STIR Challenge 2024

Description

The STIR Challenge is designed to quantify methods in deformable tracking, mapping and reconstruction. STIR is a part of EndoVis challenge at MICCAI 2024 (October 6-10, 2024). STIR will be quantified using a non-public test split of the publicly available STIR Dataset. The dataset is a stereo dataset with ground truth labels created using Infrared Tattoos1

Objective

Track points accurately and efficiently in videos

Motivation

Robust tracking and mapping in deformable scenarios is essential to enable downstream tasks in medical computer vision, such as motion compensation, and subtask automation2. A robust means to evaluate these methods for clinical efficacy requires a large labelled dataset, and a standardized evaluation. The STIR challenge provides that.

Example with MFTs3

Here are some examples of a baseline method (MFTs) on the STIR dataset.

Timeline

  • April 3 Challenge launch, 2D Evaluation Release
  • Mid-June 3D Quantification Release
  • October 6-10 MICCAI 2024
  • Early August Docker Instructions Available
  • September 23 Docker Container Submission Deadline
  • September 30 Report Submission Deadline
  • October 6-10 MICCAI 2024, challenge results (optional)

Sponsors

Sponsors: NVIDIA, ImFusion, Intuitive

Organizers

NameInstitution
Adam SchmidtIntuitive & UBC
Mert KaraogluImFusion & TUM
Soham SinhaNVIDIA
Alexander LadikosImFusion
Omid MohareriIntuitive
Simon DiMaioIntuitive
Tim SalcudeanUBC

Contact

Quick questions about our challenge can be posed via email or synapse. Email adam . schmidt at intusurg . com to be added, or register for the challenge above.

References

1- Schmidt, Adam, Omid Mohareri, Simon DiMaio, and Septimiu E. Salcudean. “STIR: Surgical Tattoos in Infrared.” IEEE Transactions on Medical Imaging (2024). 2- Schmidt, Adam, Omid Mohareri, Simon DiMaio, Michael C. Yip, Septimiu E. Salcudean. “Tracking and mapping in medical computer vision: A review.” Medical Image Analysis. 2024 Mar 2:103131. 3- Neoral, Michal, Jonáš Šerých, and Jiří Matas. “Mft: Long-term tracking of every pixel.” In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp. 6837-6847. 2024.