Please submit your papers at https://cmt3.research.microsoft.com/HVSMR2016.
- Our online system for uploading segmentation results and retrieving accuracy metrics is open:
- Please register on the challenge submission system.
- Select the training or testing phase for the data format that you are working with.
- Click "Participate in this phase".
- Click "Submit your results".
- Upload your algorithm's segmentation results.
- The system will automatically calculate the accuracy metrics - these should be the metrics you report in your paper.
- You must upload your results to the online system to participate in the challenge.
- Traditional MICCAI format papers are solicited describing each group’s segmentation strategy and their results on the provided training dataset.
- Following the review process, authors of accepted papers will re-run their algorithms on the testing dataset and submit a revised paper that includes the testing results and addresses the reviewer comments. You may want to leave some space in your initial paper for presenting your results on the testing dataset.
- See the list of important dates for the deadlines for each step described above.
- Segmentation algorithms may be automatic or semi-automatic.
- For each type of data format (orientation and cropping), entries will be ranked based on their average Dice score compared to the expert manual segmentation on the testing dataset. Ancillary measures of segmentation accuracy, e.g. average surface distance and symmetric Hausdorff distance, will also be calculated and will be used when determining the final rankings and comparing algorithms that were evaluated using different data formats.
We also welcome submission of papers related to the workshop’s theme of image analysis for congenital heart disease (CHD) cases. Example topics include:
- Image Segmentation
- Image Registration
- Statistical atlases and population modeling
- Surgical planning
- 3D-printing heart models
- Intraoperative guidance
- Computer aided diagnosis
Papers should be up to 8 pages long and follow the submission guidelines of the main MICCAI conference. Please ensure that you remain anonymous in your submission, as explained in the requirements for the main MICCAI conference.
Reviews will be double-blind. Please go to the list of important dates to see the timeline for submissions and reviews.
HVSMR papers will be published in Springer Lecture Notes in Computer Science in conjunction with the RAMBO workshop.
We are also interested in co-writing a journal article with the challenge participants to report the results of the submitted whole-heart segmentation algorithms on the open congenital heart disease dataset. Details coming soon.
Awards for the best segmentation algorithms will be provided by: