LDTM-MICCAI2024

Aims and Scope

Serial imaging is an integral part of the clinical workflow and is routinely employed to track the progression of disease or to assess response to treatment. For example, patients with cancer are often followed up longitudinally using radiological imaging (e.g., CT/MR/PET) to identify and track lesions, and assess treatment response. Recent developments in AI and machine learning have shown promise in automating or improving parts of the clinical workflow, from being able to find lesions in various organs, to classification and diagnosis of diseases. Unfortunately, despite the key role of serial imaging in the clinical workflow, developing AI systems that can track or model disease progression by learning from and exploiting longitudinal imaging has not received much attention until recently. Besides their role in the everyday clinical workflow, such models can enhance biological understanding of diseases to shape prevention and intervention strategies, inform clinical trial design, and support clinical decision making, such as patient diagnosis and prognosis.Therefore, in this workshop, we solicit submissions which adhere to the general theme of tracking and modeling disease progression with imaging and/or multimodal data.

Two areas are of particular interest:

Area 1: Cancer and concomitant diseases, where possible themes include tasks, such as detection, classification or segmentation of lesions or other abnormalities with or without multiple time points, combining imaging with other multi-modal data (e.g. radiology reports, electronic medical records), combining imaging and large language models (LLMs) to identify and track findings and disease status over time as well as mechanistic models.
Area 2: Neurodegenerative diseases, where data driven disease progression modeling (DPM), often using imaging biomarkers, has received increasing interest recently. In our recent publications in Nature Reviews Neuroscience, we provide comprehensive overviews of the current state-of-the-art phenomenological and pathophysiological models. Phenomenological models simultaneously estimate the disease progression trajectories from observed data together with the corresponding temporal axis and have important applications in patient subtyping and staging. Pathophysiological models aim to understand underlying disease mechanisms by modeling connectome-moderated pathology appearance and spreading, brain networks, and other dynamical systems models. Key emerging topics include: integrating pathophysiological and phenomenological models, pathophysiological models of the interaction of multiple disease mechanisms, and the role of deep learning in DPM.

The workshop is not limited to the above disease areas and we welcome submissions related to other diseases or applications as well. While the focus is on longitudinal data, we also welcome submissions which perform DPM with cross-sectional data. Finally, we also welcome and encourage submissions related to new datasets that can enable the research of tomorrow.

The scope of the conference includes, but is not limited to:

Detection, classification, segmentation of abnormal anatomical structures. Areas of interest include, but are not limited to the following: lesions, tumors, lymph nodes, neurodegenerative changes etc.
Disease progression modeling
Longitudinal data analysis
Follow-up assessment
Disease tracking and outcome prediction with longitudinal and/or multi-modal data
Surgical and Computer-Assisted interventions to determine/track changes
Histopathological analysis and change tracking over time
Image registration of longitudinal series for subsequent assessment
Multi-modal data fusion for detection, diagnosis, and image-guided interventions
New datasets and metrics

Call for Papers

We solicit submissions consistent with the aims and scope of the workshop. The following types of submissions will be accepted:

Full Papers: We encourage the submissions of full-length papers (upto 10 pages excluding references) describing new work that has not been previously published or accepted for publication. Accepted papers will be presented as orals and/or posters.
Extended Abstracts: We encourage the submissions of short papers ( upto 3 pages + 1 page references) describing new, previously, or concurrently published research or work-in-progress. The paper will be presented as a poster.
Accepted manuscripts from MICCAI main conference: We also invite authors of an accepted manuscript at MICCAI which is of interest for our workshop to present their work in an oral presentation at our workshop to strengthen the community around this topic.

All full papers and extended abstracts must adhere to the formatting and style used by the MICCAI main conference -- they should be anonymized and use the manuscript templates available at Lecture Notes in Computer Science. Accepted full papers will be published as a Springer Lecture Notes in Computer Science (LNCS) series.