Ben Glocker

TL;DR: Morpho-MNIST is introduced, a framework that aims to answer: "to what extent has my model learned to represent specific factors of variation in the data" and a set of quantifiable perturbations to assess the performance of unsupervised and supervised methods on challenging tasks such as outlier detection and domain adaptation.

TL;DR: In this paper, the capability of various state-of-the-art approaches for confidence-based OOD detection through a comparative study and in-depth analysis was evaluated using chest X-rays.

TL;DR: A frailty index specific to traumatic brain injury was developed and externally validated and could help to individualise rehabilitation approaches aimed at mitigating effects of frailty in patients withtraumatic brain injury.

TL;DR: It is shown how erroneous and noisy segmentation masks can be improved using Post-DAE, a post-processing method based on denoising autoencoders to improve the anatomical plausibility of arbitrary biomedical image segmentation algorithms.

TL;DR: Random Forest Regression Voting Constrained Local Models (RFRV-CLMs) are evaluated for vertebral fracture assessment and show that, while they lead to slightly poorer median errors than AAMs, they are much more robust, reducing the proportion of fit failures by 68\%.