avatarGaurav

Summary

TrainingData.io is providing a free collaborative workspace with an open-source COVID-19 radiology dataset to facilitate the development of diagnostic tools by data scientists and radiologists.

Abstract

In response to the urgent need for diagnostic tools to identify COVID-19, TrainingData.io has established a free collaborative workspace that includes an open-source dataset of chest X-rays and CT scans. This initiative aims to support data scientists and radiologists in annotating training data for machine learning models, with the dataset initially seeded from three open-source repositories. The platform encourages the contribution of de-identified imaging data from healthcare institutions worldwide, particularly from regions heavily affected by the pandemic, to diversify the dataset and expedite the creation of diagnostic tools. The use of the workspace for this purpose is offered at no cost, and TrainingData.io also provides expertise in building segmentation models using NVIDIA Clara's U-Net architecture to assist the community in developing open-source diagnostic models.

Opinions

  • The lack of diagnostic tools and proper testing is a significant issue in combating the COVID-19 outbreak.
  • Collaboration between data scientists and radiologists is crucial for building effective machine learning models for COVID-19 diagnosis.
  • There is a strong emphasis on the importance of diverse and comprehensive data for the development of robust diagnostic tools.
  • The initiative supports the open-source community and independent researchers by providing free access to necessary resources and tools.
  • The contribution of de-identified data from various sources is encouraged to enhance the dataset's representativeness and utility.
  • TrainingData.io is committed to adhering to licensing laws while supporting the global effort against COVID-19.

COVID-19 Radiology Dataset (chest XRay & CT) for Annotation & Collaboration (Part 1)

There is an urgent need for diagnostic tools to identify COVID-19. In the initial stages of this outbreak all countries, including the US, have faced one major problem — lack of diagnostic tools and proper testing. To be able to build diagnostic tools, data scientists are scrambling to get their hands on whatever little data is available.

Given the need for diagnostic tools to combat COVID-19 pandemic, TrainingData.io is providing a free collaborative workspace preloaded with open-source dataset. This collaborative workspace allows data-scientists and radiologists to share annotations of training data to be used for training machine learning models for COVID-19. This dataset is seeded with following datasets:

Anyone can create and download annotations by following this link

COVID-19 Training Data for machine learning

Open-source dataset for research: We are inviting hospitals, clinics, researchers, radiologists to upload more de-identified imaging data especially CT scans. The purpose is to make available diverse set of data from the most affected places, like South Korea, Singapore, Italy, France, Spain, USA. This contribution to open-source community will help independent researchers to build diagnostic tools faster. Product usage involved in this case will be completely free. Our intention is to help global effort to fight COVID-19 while following all licensing laws.

How to create and download annotations for COVID-19 Dataset?

Start here: https://app.trainingdata.io/v1/td/login

https://www.trainingdata.io

Build ML models using NVIDIA Clara train SDK: Team at TrainingData.io has expertise in building segmentation models using U-Net on NVIDIA Clara. We would like to offer our services to build open source models for the community. Here is a sample model to detect vertebral bodies made using 2D U-Net model in NVIDIA Clara train SDK.

Contact: For further information reach out to us: [email protected]

Current status: Workspace has 429 distinct images from 319 distinct patients, 369 CT images, 60 XRay images.

How can I download the dataset?: Only annotations (masks) created by community can be downloaded from TrainingData.io. Original images of the dataset cannot be downloaded from our website. To download original images, please visit the respective sources.

Cost: Irrespective of limits on free-usage, there will zero cost for using our product for work on this COVID-19 dataset.

Part II in this series: Automatic detection of COVID-19 infection in chest CT using NVIDIA Clara on TrainingData.io

Machine Learning
Image Processing
Data Science
Data Analysis
Health
Recommended from ReadMedium