MICCAI 2022 MELA Challenge: International Competition for Automated Mediastinal Lesion Detection in CT Scans

The internationally renowned premier conference on medical imaging, MICCAI 2022, will be held in Singapore this October. As a globally recognized academic conference in the fields of Medical Image Computing and Computer-Assisted Intervention, MICCAI attracts enthusiastic participation from professionals, top universities, and leading medical technology institutions both domestically and internationally each year through its various challenge competitions.

The MICCAI 2022 MELA International Challenge on Mediastinal Lesion Detection was jointly organized by Wuhan University, Shanghai Pulmonary Hospital Affiliated to Tongji University, and Diannai (Shanghai) Biotechnology Co., Ltd., with generous sponsorship from JD Explore Academy and the Specialized Alliance for the Promotion and Enhancement of Clinical Competencies in General Thoracic Surgery. Focusing on clinical needs in the field of mediastinal lesions, the competition provides the first large-scale benchmark dataset annotated by experts, aiming to further explore the research and clinical application value of automated screening for mediastinal lesions.

The mediastinum is the cavity located between the two lungs, containing various structures such as the thymus, heart, esophagus, and trachea. Lesions in this region are characterized by their complexity and heterogeneity. Detection of mediastinal lesions is a critical task in clinical practice; early identification plays a significantly positive role in the subsequent diagnosis and treatment of related diseases. However, the low contrast of the mediastinal area and the complexity of its lesions pose substantial challenges to manual screening and diagnosis.

1. April 1, 2022: Registration opened;

2. April 15, 2022: Release of training and validation datasets; competition launch;

3. July 1, 2022: The test set was released, and the competition registration channel closed;

4. July 3, 2022: The submission portal for competition results opens, and the leaderboard becomes available;

5. July 17, 2022: The submission portal for competition results closed.

The task of this competition is automated detection of mediastinal lesions in CT images. The dataset for this competition comprises 1,100 CT image sets, including 770 for training, 110 for validation, and 220 for testing. Each CT image set contains one or more mediastinal lesions, with annotations provided as three-dimensional bounding boxes.

1. The champion team receives$5,000Prizes; 2-3 winners will each receive$1,000Bonus; 4th to 5th place will each receive$500Bonus;

2. The organizers will cover the registration fees for the top 10 participating teams to attend the MICCAI 2022 satellite event, and these teams will receive certificates;

3. The top-ranked teams (up to 10) will be invited to write a conference summary paper, which may be published in leading medical imaging journals.

4. The submission portal for competition results will be open for only two weeks (July 3 to July 17);

5. Participants may submit up to 5 entries per day;

6. The top three competing teams are required to submit a four-page solution;

7. The use of other open-source data is permitted, but must be described in detail in the solution;

8. Any form of manual annotation on the training, validation, and test sets is prohibited. Automated algorithmic annotation is permitted but must be described in detail in the solution;

9. The competition organizers encourage participants to open-source their code;

10. The competition organizers reserve the right to disqualify non-compliant submissions.

The evaluation of this competition is based on FROC curve analysis, which assesses the average sensitivity of predictions submitted by participants at different FP (false positive) levels.

Register for a Grand Challenge account and participate in the MICCAI 2022 MELA Challenge. Official website:https://mela.grand-challenge.org/ 

Long-press the QR code below to go directly to the competition's official website

 * The following list is in no particular order.