CarOtid vessel wall SegMentation and atherosclerOsis diagnosiS challenge (COSMOS 2022)¶
Welcome to join the MICCAI COSMOS Challenge in person or online:¶
Welcome to join the SMRA grand challenge session in person or online:¶
August 24, 8:00 am-9:00 am (Los Angeles time), AI Grand Challenge
Join Zoom Meeting
https://us02web.zoom.us/j/83794872586?pwd=a2JUTjh1N0ljTGRZcmk0dW1GMmV2U
Meeting ID: 837 9487 2586
Passcode: 255242
Agenda:
Updates:
(NEW!) COSMOS 2022 Challenge Invitation to Present at SMRA and MICCAI¶
Dear participants,
Congratulations! We are willing to invite the top 10 teams in the CarOtid vessel wall SegMentation and atherosclerOsis diagnosiS challenge (COSMOS 2022) to present your work at the SMRA 2022 (Aug 23-26, 2022, https://society4mra.org/index.php/smra-2022-annual-meeting/) and MICCAI 2022 (Sept 18-22, 2022, https://conferences.miccai.org/2022/en/) events. Please let us know your answers to the following by August 8th 2022 (Please send your email to vws.challenge.2022@gmail.com).
1. Are you willing to present your work at SMRA 2022 (COSMOS 2022
session will be from 08:00am to 09:00am, PDT time, on Wednesday, Aug 24,
2022)?
- Organizers will collect and play the presentation videos from
participants.
- Please participate in the session, there will be a Q&A session, you
can choose to participate online (A ZOOM meeting link will be posted
soon) or in person (Los Angeles, CA).
2. Are you willing to present your work at MICCAI 2022 event (Sept 18,
2022)?
- More information will be available soon.
If yes, please provide us with the following information so that we can organize your presentation:
To provide by August 8th 2022
1.Team Name:
2.Name of Presenter:
3.Affiliation/Institution:
To provide by August 18th 2022
A 4-minute recorded video with slides to present your work. Notably, there will be an introduction of the background ahead of all the talks, considering the limited time, PLEASE DO NOT include Research Background in your talk.
Please upload your video to Baidu Cloud or Google Drive and email us the download link.
We suggest preparing your talk and video with PowerPoint or Keynote, please see the following instructions:
PowerPoint:
Keynote:
• https://support.apple.com/en-ca/guide/keynote/tan8a5df9cc5/mac
• https://support.apple.com/en-us/HT202220
To provide by August 20th 2022
1.Team Member Names:
2.Written Report in the format below:
Suggested Report Format
2 (minimum) to 4 (maximum) pages, follow MICCAI formatting guidelines (https://www.springer.com/gp/computer-science/lncs/conference-proceedings-guidelines)
Sections
- Rationale for choice of algorithm
- Description of algorithm with figure/flowchart/network structure chart
- Training procedure (input, output, data separation, data augmentation, loss function, optimizer, learning rate, etc.)
- Performance on training/validation set
- Model interpretability/explainability/robustness evaluations
- Limitations/Possible improvements of algorithm in future
- Challenges in vessel wall segmentation compared to other machine/deep learning tasks
To provide by September 1st 2022 (subject to change)
MICCAI presentation slides (10-minute recorded presentation explaining your method, training, and performance)
Each team will be allotted a total of 10 minutes for presentation and 5 minutes for questions/discussion.
Detailed instructions on the export format and upload location will be made available as the September 1st deadline approaches.
Please feel free to contact us if you have any questions (vws.challenge.2022@gmail.com).
2022/07/27：The COSMOS 2022 deadline (2022.7.31 23:59 EST) is approaching! Please do not hesitate to submit your Final Evaluation results!¶
Notably, please do submit your results of the test dataset
(https://drive.google.com/file/d/1IoXycGFX8o1nCLb8F_xsJZcZp-XzWqON/view?usp=sharing
or https://zenodo.org/record/6843257#.YtIsYdJBxH4) in the "Final
Evaluation" (NOT the "Validation"), OR YOUR RESULTS WOULD NOT BE
CORRECTLY EVALUATED AND RANKED.
Once you submit your results correctly, you can see your score and
ranking immediately in "Leaderboards".
Top 10 teams in the Final Evaluation will be invited to present their work at the SMRA 2022 (Aug 23-26) and MICCAI 2022 (Sep 18). There is still a big chance for you to win a place.
- Please leave your email in the submission comment for further communication.
- You can submit your results multiple times, but maximal twice per day.
If you have any questions, please contact us (vws.challenge.2022@gmail.com).
2022/07/16: Final evaluation phase of COSMOS 2022 is open now! Please note, the submission deadline has been extended to 2022.7.31 23:59 EST.
