Unity Biobank
“The Best Data, Annotated by the Best Experts”
Creating a unique AI-targeted biobank of expertise, supported by the
BHF, BSE, and NIHR.
Open Projects
Available Datasets
Description
We are building the UK’s largest and most diverse open-access biobank of echocardiographic images and annotations.
This collaborative effort fuels the development of AI algorithms that revolutionise cardiac care through improved image processing, acquisition, and interpretation.
Project Goals
- Establish the largest UK open-access biobank of echocardiographic images.
- Support AI research in image processing, acquisition, and interpretation.
- Ensure diversity and quality in annotated datasets.
Methods
Active Learning
We select cases using active learning, prioritising uncertain or informative samples.
This reduces annotation effort while maximising AI model improvement.
👉 Read more (CMPB 2024)
Quality Ranking
For image quality projects, we apply the Glicko-2 rating system, originally designed for competitive games, to dynamically rank image quality from expert feedback.
👉 Read more (JMAI 2022)
Unity Labelling Platform
Accessible across Safari (macOS/iOS) and Chrome (Windows/Linux/Android).
Secure Google login required.
👉 unityimaging.net
Who Should Participate?
- Echocardiographers: Expertise in interpretation.
- Cardiologists: Assess diagnostic impact of foreshortening, border visibility, etc.
- Sonographers: Provide acquisition expertise and optimisation insights.
Your Role
- Review echocardiographic videos/images presented in structured projects.
- Assess & Rank according to project guidelines (training videos provided).
- Provide feedback (optional) to refine annotation protocols.
Work With Us
- Contribute to groundbreaking research in cardiac imaging AI.
- Earn competitive compensation for your time and expertise.
- Flexible schedule — work remotely, set your hours.
- Make a real difference in patient care worldwide.
- Expand your professional network by collaborating with leading experts.
- Enhance your skills in annotation and medical imaging.
Open Projects
Image Quality – A4C View
Image Quality – PLAX View
Image Classification
Completed Projects
Available datasets (images and labels):
- TTE47 — Dataset Card & Reference Gallery
- UnityLV-MultiX — Multi-vendor, Multi-view Left Ventricular Echocardiography Dataset
- UNITY EchoPhase — Multibeat Echocardiographic Phase Detection
- UNITY MInD — Mitral Inflow Doppler Dataset
- UNITY EchoTDI — Automated Multibeat Tissue Doppler Echocardiography Analysis
Code Repositories
All datasets and supporting code are available at our GitHub:
👉 github.com/intsav
Professor Massoud Zolgharni
University of West London
📧 Massoud.Zolgharni@uwl.ac.uk
Professor Darrel Francis
Imperial College London
📧 d.francis@imperial.ac.uk
