46822394_ViT_ADNC

[Ei3-kw]

Hi Shakes,

Task

Classify Alzheimer’s disease (normal and AD) of the ADNI brain data (see Appendix for link) using one of the latest vision transformers such as the GFNet [6] set having a minimum accuracy of 0.8 on the test set.

This project implements a Vision Transformer (ViT) based classification system for analysing brain images from the ADNI (Alzheimer's Disease Neuroimaging Initiative) dataset. The model classifies brain images into different categories: Cognitive Normal (CN), Mild Cognitive Impairment (MCI), Alzheimer's Disease (AD - True MCI), and Subjective Memory Complaints (SMC).

What's Included

• Complete ViT implementation for brain image classification (CN vs MCI)
• Training framework with optimisations and early stopping
• Comprehensive evaluation system with visualisations
• Full documentation and usage instructions

File Structure

.
├── README.md
├── dataset.py        # Data loading and preprocessing
├── modules.py        # Model architecture definitions
├── train.py         # Training and optimisation scripts
├── predict.py       # Evaluation and prediction scripts
├── AD_NC/           # ADNI dataset directory
│   ├── train/
│   │   ├── AD/
│   │   └── NC/
│   └── test/
│       ├── AD/
│       └── NC/
└── checkpoints      # Trained models

Key Results

• Overall accuracy: 84.7%
• Strong CN identification (95.7% recall)
• High confidence in MCI detection (94.4% precision)
• Full evaluation metrics and visualisations included

Please follow the instructions in README.md to install dependencies and experiment with it.

Have an extension till Oct 30th 4PM

GLHF,
Ella

[shaivikaaaa]

@gayanku

[shaivikaaaa]

This is an initial inspection

Difficulty : Hard
Task : 5

1. Recognition Problem:

• The code structure looks ok
• missing data.py file
• no images are there for brain with AD and normal
• no train test accuracy graph
• Use pre-trained model and have additional files classes that are not part of the dataset @shakes76

2. Commit Log:

• commit messages are good
• All commits done within 2 days

3. Documentation:

Requirement:

Tittle: Done
Description of the algorithm: Done
Problem that it solves: Done
How it works in a paragraph (example any pre-processing): Done
Figure/visualisation (any input, output & results): Done
Dependencies: Done

4. Pull Request

Looks ok

Feedback:

• add images for normal and AD - brain

[Ei3-kw]

Hi,

The link to the train/ test images (https://filesender.aarnet.edu.au/?s=download&token=a2baeb2d-4b19-45cc-b0fb-ab8df33a1a24) seems expired. I have removed them from my local disk to save space. Can you provide them again so I can run my model over them?

I do have dataset.py if that's what you mean by data.py, and I'm happy to add images for normal and AD - brain, as well as train test accuracy/ loss graph once the raw images are provided.

Cheers,
Ella

[shaivikaaaa]

hi, I am not sure why the link is expired
you might just post on Ed once and see
please do upload your data.py file as well then

[Ei3-kw]

Yep I grabbed it from Rangpur, it's all g now. wdym by data.py?

[Ei3-kw]

Just updated README to include sample predictions. Do I need to update the version on Turnitin as well?

[aniketgupta17]

Observational Feedback

Pull Request:
Correctly created the Pull request from Topic Recognition Branch .
The pull request should include a clear description about the file structure .
Previous Feedback Incorporated .

File Organizing: Well-organized files.

Commit Log:
Commit messages are progressive for the Recognition Problem solved using 4 files .
Commits are not regularly made.

Documentation:
The README file should contain Conclusion and Future Improvements .
Code comments and docstrings are included.
Proper GitHub markdown formatting is used, with organized headings, lists, and code blocks.

[Ei3-kw]

Improvements are listed in Recommendations (https://github.com/Ei3-kw/PatternAnalysis-2024/tree/topic-recognition/recognition/46822394_ViT_ADNC#recommendations)
Wdym by conclusion?

[aniketgupta17]

The formal report looks if it has a concluding paragraph at the last .

[gayanku]

Marking

Good/OK/Fair Practice (Design/Commenting, TF/Torch Usage)
	Adequate design and implementation. Unwanted redundant code present.	-1
	Spacing and comments.
	Header blocks.
Recognition Problem
	OK solution to problem. Used a pre trained model.	-1
	Driver Script present.
	File structure NOT present.	-1
	Good Usage & Demo & Visualisation & Data usage.
	Module present.
	Commenting present.
	No Data leakage found.
	Difficulty : Hard. ViT (Hard Difficulty)
Commit Log
	Good Meaningful commit messages.
	Some/Adequate Progressive commits. Within 2 days.	-1
Documentation
	Readme :Acceptable. No train graphs.	-1
	Model/technical explanation :Good.
	Description and Comments :Good.
	Markdown used and PDF submitted. PDF checked.
Pull Request
	Pull Request has problems: Late submission.	-2
	No Feedback required.
	Request Description is good.
TOTAL		-7

Marked as per the due date and changes after which aren't necessarily allowed to contribute to grade for fairness.
Subject to approval from Shakes

[Ei3-kw]

• File structure presented in both README and PR
• I have extension, why late penalty?? PR opened before extended ddl

[hanemma7moud]

No description provided.

[Ei3-kw]

Please respond to my regrade request and remove my real name in the comment. I'd like to stay anonymous on GitHub.

[shakes76]

Can you email me your extension request and outcome please?

[Ei3-kw]

I have emailed you using student email

[shakes76]

Extension granted +2