Artificial Medical Intelligence Group (AMIGO) at King's College London (KCL) | AMIGO

Quality Control with Foundation Models for Radiology

Wed, 25 Mar 2026 00:00:00 +0000

Internship / Student Project Opportunity

We are developing foundation models for large-scale radiology data with a focus on robust representation learning and quality control.

The project explores how to learn general-purpose imaging representations that transfer across tasks such as scan-level quality assessment. We are particularly interested in self-supervised and weakly supervised learning strategies that can leverage large, heterogeneous radiology datasets.

During the project, students may work on:

pretraining and adapting radiology foundation models on large-scale imaging data;
learning transferable visual representations for downstream diagnostic tasks;
designing quality control pipelines to detect low-quality scans and out-of-distribution inputs;
evaluating robustness, calibration, and generalization across sites and cohorts.

We welcome applications from candidates with a strong machine learning and coding background (for example, Python/PyTorch, deep learning, and practical experience handling medical imaging data).

This is suitable as an internship or student research project for candidates interested in clinically relevant AI and translational machine learning.

How to apply

Please email the following documents:

CV
Academic transcript
A short motivation letter

Send your application to: yigit.avci@kcl.ac.uk

Contact

Mon, 24 Oct 2022 00:00:00 +0000

People

Mon, 24 Oct 2022 00:00:00 +0000

Federated Learning Interoperability Platform (FLIP)

Mon, 01 Jan 0001 00:00:00 +0000

FLIP is an open-source platform that links data from multiple NHS Trusts to enable federated training and evaluation of medical imaging AI models, while ensuring data privacy and security. Developed by the London AI Centre in collaboration with Guy’s and St Thomas’ NHS Foundation Trust and King’s College London, FLIP comprises three main components.

Secure Enclaves - Dedicated secure data storage within each partner NHS Trust’s firewall keeps sensitive patient data inside the Trust. Data from across the Trusts’ patient records systems is transferred into the secure enclave for curating and aggregation, unifying medical imaging scans from PACS and other electronic health data.

Interoperability and Data Harmonisation - Electronic healthcare records are complex and heterogeneous. FLIP uses ontological and data interoperability standards to structure and harmonise data across multiple hospitals and clinical systems, enabling AI algorithms to query, learn, and action data via an open standards-based interface.

Federated Learning and Evaluation - FLIP brings algorithms to the data within each NHS Trust’s secure enclave, without sharing information outside the secure firewall or breaking local governance rules. Algorithmic models are sent to multiple Trusts and trained on local data before being securely combined to achieve consensus. The platform supports both NVIDIA FLARE and Flower federated learning frameworks.

MONAI

Mon, 01 Jan 0001 00:00:00 +0000

MONAI is a PyTorch-based, open-source framework for deep learning in healthcare imaging. Its ambitions are:

developing a community of academic, industrial and clinical researchers collaborating on a common foundation;
creating state-of-the-art, end-to-end training workflows for healthcare imaging;
providing researchers with the optimized and standardized way to create and evaluate deep learning models.

Features

The codebase is currently under active development.

flexible pre-processing for multi-dimensional medical imaging data;
compositional & portable APIs for ease of integration in existing workflows;
domain-specific implementations for networks, losses, evaluation metrics and more;
customizable design for varying user expertise;
multi-GPU data parallelism support.

Installation

To install the current release:

pip install monai

To install from the source code repository:

pip install git+https://github.com/Project-MONAI/MONAI#egg=MONAI

Alternatively, pre-built Docker image is available via DockerHub:

# with docker v19.03+
docker run --gpus all --rm -ti --ipc=host projectmonai/monai:latest

Getting Started

Tutorials & examples are located at monai/examples.

Technical documentation is available via Read the Docs.

Contributing

For guidance on making a contribution to MONAI, see the contributing guidelines.

NiftyNet

Mon, 01 Jan 0001 00:00:00 +0000

NiftyNet is a TensorFlow-based open-source convolutional neural networks (CNN) platform for research in medical image analysis and image-guided therapy. NiftyNet’s modular structure is designed for sharing networks and pre-trained models. Using this modular structure you can:

Get started with established pre-trained networks using built-in tools
Adapt existing networks to your imaging data
Quickly build new solutions to your own image analysis problems

NiftyNet is a consortium of research organisations (BMEIS – School of Biomedical Engineering and Imaging Sciences, King’s College London; WEISS – Wellcome EPSRC Centre for Interventional and Surgical Sciences, UCL; CMIC – Centre for Medical Image Computing, UCL; HIG – High-dimensional Imaging Group, UCL), where BMEIS acts as the consortium lead.

