Skip to content

High performance convolutional neural networks for detecting possible viral pneumonia caused by COVID19 through CT scans of the lungs.

Notifications You must be signed in to change notification settings

CJP-Whitt/COVID19-Detection

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

74 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Project for finding cases of possible COVID19 induced viral pneumonia.

For an abstract overview of the project, check out our final results presentation

Training/Validation Dataset

Images set used for all training and validation available here: https://drive.google.com/file/d/1xn-7_PvUr9fiCDt0f6hCBcftW2ZG2TXo/view?usp=sharing

To run most of the files, you would have to stay inside the COVID19_Detection directory on MATLAB and access
the other files by clicking the + symbol next to the folders and then clicking the file from there.

  • SURF_SVM.m: extracts features using the SURF method.
  • splitData.mlx: Splits the UCSD data into training, testing, and validation directories.
  • pham_code.mlx: The code the nature overview paper used for transfer learning with a 80/20 random data split.
  • leaveOneOutSetup.m: Put's one patient's images into a testing directory and the rest into a training directory.
  • uscd_Data: contains testing/training/validation data splits for UCSD data.
  • UCSD_combined: contains all of UCSD data in one training folder.
  • transferLearning_ucsdSplit: Preliminary transfer learning files for 4 CNNs using a data split that was provided on UCSD dataset github.
  • transferLearning_ucsdRandomSplit: Preliminary transfer learning files using random 80/20 data split.
  • transferLearning_ucsdLeaveOneOut: Preliminary transfer learning files using leave one patient out cross validation.
  • images_and_split: contains all COVID and nonCOVID images from UCSD data.
  • finalTesting: Contains transfer learning and SVM testing files that was used for final dataset.
  • featureExtraction_ucsdLeaveOneOut: SVM feature extraction files using leave one patient out cross validation.
  • COVID19_CT_seg_20cases: Contains final testing set with 3 images from each patient from CT scans.

About

High performance convolutional neural networks for detecting possible viral pneumonia caused by COVID19 through CT scans of the lungs.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages