“SenNet + HOA – Hacking the Human Vasculature in 3D” Kaggle Competition is now LIVE
INDIANAPOLIS –The “SenNet + HOA – Hacking the Human Vasculature in 3D” Kaggle competition is now LIVE at https://www.kaggle.com/competitions/blood-vessel-segmentation.
The Cellular Senescence Network (SenNet) and the Human Organ Atlas (HOA) teams welcome contributions that help improve 3D segmentation of vasculature.
With better data on the healthy human vasculature, researchers will be able to simulate the flow of blood, oxygen, or even drugs through the vessel network. This is an exciting opportunity to get the world’s best machine learning experts aware of SenNet and to work together on tissue segmentation for vascular structures, upon which senescence neighborhoods appear to be related (in at least some tissues). The aim is to solve a fundamental problem in at least one organ, and to develop a first understanding of what vascular scaffold is needed to map senescent cells in tissue regions.
The 3D segmentation data will also be valuable for understanding how blood vasculature changes for different sex, age, and BMI. Ultimately, 3D vessel segmentation data will help pave the way towards a more complete Vascular Common Coordinate Framework (VCCF) and Human Reference Atlas (HRA) for the healthy human body.
A total prize amount of $80,000 (sponsored by Kaggle, CIFAR, and ThermoFisher) will be distributed among the five winning teams. You can join any time before January 30, 2024. Our welcome message here includes additional information and links. Please share this information widely!
Who: The competition is hosted by the National Institutes of Health Common Fund-supported Cellular Senescence Network (SenNet) Program, the Human Organ Atlas (HOA), and Indiana University. A partnership with Google is enabling the data to be shared with researchers worldwide to exchange ideas, best practices and shared information to help solve biomedical challenges.
Location/Date: Kaggle platform https://www.kaggle.com/competitions/blood-vessel-segmentation. Anyone is welcome to join before January 30, 2024. Competition ends on February 6, 2024.
Competition goal: Improve 3D segmentation of vasculature to compute the size, shape, branching angles and patterning of blood vessels in human tissue. This will make it possible to simulate the flow of blood, oxygen, or even drugs through the vessel network. The 3D segmentation data will also be valuable for understanding how blood vasculature changes during aging or disease. Ultimately, 3D vessel segmentation data will help pave the way towards a more complete Vascular Common Coordinate Framework (VCCF) and Human Reference Atlas (HRA) for the healthy human body.
Specifically: Participants are asked to segment blood vessels and create a model trained on 3D Hierarchical Phase-Contrast Tomography (HiP-CT) data from human kidneys. Our welcome message here includes additional information and links.
Prize Details: A total prize amount of $80,000 (sponsored by Kaggle, CIFAR, and ThermoFisher) will be distributed among the five winning teams.
The human body’s organs and tissues depend on the interaction, spatial organization and specialization of cells – all 37 trillion of them. Researchers make sense of cellular functions and relationships with the Vasculature Common Coordinate Framework (VCCF). The VCCF maps cells using the blood vasculature in the human body as the primary navigation system. The framework crosses all scale levels and provides a unique way to identify cellular locations using capillary structures as an address.
Currently, human expert annotators manually trace the vascular structures – a slow process. Even with expert annotators, each new dataset takes many months to complete. Machine learning approaches using this manual data do not generalize well to new datasets because of the variability of both human anatomy and to changes in the image quality as HIP-CT technology continues to improve and change.
Kaggle participants could improve the world’s understanding of the effect of vasculature on different cells in the human body. With better data, researchers could simulate the flow of blood, oxygen or even drugs throughout the vessel network. They could also use the available organ images to augment their understanding of how blood vasculature changes as sex, age and body mass index (BMI) change.
Ultimately, participants could pave the way toward a more complete VCCF and Human Reference Atlas (HRA) that could identify how the relationship between cells affect human health.
Senescence refers to a situation where cells stop dividing but release molecules absorbed by other cells. These molecular releases can help with human healing, cancer cell blockage and growth, but as humans age, senescence cells can accumulate so much they may cause harm and contribute to cardiovascular, lung, kidney, Alzheimer’s disease and arthritis. Senolytic medications can selectively kill senescence cells but researchers don’t know enough yet about senescence cells. Animal trials using the medications show promise, and human trials have been underway since 2021.
Senescence cells show a range of characteristics depending on what triggered them, the tissue origin of the cells and where they are located in the body. SenNet is working to develop maps of these cells to better understand them and develop effective therapeutics for them. SenNet will provide data and resources to the public that would otherwise be difficult to achieve through individual efforts, accelerating the ability of biomedical researchers to develop therapeutics that target cellular senescence and improve human health.
About the hosts:
SenNet was established to comprehensively identify and characterize the differences in senescent cells across the body, across various states of human health and across the lifespan. SenNet provides publicly accessible atlases of senescent cells and develops innovative tools and technologies that build upon previous advances in single-cell analysis. Learn more: https://commonfund.nih.gov/senescence
HOA is a digital atlas containing 3D multi-resolution imaging datasets, created at the world’s brightest synchrotron, the European Synchrotron Radiation Facility. Using an imaging technique called Hierarchical Phase-Contrast Tomography (HIP-CT). HIP-CT spans a previously poorly explored scale in researchers’ understanding of human anatomy, from microns to entire intact organs. The technology enables researchers to see into the human body at very high resolutions and generate huge amounts of data more quickly than could traditionally be generated. Learn more: https://www.esrf.fr/HUB/HOAHub
Media contact: Nancy Larner Ruschman, Project Manager, Cyberinfrastructure for Network Science Center at the Luddy School of Informatics, Computing, and Engineering
Department of Intelligent Systems Engineering at Indiana University