AI-powered tools tackling oesophageal cancer
Felix Zhou read Engineering at St John’s College, Cambridge graduating in 2013. He obtained his PhD at Oxford in machine learning and computer vision working on the analysis of biological motion in microscopy videos. He is currently a postdoctoral fellow in the Lu Lab at the Ludwig Cancer Institute. Felix is addressing two key clinical problems in Oesophageal Cancer (OeCa) using computational methods.
To improve the current poor outcomes experienced by patients with OeCa, research across Oxford is underway to allow earlier detection of early stage disease, and to address the issue of poor response rates to standard treatments. Felix’s research seeks to:
• Felix is working closely with collaborators in the Big Data Institute and John Radcliffe Hospital, developing AI-powered augmented reality (AR) tools to assist clinicians in accurately detecting and mapping the location of oesophageal tumours in endoscopy. A significant output of this work is a comprehensive image restoration framework for endoscopy frames corrupted by imaging artefacts. To support scientific reproducibility Felix and his colleagues  compiled the first comprehensive image dataset, collected from 6 international institutions and  established the Endoscopy Artefact Detection Challenge.
• Felix and colleagues are developing personalised treatments to overcome heterogeneity-driven resistance, a common problem with current standard treatments such as chemotherapy. This involves high-throughput drug screens using miniature organs grown from patient biopsies (organoids), and applying timelapse imaging to monitor growth and interaction dynamics. Felix develops computational analyses to characterize the spatio-temporal development of individual organoids in these screens to more specifically assess a tumour’s sensitivity to existing and new drugs. This builds upon a comprehensive motion analysis framework that Felix recently developed called Motion Sensing Superpixels (MOSES)  which has already been shown to capture subtle interaction changes between similar epithelial cell lines.
Find out more about our research below
1 Ali, Sharib, et al. “A deep learning framework for quality assessment and restoration in video endoscopy.” arXiv preprint arXiv:1904.07073 (2019).
2 Ali, Sharib, et al. “Endoscopy artifact detection (EAD 2019) challenge dataset.” arXiv preprint arXiv:1905.03209 (2019).
3 Zhou, Felix Y., et al. “Motion sensing superpixels (MOSES) is a systematic computational framework to quantify and discover cellular motion phenotypes.” eLife 8 (2019): e40162.