DeepCor Logo

DeepCor-Denoising

A deep-learning based fMRI denoising tool using Contrastive Variational Autoencoders (CVAEs).

DeepCor Feature Figure

What is DeepCor?

DeepCor is a state-of-the-art fMRI denoising tool designed to enhance the quality of task-based and resting-state fMRI data. By leveraging Contrastive Variational Autoencoders (CVAEs), DeepCor effectively separates neural signals from noise, providing researchers with cleaner data for more accurate analysis.


Based on the CVAE architecture (Abid & Zou, 2019), DeepCor represents a significant step forward in neuroimaging data preprocessing.

How to Use DeepCor

DeepCor v2 (Latest)

The latest version of DeepCor is available as a Python toolbox. It supports GPU acceleration for faster processing.


View on GitHub →

Google Colab

Try DeepCor immediately without local installation using our Google Colab notebook. Ideal for testing and small datasets.


Run in Colab →

DeepCor v1

Access the original research code used in our initial publications. Useful for reproducibility.


View v1 Code →

Publications & News

DeepCor: Denoising fMRI Data with Contrastive Autoencoders
Zhu*, Y., Aglinskas*, A. & Anzellotti, S. (*co-first authors)
Nature Methods (2025)
Read Article
Denoising fMRI data using Contrastive Variational Autoencoders
Aglinskas, A., Zhu, Y & Anzellotti, S.
NeurIPS Workshop "Data and Brain and Mind" (Dec 2025)
View Poster PDF View Code
DeepCor (Preprint)
Zhu*, Y., Aglinskas*, A. & Anzellotti, S.
bioRxiv (2023)
Read Preprint

The Team

Yu Zhu
Yu Zhu
Brown University, USA
Aidas Aglinskas
Aidas Aglinskas
Boston College, USA
Stefano Anzellotti
Stefano Anzellotti
Boston College, USA