On 12/31/2020, VinBigdata Chest X-ray Abnormalities Detection Challenge was officially launched on Kaggle platform. As a global playground for the Data Science and Artificial Intelligence community, this is an opportunity for domestic and international talents to access unique medical image data sources of Vietnamese people, competing for awards with a total value up to 50,000 USD.

VinBigdata Chest X-ray Abnormalities Detection Challenge is the first international medical imaging competition organized by Vietnamese organizations on the online platform Kaggle.com. With the challenge that requires to localize and classify 14 types of thoracic abnormalities from 18,000 chest radiographs, it helps tech- aholic to get closest to the work of a radiologist in the hospital, thereby developing appropriate solutions, solving practical medical problems.

Competition: VinBigdata Chest X-ray Abnormalities Detection Challenge

• Organizer: Vingroup Big Data Institute
• Duration: December 31, 2020 – March 31, 2021.
• Participants: Technology community worldwide
• The contest is held online on the Kaggle.com platform – the world’s largest community of Artificial Intelligence and Data Science today.
• Join the competition here.

1. TASK

Kagglers will be challenged to build ML algorithms to detect visual pattern of abnormal findings on chest x-ray scans. Each abnormal lesion should be located via a bounding box. Specifically, you’ll automatically localize and classify 14 types of thoracic abnormalities from chest radiographs. You’ll work with a dataset consisting of 18,000 scans that have been annotated by experienced radiologists. You can train your model with 15,000 independently-labeled images and will be evaluated on a test set of 3,000 images.

2. DATA

Data was collected from 108 Central Military Hospital and Hanoi Medical University Hospital. 100% of the data was de-identified (all patient identification information has been deleted or replaced by random values. (course) and retains only the necessary information for image analysis and processing). Each image provided by the contest is annotated by experienced radiologists in Vietnam. For training dataset, each case was independently labeled by 3 doctors. Meanwhile, for evaluation data, each case was labeled by a panel of 5 doctors.

3. RULES

• Contestant teams: Each team consists of up to 5 people, employees of Vingroup are not allowed to participate.
• Form of evaluation: Every day, each team is submitted up to 5 results, the rankings will be continuously updated in real time, through direct comparison with the doctor’s assessment. At the end of the contest, each team can submit the best models twice, and the results will be calculated on 100% of the test data set for re-ranking.
• Note: Winning teams must submit a description of the solution and the computer program source code used to train and execute the algorithm.

4. AWARDS

• Total prize value: 50,000 USD
• Best team prize: 20,000 USD
• The SPECIAL prize for the team with 100% Vietnamese members and the most impressive result: $8,000

5. ABOUT KAGGLE

Kaggle.com is Google’s online platform, specializing in organizing competitions in the areas of Artificial Intelligence and Data Science. It is the most famous platform in the world in this field with a large community of more than 5 million registered and 150,000 ranked contestants. Kaggle competitions are strictly assessed on the quality of the data as well as the difficulty and impact of the problem. Many large companies like Google, Amazon, Facebook, Microsoft, GE, Intel and major research centers like the North American Photovoltaic Association (RSNA) have held data contests on Kaggle. VinBigdata is the first organization in Vietnam to hold a contest on this platform, in the Top 10 companies in the world with the highest award value on Kaggle.

VinBigdata Chest X-ray Abnormalities Detection Challenge has been implemented since December 31, 2020. It lasts till March 31, 2021. The competition is one of our activities to promote research and create solutions and products for Vietnamese people, based on Vietnamese data.

Individuals and organizations can attend HERE.