. Fuse Hiroki; Oishi Kota; Maikusa Norihide; Fukami Tadanori; (2018). Detection
of Alzheimer's Disease with Shape Analysis of MRI Images. Japanese
Alzheimer’s Disease Neuroimaging Initiative.
2. H. M. Tarek Ullah; Zishan Ahmed Onik; Riashat Islam; Dr. Dip Nandi; (2018).
Alzheimer’s Disease And Dementia Detection From 3D Brain MRI Data Using
Deep Convolutional Neural Networks. 2018 3rd International Conference for
Convergence in Technology.

Dept.Of 3
CSE,GMIT,Bharathinagara 6
Detection of Alzheimer’s disease using Deep 2024-
learning 25
3. Neelaveni, J.; Devasana, M.S.Geetha (2020). 2020 6th International Conference
on Advanced Computing and Communication Systems (ICACCS) - Alzheimer
Disease Prediction using Machine Learning Algorithms.
4. Marcus, Daniel S.; Wang, Tracy H.; Parker, Jamie; Csernansky, John G.; Morris,
John C.; Buckner, Randy L. (2007). Open Access Series of Imaging Studies
(OASIS): Crosssectional MRI Data in Young, Middle Aged, Nondemented, and
Demented Older Adults. Journal of Cognitive Neuroscience.
5. Malone, Ian B.; Cash, David; Ridgway, Gerard R.; MacManus, David G.;
Ourselin, Sebastien; Fox, Nick C.; Schott, Jonathan M. (2013). MIRIAD—Public
release of a multiple time point Alzheimer's MR imaging dataset.
6. Fan, Zhao; Xu, Fanyu; Qi, Xuedan; Li, Cai; Yao, Lili (2019). Classification of
Alzheimer’s disease based on brain MRI and machine learning. Neural Computing
and Applications.
7. Acharya, U. Rajendra; Fernandes, Steven Lawrence; WeiKoh, Joel En; Ciaccio,
Edward J.; Fabell, Mohd Kamil Mohd; Tanik, U. John; Rajinikanth, V.; Yeong,
Chai Hong (2019). Automated Detection of Alzheimer’s Disease Using Brain MRI
Images– A Study with Various Feature Extraction Techniques. Journal of Medical
Systems.
8. Eke, Chima Stanley; Jammeh, Emmanuel; Li, Xinzhong; Carroll, Camille;
Pearson, Stephen; Ifeachor, Emmanuel (2020). Early Detection of Alzheimer's
Disease with Blood Plasma Proteins using Support Vector Machines. IEEE Journal
of Biomedical and Health Informatics.
9. R. Jain, N. Jain, A. Aggarwal and D. J. Hemanth (2019), Convolutional neural
network based Alzheimer’s disease classification fromagnetic resonance brain
images.

10. Ge, Chenjie; Qu, Qixun; Gu, Irene Yu-Hua; Store Jakola, Asgeir (2019). 2019
IEEE International Conference on Image Processing (ICIP) - Multiscale Deep
Convolutional Networks for Characterization and Detection of Alzheimer’s
Disease Using MR images.
11. Stoyanov, Danail; Taylor, Zeike; Kia, Seyed Mostafa; Oguz, Ipek; Reyes, Mauricio;
Martel, Anne; Maier-Hein, Lena; Marquand, Andre F.; Duchesnay, Edouard;

Dept.Of 3
CSE,GMIT,Bharathinagara 7
Detection of Alzheimer’s disease using Deep 2024-
learning 25
Löfstedt, Tommy; Landman, Bennett; Cardoso, M. Jorge; Silva, Carlos A.; Pereira,
Sergio; Meier, Raphael (2018). Understanding and Interpreting Machine Learning in

Dept.Of 3
CSE,GMIT,Bharathinagara 8
Detection of Alzheimer’s disease using Deep 2024-
learning 25
Visualizing Convolutional Networks for MRI-Based Diagnosis of Alzheimer’s Disease. First
International Workshops.
12. Biju, K.S.; Alfa, S.S.; Lal, Kavya; Antony, Alvia; Akhil, M Kurup (2017).
Alzheimer’s Detection Based on Segmentation of MRI Image. Procedia Computer
Science.
13. Khan, A., & Zubair, S. (2020). A Machine Learning-based robust approach to
identify Dementia progression employing Dimensionality Reduction in Cross-
Sectional MRI data. 2020 First International Conference of Smart Systems and
Emerging Technologies.
14. Tanveer, M., Richhariya, B., Khan, R. U., Rashid, A. H., Khanna, P., Prasad, M.,
& Lin, C. T. (2020). Machine Learning Techniques for the Diagnosis of
Alzheimer’s Disease. ACM Transactions on Multimedia Computing,
Communications, and Applications.
15. Subramanyam Rallabandi, V. P., & Seetharaman, K. (2021). Machine Learning-
Based Classification of Dementia Types: MRI Study. 2021 International
Conference on Artificial Intelligence and Smart Systems.
16. Mehmood, A., Yang, S., Feng, Z., Wang, M., Ahmad, A. S., Khan, Rizwan
Khan, Muazzam Maqsood, Yaqub, M. (2021). A Transfer Learning Approach
for Early Diagnosis of Alzheimer’s Disease on MRI Images.
17. A. De and A. S. Chowdhury (2021), DTI based Alzheimer’s disease classification
with rank modulated fusion of CNNs and random forest, Expert Syst. Appl.
18. J. Shi, X. Zheng, Y. Li, Q. Zhang and S. Ying (2018), Multimodal neuroimaging
feature learning with multimodal stacked deep polynomial networks for diagnosis
of Alzheimer’s disease.

