Details

Autor(en) / Beteiligte

• Buddenkotte, Thomas
• Escudero Sanchez, Lorena
• Crispin-Ortuzar, Mireia
• Woitek, Ramona
• McCague, Cathal
• Brenton, James D.
• Öktem, Ozan
• Sala, Evis
• Rundo, Leonardo

Titel

Calibrating ensembles for scalable uncertainty quantification in deep learning-based medical image segmentation

Ist Teil von

• Computers in biology and medicine, 2023-09, Vol.163, p.107096-107096, Article 107096

Ort / Verlag

United States: Elsevier Ltd

Erscheinungsjahr

2023

Quelle

MEDLINE

Beschreibungen/Notizen

• Uncertainty quantification in automated image analysis is highly desired in many applications. Typically, machine learning models in classification or segmentation are only developed to provide binary answers; however, quantifying the uncertainty of the models can play a critical role for example in active learning or machine human interaction. Uncertainty quantification is especially difficult when using deep learning-based models, which are the state-of-the-art in many imaging applications. The current uncertainty quantification approaches do not scale well in high-dimensional real-world problems. Scalable solutions often rely on classical techniques, such as dropout, during inference or training ensembles of identical models with different random seeds to obtain a posterior distribution. In this paper, we present the following contributions. First, we show that the classical approaches fail to approximate the classification probability. Second, we propose a scalable and intuitive framework for uncertainty quantification in medical image segmentation that yields measurements that approximate the classification probability. Third, we suggest the usage of k-fold cross-validation to overcome the need for held out calibration data. Lastly, we motivate the adoption of our method in active learning, creating pseudo-labels to learn from unlabeled images and human–machine collaboration. [Display omitted] •Existing approaches fail to approximate the classification probability.•Ensembles of models with differing sensitivity/precision express uncertainty.•Calibrated ensembles can approximate classification probability on unseen data.•Uncertainty quantification (UQ) might be used to better process or use unlabeled data.•Novel UQ method for pixel-wise probability heatmaps in 3D medical segmentation.