Publications

Hereunder are three of my previous publications that you may want to check out. Don’t hesitate to have a look at my scholar profile to see my full list of publications.

My Thesis: Advancing automation in radiation therapy through artificial intelligence: decision support and uncertainty estimation

AI applied to cancer research is a dream but patient life is at stake so AI must be safe. This thesis contributes by proposing radiation therapy dose prediction model architectures and extending applications such as treatment decision support and automatic treatment planning. We built different quality assurance methods and designed a deployment interface.

Can input reconstruction be used to directly estimate uncertainty of a dose prediction U‐Net model?

In this study, we modify the CNN (HDU-Net) model architecture to obtain a direct uncertainty estimation method and apply it for a radiotherapy dose prediction. While simple, it allows to successfully flag out-of-distribution data and give information on the quality of the output in a single pass. This work was selected as ICCR Rising star competitor. You can check out the code repository.

PARROT: An end-to-end open source workflow of AI-assisted treatment planning and decision support

Here, we present PARROT: a user-friendly graphical interface built to deploy AI models for research clinicians. It is a free, open-source web platform that facilitates the use of AI delineation and dose prediction models and the visualization of the models outputs. The treatment decision support shows clinical evaluation tools to compare dose distributions and estimate of normal tissue complication probabilities.