Fast Uncertainty Quantification for Deep Learning-based MR Brain Segmentation

Benjamin Lambert (1, 2) , Florence Forbes (3) , Senan Doyle (2) , Alan Tucholka (2) , Michel Dojat (1)

1 GIN - [GIN] Grenoble Institut des Neurosciences
2 Pixyl Medical [Grenoble]
3 STATIFY - Modèles statistiques bayésiens et des valeurs extrêmes pour données structurées et de grande dimension

Quantifying the uncertainty attached to Deep Learning models predictions can help their interpretation, and thus their acceptance in critical fields. Yet, current standard approaches rely on multi-steps approaches, increasing the inference time and memory cost. In clinical routine, the automated prediction has to integrate into the clinical consultation timeframe, raising the need for faster and more efficient uncertainty quantification methods. In this work, we propose a novel model, named as BEHT, and evaluate it on an automated segmentation task of White-Matter Hyperintensities from T2-weighted FLAIR MRI sequences of Multiple-Sclerosis (MS) patients. We demonstrate that this approach outputs predictive uncertainty much faster than the state-of-the-art Monte Carlo Dropout approach, with a similar-and even slightly better-accuracy. Interestingly, our approach distinguishes 2 distinct sources of uncertainties, namely aleatoric and epistemic uncertainties.

Domaines: Statistiques [math.ST]

Benjamin Lambert, Florence Forbes, Senan Doyle, Alan Tucholka, Michel Dojat. Fast Uncertainty Quantification for Deep Learning-based MR Brain Segmentation. EGC 2022 - Conference francophone pour l'Extraction et la Gestion des Connaissances, Jan 2022, Blois, France. pp.1-12. ⟨hal-03498120⟩