Important Dates

Event Date
Full Paper Deadline June 29th, 2024; 11:59 PM PST
Notification of Acceptance July 15th, 2024; 11:59 PM PST
Camera-ready Deadline July 30th, 2024; 11:59 PM PST
License to publish Deadline August 16th, 2024; 11:59 PM PST
TGI3 workshop date October 10th, 2024 (MICCAI 2024 Satellite Events Day 2)

Workshop Program

Time Topic
Location Jade room at Palmeraie Palace
13:30 - 13:40 Welcome Remarks by Chao Chen
13:40 - 14:25 Keynote 1: Aasa Feragen from Technical University of Denmark
  Extracting and analyzing curvilinear structures from low quality images
14:30 - 15:00 Oral Session – 15 min * 4 presentations
  Multi-Factor Component Tree Loss Function: A Topology-Preserving Method for Skeleton Segmentation from Bone Scintigrams by Anh Nguyen (Tokyo University of Agriculture and Technology), Jean Cousty (ESIEE), Yukiko Kenmochi (CNRS), Shigeaki Higashiyama, Joji Kawabe, Akinobu Shimizu
  ccDice: A Topology-Aware Dice Score Based on Connected Components by Pierre Rougé (Université Reims Champagne-Ardenne, CRESTIC & INSA Lyon, CREATIS), Odyssée Merveille (Creatis), Nicolas Passat (Université Reims Champagne-Ardenne)
15:00 - 15:45 Keynote 2: Moo K. Chung from University of Wisconsin-Madison
  Aligning Asynchronous Human Brain Networks through Persistent Homology
15:45 - 16:30 Coffee Break
16:30 – 17:30 (Main poster room) Poster Session

Poster Schedule

ID Title
1 Analyzing Brain Tumor Connectomics using Graphs and Persistent Homology
2 A Bispectral 3D UNet for Rotation Robustness in Medical Segmentation
3 Restoring Connectivity in Vascular Segmentations using a Learned Post-Processing Model
4 Exploitation of Mapper Algorithm in Neuroimaging Applications: A Novel Framework for Outcomes Prediction
5 Topological data analysis of resting-state fMRI suggests altered brain network topology in Functional Dyspepsia: A Mapper-based parcellation approach
6 P-Count: Persistence-based Counting of White Matter Hyperintensities in Brain MRI
7 Outlier Detection in Large Radiological Datasets using UMAP
8 A Topological Comparison of the Fluorescence Imitating Brightfield Imaging and H&E Imaging
9 Topological Analysis of Seizure-Induced Changes in Brain Hierarchy Through Effective Connectivity

Submission Guidelines

Camera-ready submission

Please follow the MICCAI’24 camera-ready guidelines to prepare your final version: https://conferences.miccai.org/2024/en/CAMERA-READY-SUBMISSION-GUIDELINES.html. Specifically, we require the following files (submit them to the CMT system via a single zip file):

Contact

Feel free to email Xiaoling Hu (xihu3@mgh.harvard.edu) or Chao Chen (chao.chen.1@stonybrook.edu) if you have any questions regarding the workshop.

Call For Submissions

The significant advances in computational and data science over the past decade have had an immense impact on biomedical science and healthcare. Concurrently, researchers in the biomedical fields now face new challenges caused mainly by the nature of complex, often high-dimensional, noisy and diverse datasets.

Recent years have witnessed an increasing interest in the role topology plays in machine learning and data science. Topology offers a collection of topology-based techniques that have matured to a field known today as Topological Data Analysis (TDA). TDA provides a general and multi-purpose set of robust tools that have shown excellent performance in several real-world applications. These tools are naturally applicable to numerous types of data including images, points cloud, graphs, meshes, time-varying data and more. TDA techniques have been increasingly used with other techniques such as deep learning to enhance the performance, expressiveness, and generalizability of generic learning tasks. Furthermore, the properties of the topological tools allow discovering complex relationships and separating signals that are hidden in the data due to noise. In addition, TDA has an immense ability to extract unique features that are persistent over multiple scales. Finally, topological representations naturally lend themselves to insightful visualization making them useful for tasks that require interpretability and explainability. In particular, TDA tools can offer insights and interpretability on the data that are not readily available in main-stream computational tools such as deep learning techniques. The interpretation and explainability properties of TDA could help enhance the trust of medical professionals in intelligent systems.

All these properties of topological-based methods strongly motivate the adoption of TDA tools to various applications and domains including neuroscience, bioscience, biomedicine, and medical imaging.

This workshop will focus on using TDA techniques to enhance the performance, generalizability, expressiveness, efficiency, and explainability of the current methods applied to medical data. In particular, the workshop will focus on using TDA tools solely or combined with other computational techniques (e.g., feature engineering and deep learning) to analyze medical data including images/videos, sounds, physiological, texts and sequence data. The combination of TDA and other computational approaches is more effective in summarizing, analyzing, quantifying, and visualizing complex medical data.

This workshop will bring together mathematicians, biomedical engineers, computer scientists, statisticians and medical doctors for the purpose of showing the strength of using TDA-based tools for medical data analysis. It will also expose current applications, and provide an interdisciplinary forum for the exchange of ideas on novel applications, current challenges, and future directions.

We welcome submissions that present how TDA techniques, slowly or combined with other computational techniques, have been, or potentially could be, employed to tackle interesting problems in several areas of medical data computing and computer-assisted intervention. Topics of interest include, but are not limited to:

  1. Ensemble of topological and deep learning models for medical applications

  2. Topological-based approaches for disease diagnosis, monitoring, and prediction

  3. Topological-based approaches for classification and segmentation (e.g., level sets, graph cuts and fuzzy connectedness)

  4. Topological-based approaches for medical signal processing (e.g., images, audio, texts, etc.)

  5. Topological-based approaches for personalized medicine

  6. Shape models and analysis for medical data

  7. Topological-based learning and optimizations for medical applications

  8. Topological-based medical data registration, summarization, and enhancement

  9. Explainability, interpretability, and visualization of medical data

  10. Scalable TDA methods for medical records

  11. Topological structures for the analysis of biomedical data

Organizers

Chao Chen Bjoern Menze Yash Singh
Xiaoling Hu Johannes C. Paetzold Saumya Gupta
Colleen Farrelly Quincy Hathaway Rahul Paul
Paul Rosen Jennifer Rozenblit