Overview:

Medical imaging is a key component of modern healthcare, facilitating a wide array of diagnostic and therapeutic applications. Data-driven computer vision and AI solutions for medical imaging thereby represent a great potential to make a real-life impact by improving patient care.

However, safety requirements associated with healthcare pose major challenges for this research field, especially regarding data curation. Collection and annotation of medical data is often resource-intensive due to the need for medical expertise. At the same time, data quality is of the highest importance to ensure safe and fair usage in clinical settings. As a result, efficient data curation and validation as well as learning from small data are important areas of research. Synthetic data generation and augmentation are further promising directions, which themselves, however, pose challenges regarding quality, bias, and utility.

In addressing these demands, data engineering emerges as a crucial driver in advancing medical imaging research into deployment. Nevertheless, it is challenging to fulfill all the needs of task-specific applications via traditional methods. To bridge the gap, this workshop aims to encourage the discussion on topics related to data curation and augmentation for medical applications to tackle the challenges of limited or imperfect data in the real-world medical application.

Scope:

We invite topics related to medical imaging and medical computer vision which include but are not limited to:

• Data selection and curation
  • clinical data collection & analysis pipelines
  • multi-modal and/or longitudinal data management & analysis
  • data selection techniques
  • data bias analysis
  • annotation tools and strategies
• Data synthesis and augmentation
  • generative modeling
  • neural rendering
  • style transfer
  • physics-based simulation
  • sim2real gap
  • label augmentation
• Learning with limited & imperfect data
  • domain adaptation/generalization
  • few-shot/zero-shot learning
  • Semi-/self-/un-/weakly-supervised learning
  • active learning
  • multi-modal learning
  • noisy & incomplete data
• Data verification and quality assessment
  • annotation consensus
  • anonymization & privacy
• Effects of Data Domain Representation on Clinical Outputs
  • case studies of AI for rare diseases
  • downstream impact of data bias in augmentation
  • downstream impact of synthetic data
• From Data to Results: Learning from Success
  • showcasing data-centric solutions with clinical impact

Submission:

Papers can be submitted via [to be announced] until [to be announced]. The submission should be anonymous and follow the CVPR conference format and template (8 pages excl. references). Supplementary material can be submitted as well.

A double-blind reviewing process is adhered to guarantee paper quality. There will be a single round of reviews without a rebuttal. Decisions on acceptance will be announced by [to be announced]. In submitting a paper, authors implicitly acknowledge that no paper of substantially similar content has been or will be submitted to another conference or workshop until the decisions have been made by our workshop.

back