Scribd
Upload
17 views2 pages

About Dataset Dementia

tq

Uploaded by

anuchandru003
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
17 views2 pages

About Dataset Dementia

tq

Uploaded by

anuchandru003
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as TXT, PDF, TXT or read online on Scribd
You are on page 1/ 2

About Dataset

Alzheimer MRI Preprocessed Dataset (128 x 128)

The Data is collected from several websites/hospitals/public repositories.


The Dataset is consists of Preprocessed MRI (Magnetic Resonance Imaging) Images.
All the images are resized into 128 x 128 pixels.
The Dataset has four classes of images.
The Dataset is consists of total 6400 MRI images.
Class - 1: Mild Demented (896 images)
Class - 2: Moderate Demented (64 images)
Class - 3: Non Demented (3200 images)
Class - 4: Very Mild Demented (2240 images)
Motive
The main motive behind sharing this dataset is to design/develop an accurate
framework or architecture for the classification of Alzheimers Disease.

References

https://adni.loni.usc.edu/
https://www.alzheimers.net/
https://www.kaggle.com/datasets/jboysen/mri-and-alzheimers
https://ieeexplore.ieee.org/document/9521165
https://catalog.data.gov/dataset/alzheimers-disease-and-healthy-aging-data
https://www.nature.com/articles/s41598-020-79243-9
https://cordis.europa.eu/article/id/429468-the-final-epad-dataset-is-now-available-
on-the-alzheimer-s-disease-workbench

-----------------------------------------------------------------------------------
-----------------------------------------------------------------------------------
-----------------------------------------------------------------------------------
----------------------------------------------------------------------

Alzheimer Detection and Classifier¶


This notebook(codes) can be also downloaded from secret gist:
https://tinyurl.com/2p8a9847
. Problem Definition¶
Alzheimer is known the most for the causes of dementia. It takes 2/3 of the whole
dementia popilation, while the cause is still unknown. According to papers from
Lancet neurology, even the old theory such as neural inflations, which was
disregarded as the cause compared to tau-protein, beta-amiloids, and genetic
factors (Apo E4).

Before I start, I'll have to explain a bit more about the Alzheimer, because, the
data contains 4 classes of Alzheimer that depends on the severity of dementia.

There are degrees of severity in Alzheimer.

Very mildly demented : This is the stage where patient starts to forget where they
put their stuff, other people's names recently, etc. It is hard to detect through
cognitive ability test.

Mildly demented : This is the stage where patients don't remember the words, can't
find their way to the destination, loss of focus and work-abilities. This is also
the stage where patients even forget that they are losing memory. From this stage,
with cognitive testing, it can be found.

Moderately demented : Starts to forget the recent activities, important old


histories, have hard time calculating the budget, hard to go outside alone, and
loss of empathy.
There are 3 more stages in the moderately dementia, which in the terminal stage,
the patient can't move on their own, while they lose the ability to speak. But I
assume that the current dataset from Kaggle
[https://www.kaggle.com/datasets/sachinkumar413/alzheimer-mri-dataset] is
considering all the these stages merged inside 'Moderately demeted' or not even
considered.

Knowing these stages are important because the faster the stage the patient is at,
the treatment will have higher effect in terms of slowing the process. If the
dementia is found during the moderately demented stage, it is known that the
patient will pass away in 3 years. (One of the reported case is a rythm guitarists
Malcom from the band AC/DC was diagnosed severe dementia at 2014.)

Thus, having an AI that detects alzheimer dementia in the early stage can allow
longer life expactancy from the patient as well as higher life quality overall from
the slowdown of dementia.

As Alzheimer can not only be found with cognitive ability testing, but also through
MRI or CT by looking at the ventricles of the brain and cortical atrophy, the
theoratical foundation on this project is solid. Doctors find the patient with
Alzheimer's have a brain that have enlarged ventricles (that lies in the center of
the brain) as well as thinner cortical grey area of the brain.

2. Solution Specification
This project will go through various ML models from basic (Principle Component
Analysis(PCA), Linear Discriminant Analysis(LDA), Support Vector Machine with 3
different kernels starting with linear to rbf kernel, and finally Convolutional
Neural Network models VGG16 for Detection and EfficientNetB0 for Classification)
and see which type works the best.

This project will contain 2 main sections. Alzheimer Detection and Alzheimer
Classification. Detection models (PCA, SVM, LDA, VGG16) will be making judgements
about the test data to see whether this person's brain image shows Alzheimer signs.

On the other hand, Alzheimer Classification will be done through training the data
with different severity for Alzheimer. Actually, this is how the original data is
made, and this model can be used as a detector as well. As ML model will provide
what stage the patient is, it may be used as a supportive tool for doctors to
diagnose Alzheimer's Disease. It brings some light in the practical world as it may
find early stages of AD, which can increase life expectancy as well as increased
life quality with supportive treatments.

Summary: I will try to implement diverse models for Alzheimer detection(non vs with
alzheimer) as well as Alzheimer Classifier(non vs very mild vs mild vs moderate)

Prepare the data in 2 different ways.


Alzheimer Detection: Whether the patient has the alzhiemer or not (non vs all other
categories)
Alzheimer Classifier: Define what stage the patient is in the alzheimer.
The models that are going to be tested here are

PCA for Alzheimer Detection


LDA for Alzheimer Detection
SVM for Alzheimer Detection and Alzheimer Classifier
CNN for Alzheimer Detection(VGG16) and Alzheimer Classifier(EfficientNetB0)

You might also like