UCL Centre for Medical Image Computing


Benjamin Billot and Lydia Neary-Zajiczek - CMIC/WEISS Joint Seminar Series

03 November 2021, 1:00 pm–2:00 pm

Benjamin Billot and Lydia Neary-Zajiczek - talks as part of CMIC/WEISS Joint Seminar Series

Event Information

Open to





UCL Centre for Medical Image Computing and Wellcome/EPSRC Centre for Interventional and Surgical Sciences

Speaker: Benjamin Billot

Title: SynthSeg: Domain Randomisation for Segmentation of Brain MRI Scans of any Contrast and Resolution

Abstract: When applied to segmentation of brain MRI scans, convolutional neural networks (CNNs) have difficulties generalising to new target domains, especially with unseen contrast and resolution. Here we introduce SynthSeg, a CNN trained with synthetic data sampled from a generative model inspired by Bayesian segmentation. Crucially, we adopt a domain randomisation strategy where we fully randomise the generation parameters to maximise the variability of the training data. Consequently, SynthSeg can segment preprocessed and unpreprocessed real scans of any target domain, without retraining or fine-tuning. We demonstrate SynthSeg on 5,500 scans of 6 modalities and 10 resolutions, where it exhibits unparalleled generalisation compared to supervised CNNs, test time adaptation, and Bayesian segmentation.


Speaker: Lydia Neary-Zajiczek

Title: Image Resolution In Digital Pathology: How Important Is It?

Abstract: The digitization of pathology is a key mitigation strategy for the critical staffing shortages facing most departments, but despite its clear benefits, adoption has been limited. Historic perceptions of inferior digital image quality has led to samples being digitized at very high spatial resolution, increasing costs and resulting in enormous digital files that are challenging to work with, both for pathologist assessment and for training automated image classification tools. In this talk I will describe image resolution from an optical physics perspective, in comparison to how this term is generally used when describing annotated microscopy datasets in the context of training image classification algorithms. I will discuss how optical resolution and sampling resolution differ, particularly how these parameters define the information content of a microscopy image.


Chair: Evans Mazomenos