Skip to content

Feature extraction of Open Access Series of Imaging Studies (OASIS) using Keras applications.

Notifications You must be signed in to change notification settings

caiocarneloz/koasis

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 

Repository files navigation

koasis

Keras pre-trained deep neural networks for Open Access Series of Imaging Studies (OASIS) feature extraction.

Getting Started

Dependencies

You need Python 3.7 or later to use koasis. You can find it at python.org.

You aso need numpy, pandas and keras packages, which is available from PyPI. If you have pip, just run:

pip install numpy
pip install pandas
pip install keras

Installation

Clone this repo to your local machine using:

git clone https://github.com/caiocarneloz/keras-oasis.git

Features

With this code, its possible to:

  • Retrieve all the labels from clinical info (considering CDR)
  • Extract features from any OASIS-1 image using Keras Applications
  • Generate a dataset containing all non-zero features and corresponding labels

Usage

Actually, you need to create a folder containing all OASIS images and CDR info. Having this, it is necessary just call the oasis_extract function sending the folder path:

oasis_extract('folder/')

Output

As output, for each pre-trained model, a .csv file containing the features and labels is generated:

            0         1         2    3  ...       6223       6224  6225       6226
0    0.751712  0.000000  4.072742  0.0  ...  42.221912   0.000000   0.0    Control
1    2.542512  0.000000  3.418418  0.0  ...  13.266520   5.047302   0.0    Control
2    0.904695  0.000000  3.447591  0.0  ...   7.464177   0.000000   0.0  Alzheimer
3    1.975773  0.000000  5.956251  0.0  ...  41.108479   0.000000   0.0    Control
4    0.000000  0.000000  0.584957  0.0  ...   2.805410   4.802437   0.0    Control
5    1.465851  0.000000  5.688468  0.0  ...  53.284348   0.000000   0.0    Control
6    0.000000  0.000000  6.295202  0.0  ...  20.517311   0.000000   0.0    Control
7    0.565955  0.000000  5.364164  0.0  ...  29.104387   0.000000   0.0    Control
8    2.452629  0.000000  3.623104  0.0  ...   8.846025   5.419124   0.0    Control
9    1.460919  0.000000  2.128028  0.0  ...  43.439747   0.000000   0.0    Control
10   3.632581  0.000000  4.830678  0.0  ...  23.841797  14.104014   0.0    Control
11   3.157856  0.000000  4.083702  0.0  ...  17.051870   6.060349   0.0  Alzheimer

Citation

If you use OASIS dataset, please cite the original publication:

MARCUS, Daniel S. et al. Open Access Series of Imaging Studies (OASIS): cross-sectional MRI data in young, middle aged, nondemented, and demented older adults. Journal of cognitive neuroscience, v. 19, n. 9, p. 1498-1507, 2007.

About

Feature extraction of Open Access Series of Imaging Studies (OASIS) using Keras applications.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages