Learning with Incomplete Medical Data 

Special Session on LEARNING WITH INCOMPLETE MEDICAL DATA

The 49th IEEE International Conference on Acoustics, Speech, & Signal Processing - ICASSP 2024

Technical Scope

The sources of medical data are diverse and include various modalities, e.g., 1) 1D signals such as EEG/ECG signals, and 2) 2D images such as gigapixel histopathology images, and 3) 3D images such as MRI images and OCT images, and 4) videos such as ultrasound image sequences etc. and the data scale is rapidly expanding along with the acceleration of population aging. To alleviate the workload pressure caused by these big medical data, numerous machine learning methods have been developed and datasets have been created. However, on one hand, due to the huge costs associated with data collection and annotation, most existing datasets have limitations such as small scale, or missing data modalities, or missing partial labels, or even completely missing labels. On the other hand, due to privacy and security concerns, there are strict restrictions in medical data sharing, which often leads to that the local system only can access the medical data contained within itself rather than having the access right to all medical data across multiple local systems. These lead to existing models learned with those incomplete data be far from real-world clinical applications. Thus, it becomes urgent to develop novel machine learning methods for incomplete medical data. 

The aim of this special session is to serve as a premier platform to foster collaborations with researchers working on a variety of aspects of machine learning and medical data analysis, and inspire novel machine learning methods towards incomplete medical data for real-world clinical applications, and facilitate in-depth exploring and discussion. 

We invite submissions of original unpublished technical papers on topics including but not limited to: 

• semi-/self-supervised/unsupervised learning with incomplete medical data

• few-shot learning with incomplete medical data

• partial-supervised learning with incomplete medical data

• federated learning methods with medical data

• medical data synthesis methods

Submission Guidelines

• Prospective authors are invited to submit full-length papers of not more than four pages of technical content including illustrations, with an optional fifth page containing only references. Document preparation please prefer to https://cmsworkshops.com/ICASSP2024/papers/paper_kit.php

• Papers must be submitted by 6 September, 2023. 

• All the papers will go through peer review.

• Paper submissions to this special session are not through the regular ICASSP website. Please directly click here to submit your special session paper and select the Special Session ✨20.11: Learning with Incomplete Medical Data✨ 

Important Dates

• September 6, 2023: Deadline for paper submissions

• November 9, 2023: Reviews available to authors

• November 9-15, 2023: Author response period

• December 13, 2023: Notification of paper acceptances

• January 11, 2024: Camera ready paper deadline

• January 30, 2024: Deadline for author registration

• April 14-19, 2024: Conference dates

Organizers

Qing Liu Robert Jenssen Guoying Zhao

Central South University, China The Arctic University of Norway, Norway University of Oulu, Finland

Email: qing.liu.411@gmail.com Email: robert.jenssen@uit.no Email: guoying.zhao@oulu.fi