Brain MRI from a subject with Alzheimer's disease, from the Open Access Series of Imaging Studies (OASIS) dataset.

Organ Shapes

Organ shapes contain pivotal information about our health. For instance, characterizing brain shapes allows clinicians to quantify the progression of Alzheimer's disease. Methods of automatic diagnosis often compare a new subject's organ shape with a pre-computed "template" organ shape, i. e. the standard shape for a healthy subject. 

Our research explores statistical learning on organ shapes, and:

  • studies methods allowing to compute with organ shapes in the small data regimes (smaller dataset sizes),
  • quantifies the uncertainty associated to the computation of an organ's template, which impacts the uncertainty associated with a computational diagnosis relying on it.


Check out our mathematical formulation of this statistical issue!


data sources in medicine

Computer-assisted diagnosis requires merging heterogeneous medical data sources.

BioShapes in Computer-Aided Diagnosis

Computer-aided diagnosis has shown stellar performance in providing accurate medical diagnoses from biomedical images. Yet this field typically focuses on harvesting the signal provided by a single (and often, very accurate) imaging modality.

Our research delves into statistical approaches to computer-aided diagnosis that:

  • integrates heterogeneous sources of information: from biomedical images and shapes to demographic data,
  • quantifies the uncertainty on the data sources,
  • outputs computer-aided diagnosis with associated notion of reliability.

Want to learn more?

Check out our solution for Covid-19 diagnosis, that won the 1st prize at the Covid-19 Grand Challenge (100k$)!

- Look at the research of our collaborator Claire Donnat!


Relevant Publications