Islam, Jyoti; Zhang, Yanqing (2018). Brain MRI analysis for

Alzheimer’s disease diagnosis using an ensemble system of deep
convolutional neural networks. Brain Informatics.
19. Mandal, Pravat K.; Mahajan, Rashima; Dinov, Ivo D.; Mandal, Pravat K.
(2012). Structural Brain Atlases: Design, Rationale, and Applications in
Normal and Pathological Cohorts. Journal of Alzheimer's Disease.
20. Jing Zhang; Chunshui Yu; Guilian Jiang; Weifang Liu; Longzheng Tong (2012).
3D texture analysis on MRI images of Alzheimer’s disease.
Dept.Of 3
CSE,GMIT,Bharathinagara 9
Detection of Alzheimer’s disease using Deep 2024-
learning 25
21. Krishnakumar, V.; Parthiban, Latha (2019). A Novel Texture Extraction
Technique with T1 Weighted MRI for the Classification of Alzheimer's
Disease. Journal of Neuroscience Methods.
22. Xiao, Zhe; Ding, Yi; Lan, Tian; Zhang, Cong; Luo, Chuanji; Qin, Zhiguang
(2017). Brain MR Image Classification for Alzheimer’s Disease Diagnosis
Based on Multifeature Fusion. Computational and Mathematical Methods in
Medicine.
23. Ben Ahmed, Olfa; Benois-Pineau, Jenny; Allard, Michèle; Ben Amar, Chokri;
Catheline, Gwénaë le (2015). Classification of Alzheimer’s disease subjects from
MRI using hippocampal visual features. Multimedia Tools and Applications.
24. Beheshti, I.; Demirel, H. (2015). Feature-ranking-based Alzheimer’s
disease classification from structural MRI. Magnetic Resonance Imaging.
25. Zhou, Qi; Goryawala, Mohammed; Cabrerizo, Mercedes; Wang, Jin; Barker,
Warren; Loewenstein, David A.; Duara, Ranjan; Adjouadi, Malek (2014). An
Optimal Decisional Space for the Classification of Alzheimer’s Disease and Mild
Cognitive Impairment. IEEE Transactions on Biomedical Engineering.
26. Prashanthi Vemuri; Jeffrey L. Gunter; Matthew L. Senjem; Jennifer L. Whitwell;
Kejal Kantarci; David S. Knopman; Bradley F. Boeve; Ronald C. Petersen; Clifford
R. Jack Jr. (2008). Alzheimer's disease diagnosis in individual subjects using
structural MR images: Validation studies. Neuro Image.
27. Liu, Manhua; Zhang, Daoqiang; Shen, Dinggang (2014). Hierarchical fusion of
features and classifier decisions for Alzheimer's disease diagnosis. Human Brain
Mapping.
28. Khedher, L.; Ramírez, J.; Górriz, J.M.; Brahim, A.; Segovia, F. (2015). Early
diagnosis of Alzheimer s disease based on partial least squares, principal component
analysis and support vector machine using segmented MRI images.
Neurocomputing.
29. Ben Ahmed, Olfa; Mizotin, Maxim; Benois-Pineau, Jenny; Allard, Michèle; Catheline,
Gwénaëlle; Ben Amar, Chokri (2015). Alzheimer's disease diagnosis on structural
MR
images using circular harmonic functions descriptors on hippocampus and posterior
cingulate cortex. Computerized Medical Imaging and Graphics.

Dept.Of 4
CSE,GMIT,Bharathinagara 0
Detection of Alzheimer’s disease using Deep 2024-
learning 25
30. S. Dubey, Alzheimer’s Dataset, (2019), Available at:
https://www.kaggle.com/tourist55/alzheimers-dataset-4-class-of-images

Dept.Of 4
CSE,GMIT,Bharathinagara 1
Detection of Alzheimer’s disease using Deep 2024-
learning 25
APPENDIX:
Screen Shots:

Dataset Overview

Dept.Of 4
CSE,GMIT,Bharathinagara 2
Detection of Alzheimer’s disease using Deep 2024-
learning 25

a) CNN Model Layers

Dept.Of 4
CSE,GMIT,Bharathinagara 3
Detection of Alzheimer’s disease using Deep 2024-
learning 25

b) Training Model Accuracy

c) Training Model AUC

Dept.Of 4
CSE,GMIT,Bharathinagara 4
Detection of Alzheimer’s disease using Deep 2024-
learning 25

d) Training Model loss

e) Detailed Final Model Analysis Of Each Category

Dept.Of 4
CSE,GMIT,Bharathinagara 5
Detection of Alzheimer’s disease using Deep 2024-
learning 25

f) Visualization Of The Prediction

Dept.Of 4
CSE,GMIT,Bharathinagara 6
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Dept.Of 4
CSE,GMIT,Bharathinagara 7
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Dept.Of 4
CSE,GMIT,Bharathinagara 8
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Dept.Of 4
CSE,GMIT,Bharathinagara 9
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Dept.Of 5
CSE,GMIT,Bharathinagara 0
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Dept.Of 5
CSE,GMIT,Bharathinagara 1
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Dept.Of 5
CSE,GMIT,Bharathinagara 2
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Dept.Of 5
CSE,GMIT,Bharathinagara 3
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Dept.Of 5
CSE,GMIT,Bharathinagara 4
Detection of Alzheimer’s disease using Deep 2024-
learning 25

Dept.Of 5
CSE,GMIT,Bharathinagara 5
Detection of Alzheimer’s disease using Deep 2024-
learning 25

4
9