Dear all,

 

We are welcoming paper submissions to the workshop on "Uncertainty for safe utilization of machine learning in medical imaging”  that will be held on the 27^th September as satellite event of MICCAI 2021.

More details on the workshop can be found at https://unsuremiccai.github.io

Deadline for submission is on the 25th June 2021

We are looking forward to your submissions

Christian Baumgartner

Adrian Dalca

Carole Sudre

Ryutaro Tanno

Koen Van Leemput

Sandy Wells


Call for Papers

Submission deadline 25^th June 2021 - https://unsuremiccai.github.io

Overview
With the rise and influence of machine learning (ML) in medical application and the need to translate newly developed techniques into clinical practice, questions about safety and uncertainty over measurements and reported quantities have gained importance. Obtaining accurate measurements is insufficient, as one needs to establish the circumstances under which these values generalize, or give appropriate error bounds for these measures. This is becoming particularly relevant to patient safety as many research groups and companies have deployed or are aiming to deploy ML technology in clinical practice.

The purpose of this workshop is to develop awareness and encourage research on uncertainty modelling to ensure safety for applications spanning both the MIC and CAI fields. In particular, this workshop invites submissions to cover different facets of this topic, including but not limited to: detection and quantification of algorithmic failures; processes of healthcare risk management (e.g. CAD systems); robustness and adaptation to domain shifts; evaluation of uncertainty estimates; defence against noise and mistakes in data (e.g. bias, label mistakes, measurement noise, inter/intra-observer variability). The workshop aims to encourage contributions in a wide range of applications and types of ML algorithms. The use or development of any relevant ML methods are welcomed, including, but not limited to, probabilistic deep learning, Bayesian nonparametric statistics, graphical models and Gaussian processes. We also aim to ensure broad coverage of applications in the context of both MIC and CAI, which are categorized into reporting problems (descriptions of image contents) such as diagnosis, measurements, segmentation, detection, and enhancement problems (addition of information) such as image synthesis, registration, reconstruction, super-resolution, harmonisation, inpainting and augmented display.


Scope
We accept submissions of original, unpublished work on safety and uncertainty in medical imaging, including (but not limited to) the following areas:
• Uncertainty quantification in any MIC or CAI applications
• Risk management of ML systems in clinical pipelines
• Defending against hallucinations in enhancement tasks (e.g. super-resolution, reconstruction, modality translation)
• Robustness to domain shifts
• Measurement errors
• Modelling noise in data (e.g. labels, measurements, inter/intra-observer variability)
• Validation of uncertainty estimates
• Active Learning
• Confidence bounds
• Posterior inference over point estimates
• Bayesian deep learning; Graphical models
• Gaussian processes
• Calibration of uncertainty measures
• Bayesian decision theory

Submission Format
Submissions must be 8-page papers (excluding references) following the Springer LNCS format. Author names, affiliations and acknowledgements, as well as any obvious phrasings or clues that can identify authors must be removed to ensure anonymity. Note that the 8 page limit refers only to the main content. Including references and acknowledgements the submission may exceed 8 pages.