Skip to content
@farialab

ADS++

Tools for stroke-focused medical imaging.

ADS++ (Acute-stroke Detection Segmentation)

About

ADS++ , is an open-source PyTorch-based toolbox offering a modular, clinically-oriented, and highly interoperable suite of tools specifically designed for multimodal stroke imaging. Building upon our established Acute-stroke Detection Segmentation (ADS) framework, ADS++ provides a streamlined environment that encourages both the use of a complete, validated workflow and the flexible integration of its individual components with external tools and pipelines. The core philosophy of ADS++ is to deconstruct the monolithic toolkit into powerful, independently useful modules, each excelling at its specific task while maintaining seamless connectivity within the ADS++ workflow and with the broader neuroimaging ecosystem.

ADS++ allows you to:

  • Effortlessly manage and standardize diverse multimodal stroke datasets (DICOM, NIfTI, BIDS, remote S3/HTTP sources) through its adaptive Ads-IO module, eliminating cumbersome format conversions.
  • Implement robust and reproducible preprocessing pipelines using Ads-Prep, incorporating tools like SynthStrip, ANTs, and TorchIO for tasks like skull stripping, registration, and normalization, with built-in quality control.
  • Utilize and develop state-of-the-art deep learning models for stroke lesion segmentation (e.g., our flagship DAGMNet) and other tasks within Ads-DL, with support for a model zoo (UNet, MONAI integration) and flexible training powered by PyTorch Lightning.
  • Generate clinically relevant quantitative analyses and automated reports through AdsReport, including volumetric analysis, territorial mapping, and integration of explainable AI techniques for better model understanding.
  • Quantify stroke characteristics and predict outcomes using machine learning models within Ads-ML, with tools like auto-sklearn to optimize model selection.
  • Easily deploy validated models to various environments, including Docker containers for reproducibility and optimized models for edge devices with Ads-Deploy.
  • Develop new, specialized modules or integrate existing external tools more rapidly by leveraging the clear interfaces and standardized data structures within ADS++.

OpenADS, is released as the first open-source implementation of the ADS framework to provide a stable, reproducible, and engineering-ready reference pipeline for multimodal stroke imaging. The current system integrates the core components of the ADS methodology into a single, end-to-end workflow that handles data preparation, spatial normalization, deep learning–based lesion segmentation, quantitative analysis, and automated report generation. By open-sourcing OpenADS, we aim to make the full processing pipeline transparent and reproducible, enable researchers and developers to evaluate and extend the methods in practical environments, and establish a concrete baseline implementation that will inform the future modular architecture envisioned in the ADS++ ecosystem.

Popular repositories Loading

  1. OpenADS OpenADS Public

    Python 3

  2. .github .github Public

    Codebase overview

Repositories

Showing 2 of 2 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…