Skip to content

PiPNet3D: Patch-Based Intuitive Prototypes for Interpretable 3D Images Classification

Notifications You must be signed in to change notification settings

desantilisa/PIPNet3D

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

45 Commits
 
 
 
 
 
 
 
 

Repository files navigation

PIPNet3D

PiPNet3D: Patch-Based Intuitive Prototypes for Interpretable 3D Images Classification

We present PIPNet3D, a part-prototype neural network for volumetric images.

We applied PIPNet3D to the binary classification of Alzheimer's Disease from 3D structural Magnetic Resonance Imaging (sMRI, T1-MRI).

We assess the quality of prototypes under a systematic evaluation framework, propose new functionally grounded metrics to evaluate brain prototypes and develop an evaluation scheme to assess their coherency with domain experts.

Classes (clinical cognitive decline level):

  • Cognitively Normal (CN)
  • Alzheimer's Disease (AD)

arXiv preprint: "PIPNet3D: Interpretable Detection of Alzheimer in MRI Scans"

Accepted at the iMIMIC workshop during the MICCAI-2024 event.

Overview of PIPNet

Images and labels (cognitive decline level) were collected from the Alzheimer's Disease Neuroimaging Initiative (ADNI) https://adni.loni.usc.edu (data publicity available under request) and preprocessed using data_preprocessing.py functions.

Brain atlas (CerebrA) downloaded from https://nist.mni.mcgill.ca/cerebra/.

Codes adapted from the original PIPNet

Training a PIPNet: main_train_pipnet.py

Test a trained PIPNet: main_test_pipnet.py

Link to the weights of trained PIPNet(s) available in "models" folder.

About

PiPNet3D: Patch-Based Intuitive Prototypes for Interpretable 3D Images Classification

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages