Clinical Trial Details
— Status: Active, not recruiting
Administrative data
| NCT number |
NCT05775068 |
| Other study ID # |
ARGOS |
| Secondary ID |
|
| Status |
Active, not recruiting |
| Phase |
|
| First received |
|
| Last updated |
|
| Start date |
July 1, 2021 |
| Est. completion date |
December 1, 2024 |
Study information
| Verified date |
March 2024 |
| Source |
Maastricht Radiation Oncology |
| Contact |
n/a |
| Is FDA regulated |
No |
| Health authority |
|
| Study type |
Observational
|
Clinical Trial Summary
Identifying the outline of a Gross Tumour Volume (GTV) in lung cancer is an essential step in
radiation treatment. Clinical research, such as radiomics and image-based prognostication,
requires the GTV to be pre-defined on massive imaging datasets. The ARGOS community creates
an open-source and vendor-agnostic federated learning infrastructure that makes it possible
to train a deep learning neural network to automatically segment Lung Cancer GTV on computed
tomography images. To reduce risks associated with sharing of patient data, we have used a
data-secure Federated Learning paradigm known as the "Personal Health Train" that has been
jointly developed by MAASTRO Clinic and the Dutch Comprehensive Cancer Organization (IKNL).
The successful completion of this project will deliver a highly scalable and readily-reusable
framework where multiple clinics anywhere in the world - large or small - can equitably
collaborate and solve complex clinical problems with the help of artificial intelligence and
massive amounts of data, while reducing the barriers associated with moving sensitive patient
data across borders.
Description:
Lung cancer (LC) is the single leading cancer cause of death worldwide (age-standardized rate
of 18.5 per 100,000 population), outstripping the mortality from cancers of the breast,
gastro-intestinal tract and reproductive organs. Radiotherapy (RT), often in combination with
other treatments, has an essential role in managing LC. An essential step in the RT process
is to draw the outline of the Gross Tumor Volume (GTV) in the lung on axial computed
tomography (CT) scans. The step is required for precisely directing tumoricidal radiation to
the target, and simultaneously avoiding irradiation of adjacent healthy tissue as much as
reasonably achievable.
However, tumor outlining by hand consumes a large amount of expert physician time, and has
demonstrably high levels of inter- and intra-observer variability. Part of a clinical
solution would require validated automated systems that work well for complex GTVs in a wide
variety of clinical settings. In recent times, a subclass of artificial intelligence known as
deep learning neural networks (DLNNs) has shown promising potential to assist clinicians for
such image processing tasks. The immense appeal of DLNN-based tools, if they can be safely
shown to add value into radiotherapy clinical workflow, is easily understandable - these have
the potential to significantly boost the productivity of clinicians by automating a portion
of labor-intensive work.
In respect to LC, models trained on selective data from few institutions are the norm. What
the field lacks is not simply large sample size, but sufficient diversity and heterogeneity
of subjects to represent the real world, and the means to train a DLNN on such a population.
That such a population exists among all the RT clinics around the world is indisputable,
however the question is how do we utilize data from all over the world for such a purpose.
"Federated Learning" very clearly addresses this by side-stepping a few of the administrative
complication of transferring individual-patient level data across national borders. Federated
learning is an implementation of the Personal Health Train (PHT) paradigm, where we send
research questions to each other in the form of software and exchange anonymous statistical
results (such as a DLNN model) instead of sending patient data around. Hence PHT addresses
two of the major challenges of using large-scale cancer data at a single stroke: (a) using
data for a good purpose in spite of the geographic dispersion of oncology data, and (b)
reducing privacy concerns associated sharing of private patient data across borders.
Objective
Project ARGOS will demonstrate how some of the infrastructural challenges of federated deep
learning and early clinical feasibility barriers to an LC GTV DLNN-based automated
segmentation model might be developed using a PHT approach. ARGOS adopts a global,
cooperative, vendor-agnostic and inter-disciplinary approach to AI development using
decentralized imaging datasets. As our first starting step, we will focus on less complex
clinical cases where the LC primary GTV is mostly contained inside the lung.
ARGOS plans to use existing radiotherapy planning CT delineations from several leading
radiotherapy centres throughout Europe, Asia, Oceania and North America. No new patient data
will be required because all the existing data already resides inside RT clinics as a result
of standard-of-care treatment.
The initial objective will be to train a DLNN that automatically segments the LC primary GTV
that is mostly or entirely contained in the lung parenchyma. The ARGOS partners will also
independently validate the globally-trained model on holdout validation and external test
datasets.
Sub-objectives
1. Share know-how among radiotherapy centres around the world for setting up the required
radiotherapy imaging data and metadata as "FAIR imaging data stations".
2. Offer a vendor-neutral and platform-agnostic open-source architecture for global
federated deep learning ("secure tracks").
3. Provide a registration and credentialing procedure for packaging deep learning
algorithms as a docker container software application ("docker trains").
4. Define a project governance structure and standardized operational principles, including
collaborative research agreements, data protection and intellectual property
valorization.
