HoverFast - HoverFast v1.1.1

Release note

This release contains minor bug fixes on tissue mask generation and handling missing magnification to work on.

Bug fixes

• Fix the generated mask if none was given, the model would run on all the slides
• Fix the resizing when upscaling from higher magnification if the desire magnification is unavailable

Scientific Software - Peer-reviewed - Jupyter Notebook
Published by JulienMassonnet about 1 year ago

HoverFast - HoverFast v1.1.0

We are excited to announce the release of HoverFast v1.1.0.

New Features

• Support for DAB-stained IHC nuclei segmentation: This version introduces support for segmenting DAB stained IHC nuclei, utilizing GPU based stain deconvolution.
• Custom dataset creation tutorial: We have added a tutorial notebook that enables users to create custom datasets for nuclei or any round-like shaped pathology objects. This flexibility allows users to train models on other tissue structures, utilizing HoverFast’s optimized inference.

Bug fixes

• Bug that would cause first predicted object on labeled masks of infer_roi to be skipped has been fixed

Scientific Software - Peer-reviewed - Jupyter Notebook
Published by petroslk over 1 year ago

HoverFast - HoverFast v1.0.1 - JOSS release

Release note

This release contains minor bug fixes and feedback incorporated during the JOSS review process. We thank the editor and all the reviewers for their contributions.

Bug fixes

Object labeling of infer_roi would miss the first object.

Scientific Software - Peer-reviewed - Jupyter Notebook
Published by petroslk over 1 year ago

HoverFast - HoverFast v1.0.0 - Initial Release

Release Title

HoverFast v1.0.0 - Initial Release

Release Description

We are thrilled to announce the first official release of HoverFast (v1.0.0), a high-performance tool for efficient nuclear segmentation in Whole Slide Images (WSIs). This initial release includes a robust set of features designed to facilitate rapid and accurate segmentation, supporting research and diagnostics in medical imaging.

Highlights

Features

• Whole Slide Image Inference: Perform segmentation on large histopathological images with ease.
• Region of Interest (ROI) Inference: Efficiently process specific regions within WSIs.
• Model Training: Train HoverFast on your own data using provided tools and containers.
• Command Line Interface: Versatile CLI for various operations including inference and model training.
• Docker & Singularity Support: Simplified setup and execution using containers.

Installation

• Docker: Pull and run HoverFast using Docker for streamlined setup.
• Singularity: Use Singularity for environments that support it.
• Conda: Local installation for development purposes.

Usage

• Inference Commands: Detailed commands for running inferences on WSIs and ROIs.
• Training Commands: Steps to generate datasets and train models.

Acknowledgments

• Julien Massonnet - JulienMassonnet
• Petros Liakopoulos - petroslk
• Andrew Janowczyk - choosehappy

Scientific Software - Peer-reviewed - Jupyter Notebook
Published by petroslk over 1 year ago