As some of you have already realized, we’ve had some technical glitches in the initial two days of the preliminary test phase. Due to this, we’ve decided to extend the deadline by two days, so we will close the preliminary test phase on:
Thursday, September 2nd, 2021
From September 2 – September 3, 2021, we will open the final test phase. Please note you will have to submit your final solution to this phase yourself together with a preprint link (see below).
Important: You can only submit once (1 time) in total to this phase and no leaderboard or results for this submission will be posted before the workshop. Therefore, please make sure that you have your algorithm (& docker image) set up correctly.
For the submission you will need a brief description of your algorithmic approach, available as a preprint on the preprint server of your choice (arxiv.org, medrxiv.org,…). The abstract should be around 2 pages in length and describe:
- What your general algorithmic approach is
- How you trained it
- Any aspect that you think is new and interesting to know 🙂
- (optionally) Your result on the preliminary test set
Please note that you will have to submit your preprint to arxiv.org latest by
Thursday, Sept 2, 14:00 US Eastern Time
to receive a link for the submission in time (see also https://arxiv.org/help/availability). You may need to be endorsed to be able to submit a paper. Other preprint servers may have other policies. Please keep this in mind when preparing your submission.
As a reference / example of how your preprint may look like, please have a look the preprint describing our reference approach (the one in the docker container):
Wilm, F., Breininger, K., Aubreville, M: Domain Adversarial RetinaNet as a Reference Algorithm for the MItosis DOmain Generalization (MIDOG) Challenge, https://arxiv.org/abs/2108.11269
Feel free to download and adjust our preprint’s LateX source code (zip file here). Please make sure to cite the official MIDOG challenge description (http://doi.org/10.5281/zenodo.4573977) in your preprint.
Note: A submission needs to describe your own approach, so submission of the reference docker container is not a qualified submission.
Multiple submissions by different participants: Multiple submissions by different team members are accepted. However, in order to have a fair competition: All submitted approaches (independent of team memberships) need to be sufficiently different and follow a different methodology. This will be reviewed by the organizers for each case.
The MIDOG MICCAI workshop takes place on Friday October 1st at 2pm (see here). In order to participate, you will need to register to the workshop track of MICCAI (the registration fees are 60€ / 80€ for students and 80€/100€ for non-students). We appreciate attendance of every participant at the workshop.
The best and most interesting contributions (subject to the F1 metric and a subsequent peer review of the approach described in the preprint) will be invited by us to an oral presentation (10 minutes) of their work on the conference. Participation in the workshop will be required to be eligible for one of the awards. All two-page abstracts passing peer review and will still qualify as abstract for the proceedings (see below), even if they are not invited for an oral presentation.
You will receive the notification of acceptance (=invitation to an oral presentation) until:
Friday, September 15th, 2021
Note: We know that this does not allow you to qualify for the early bird registration fee for MICCAI. Should you not be able to join the workshop because the difference in the fee makes it unaffordable to you, please contact us.
Challenge proceedings and summary paper
We will publish challenge proceedings of the MIDOG challenge (together with the Learn2Reg and MOOD challenges) on Springer LNCS after the conference. The best qualified submissions on the final leaderboard will receive an invitation to contribute a full paper to these proceedings (pending peer review).
We will also conduct a peer review for the two-page abstracts (as described above). All submissions that are accepted (by the peer review) but not among those invited for oral presentation / full paper will still be included in the challenge’s proceedings as abstract.
We also aim to publish a summary paper on the challenge. The first and last author of the full proceedings papers will be invited to contribute to this summary paper.
Is is OK if the method description goes beyond two pages? (for example 2.5-3 pages?)
If you are a bit above two pages, this is still okay (e.g., conclusion/references on the third page). Still, we kindly ask you to be concise in your description, also since there will be the opportunity for a full paper for all invited approaches.
The arxiv articles I submitted have always been on-hold. How about half of this? Can I post the link to my github first?
I’m having the same issue, my preprint is “on hold” on arXiv and they say on the website that it can take a few days to resolve, so I don’t know how to submit to the challenge.
Hi Sen, hi Jack,
all articles who are submitted by Thu, 14:00 ET should be announced (=get a link) latest by Friday morning (ET). We will open the final submission phase now and you will have the opportunity to submit your docker solution until tomorrow (Friday, Sept 3) evening (CEST), which should be sufficient to receive the link.
In case your article is on hold until tomorrow (=Friday) afternoon, you may use replacement link instead, however, we would then ask you to follow up with the arXiv link as soon as you receive it.
i also ran into the same “on hold” issue this morning. Are there any specific requirements for replacement links (on github, eg.)?
Link to pdf in a public gitlab/github repository would be great. Thank you!
I wanted to submit for the final test phase of the MIDOG challenge but got an error “The algorithm failed on one or more cases”. I am sure I used this docker container for the preliminary test phase without any error. I tried the algorithm and it showed the outputs. I also tried to re-submit this docker container and I got this error: “A submission for this algorithm container image for this phase already exists.”
I appreciate it if you could help me with this issue.
we are looking this and will be contact with you via the MIDOG forum.
I have exactly the same situation. What should I do ?
I will be in contact via email.
Already notified acceptance teams?