(NEW!) Testing Dataset has been uploaded!¶
https://drive.google.com/file/d/1IoXycGFX8o1nCLb8F_xsJZcZp-XzWqON/view?usp=sharing
or https://zenodo.org/record/6843257#.YtIsYdJBxH4
(Please note that the image size of case 60 is (432, 512, 512). When inserting contours into the QVS files, please take into account the effect of image size. )
2022/07/15: Final evaluation phase is coming!¶
- To avoid delay in responding to the participants' testing set requests, we decided to release the testing set directly (2022.7.16 00:00 EST).
- We will release annotations after the challenge to all teams that successfully submit the final results, please leave your email in the submission comment during the final evaluation phase.
- Data can be used for scientific purposes and research papers with submitting the final results and citing our challenge.
In addition, the deadline is currently set to EST (Eastern Standard Time, UTC-4).
2022/07/11: The evaluation docker for validation has been updated to fix the bug validation-submission-error.
In addition, note that the writing order of the boundary points should follow the edge order of polygons, instead of simply sorting the points by X and Y coordinates, e.g. [[0, 0], [1,0], [1,1], [0,1]] instead of [[0, 0], [1,0], [0,1], [1,1]].
2022/07/11: We also uploaded the evaluation code: evaluate_carotid_challenge_v2.ipynb (Please change "if target_wall_cont is None:" to "if target_wall_cont is None or target_lumen_cont is None:" in the evaluation code).
2022/07/09: Early submission for validation is open! Please refer to Validation Submission!
2022/07/07: The validation dataset was uploaded to https://drive.google.com/file/d/1HXrhJpN8j7PLA5Y0d0x1dv_yJ-hTuNbX/view?usp=sharing or https://zenodo.org/record/6804793#.YsZFoS1By9Z. And the submission guidelines were updated!
Background¶
Atherosclerosis is a leading cause of death worldwide, which occurs with luminal narrowing and plaque formation in multiple vascular beds, including carotid arteries. Early detection and proper treatment of carotid atherosclerosis can prevent the progression of cardiovascular disease. Magnetic resonance (MR) black-blood vessel wall imaging (BB-VWI) has been effective at visualizing normal and diseased arteries and characterizing atherosclerotic lesions, which was used in many previous clinical studies, with careful and comprehensive manual segmentation of the vessel wall. However, manual segmentation is labor-intensive and requires a high degree of training in vessel wall review. On the other hand, automatic segmentation is also challenging for complex atherosclerotic lesions and in complex arterial geometries.
Vessel wall imaging (VWI) with MRI of the carotid artery has been recognized to be able to identify atherosclerotic lesions which pose an increased risk of causing clinical events. However, traditional axial acquisition MRI sequences require a long scan. To ensure patient compliance and diagnostic image quality, a fast 3D carotid black blood MRI (volume isotropic turbo spin echo acquisition, VISTA) has been developed, which allows large coverage of carotid arteries with submillimeter isotropic resolution in the coronal acquisition, and is able to depict atherosclerotic lesion burden, severity, and luminal stenosis. The sequence has potential clinical application in identifying patients with advanced lesions but its application is limited due to the complexity of 3D image review, a large number of images available, and the lack of trained radiologists with extensive experience in the evaluation of carotid vessel wall thickness.
Challenge¶
In this challenge, the task is to segment the vessel wall from 3D-VISTA images and diagnose the atherosclerotic lesions with high accuracy and robustness. And then the clinical usable measurements such as wall thickness (difference between the lumen and outer wall contours), lumen area or stenosis percent can be derived from the vessel wall segmentation. In addition, the identification of the lumen and outer wall boundary of the vessel wall is also critical for the diagnosis of lesions. In summary, this challenge focuses on carotid vessel wall segmentation and atherosclerotic lesion diagnosis.
¶
How to download data?¶
Please register your team on the grand challenge website and join this challenge, then find the download link on the Data page (https://vessel-wall-segmentation-2022.grand-challenge.org/data/).
More details about the data format at the Data page (https://vessel-wall-segmentation-2022.grand-challenge.org/data/).
Winners¶
The challenge is endorsed by SMRA 2022 and MICCAI 2022. The top 10 teams will be announced in August 2022 and will be invited to present their work at the SMRA 2022 and MICCAI 2022 events. More details are to be discussed.
Requirements¶
- Use of images will be restricted solely to this challenge, participants must agree not to use the images in any other capacity.
- There are no publication requirements in this challenge. The participants may choose the author list (participating team) for their presentations at SMRA 2022 and MICCAI 2022. Participating teams are free to publish their own results. The challenge organizers will not publish a challenge paper first.
- There is no requirement for participants to open source their code. But publicly sharing your code is greatly encouraged.
Acknowledgment¶
Thank MICCAI 2022 and SMRA 2022 for the support of this challenge.
Thank TS Imaging for the great sponsorship and support of this challenge.