Features

Easy-to-customise interfaces of network components
Sharing networks and pretrained models
Support for 2-D, 2.5-D, 3-D, 4-D inputs*
Efficient training with multiple-GPU support
Implementation of recent networks (HighRes3DNet, 3D U-net, V-net, DeepMedic)
Comprehensive evaluation metrics for medical image segmentation

NiftyNet is not intended for clinical use.

*2.5-D: volumetric images processed as a stack of 2D slices; 4-D: co-registered multi-modal 3D volumes

Installation

Please install the appropriate TensorFlow package:
- pip install "tensorflow==1.15.*"
pip install niftynet

All other NiftyNet dependencies are installed automatically as part of the pip installation process.

To install from the source repository, please checkout the instructions.

Documentation

The API reference and how-to guides are available on Read the Docs.

Useful links

Citing NiftyNet

If you use NiftyNet in your work, please cite Gibson and Li, et al. 2018:

E. Gibson*, W. Li*, C. Sudre, L. Fidon, D. I. Shakir, G. Wang, Z. Eaton-Rosen, R. Gray, T. Doel, Y. Hu, T. Whyntie, P. Nachev, M. Modat, D. C. Barratt, S. Ourselin, M. J. Cardoso† and T. Vercauteren† (2018) NiftyNet: a deep-learning platform for medical imaging, Computer Methods and Programs in Biomedicine. DOI: 10.1016/j.cmpb.2018.01.025

BibTeX entry:

@article{Gibson2018,
 title = "NiftyNet: a deep-learning platform for medical imaging",
 journal = "Computer Methods and Programs in Biomedicine",
 year = "2018",
 issn = "0169-2607",
 doi = "https://doi.org/10.1016/j.cmpb.2018.01.025",
 url = "https://www.sciencedirect.com/science/article/pii/S0169260717311823",
 author = "Eli Gibson and Wenqi Li and Carole Sudre and Lucas Fidon and
 Dzhoshkun I. Shakir and Guotai Wang and Zach Eaton-Rosen and
 Robert Gray and Tom Doel and Yipeng Hu and Tom Whyntie and
 Parashkev Nachev and Marc Modat and Dean C. Barratt and
 S\'{e}bastien Ourselin and M. Jorge Cardoso and Tom Vercauteren",
}

The NiftyNet platform originated in software developed for Li, et al. 2017:

Li W., Wang G., Fidon L., Ourselin S., Cardoso M.J., Vercauteren T. (2017) On the Compactness, Efficiency, and Representation of 3D Convolutional Networks: Brain Parcellation as a Pretext Task. In: Niethammer M. et al. (eds) Information Processing in Medical Imaging. IPMI 2017. Lecture Notes in Computer Science, vol 10265. Springer, Cham. DOI: 10.1007/978-3-319-59050-9_28

Licensing and Copyright

NiftyNet is released under the Apache License, Version 2.0.

Acknowledgements

This project is grateful for the support from the Wellcome Trust, the Engineering and Physical Sciences Research Council (EPSRC), the National Institute for Health Research (NIHR), the Department of Health (DoH), Cancer Research UK, King’s College London (KCL), University College London (UCL), the Science and Engineering South Consortium (SES), the STFC Rutherford-Appleton Laboratory, and NVIDIA.

VTrails

Mon, 01 Jan 0001 00:00:00 +0000

A vectorial representation of the vascular network, which embodies quantitative features such as location, direction, scale, and bifurcations, has many potential cardio- and neuro-vascular applications.

VTrails stands as an end-to-end approach to extract geodesic vascular minimum spanning trees from angiographic data by solving a connectivity-optimised anisotropic level-set over a voxel-wise tensor field representing the orientation of the underlying vasculature.

VTrails has been presented at the biennial conference Information Processing in Medical Imaging (IPMI 2017) [1] [Full-Text], and further published on the IEEE Transactions on Medical Imaging journal [2] [Full-Text].

Artificial Medical Intelligence Group (AMIGO) at King's College London (KCL) | AMIGO

Quality Control with Foundation Models for Radiology

Internship / Student Project Opportunity

How to apply

Contact

People

Federated Learning Interoperability Platform (FLIP)

MONAI

Features

Installation

Getting Started

Contributing

Links

NiftyNet

Features

Installation

Documentation

Useful links

Citing NiftyNet

Licensing and Copyright

Acknowledgements

VTrails