Skip to content

tranktle/2022_mri_classification

Repository files navigation

2D_4classes_mri_classification

This project is about using Pytorch for 2D MRI image classification using transfer learning with Resnet34 with a testing accuracy is 98%.

The images are downloaded from Kaggle. These images will be classified into four categories, including

NOD: Non Demented,
VMD: Very Mild Demented,
MID: Mild Demented,
MOD: Moderate Demented.

Index Description Jupiter notebook Content data
1 Val_accuracy: 0.98
Testing accuracy: 0.68 😢
01_Resrnet34.ipynb Download data
split to train, val, test
Train with Resnet34
Testing evaluation
Reasoning
org_day
2 Val_accuracy: 0.99
Testing accuracy: 0.98 😃
Is this approach ok? 🤔
02_Resnet34.ipynb Combine data
Split data
Train model
Evaluate the model
allnew

For utility functions, please see mymodulo.py

Project Structure

  • 01-Resnet34.ipynb, 02-Resnet34.ipynb: Two main run files.
  • mymodule.py: all utility functions used in the project.
  • model: folder contains trained model.
  • Data: contains data after being downloaded.

About

2022_mri_classification

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published