DPU - Symposium 2022

Medica l Image Ana lysis & Ar t i f icia l Intel l igence Symposium 2022 Curriculum Vitae: Dr. Lorena Escudero Sánchez is an interdisciplinary researcher, specialised in Artificial Intelligence (AI), Machine Learning (ML), Deep Learning (DL), Data Science, Image Analysis and software development, applying the scientific vision and experience gained as a Particle Physicist to develop advanced cancer imaging techniques that will make an impact on society. Lorena is a Turing Fellow of The Alan Turing Institute, a Borysiewicz Interdisciplinary Fellow of the University of Cambridge, and a Rokos PDRA of Queens’ College Cambridge. She graduated from the University of Salamanca with a degree in Physics and a Master in Cosmology and Particle Physics. After a short time working at CERN, she then moved to Valencia with an FPU fellowship to complete her PhD in Neutrino Physics, working on the T2K experiment in Japan. In 2016 she moved to Cambridge and worked as a Research Associate at the Cavendish Laboratory, working mainly on pattern-recognition algorithms and related software development. She has been awarded numerous scholarships and awards, including the 2016 Breakthrough Prize in Fundamental Physics (as part of the T2K collaboration) and the PhD Award of Excellence by the University of Valencia. Lorena has led teams within worldwide collaborations, e.g. as a convenor of the Simulation and Reconstruction working group of the DUNE experiment(1000+ members), and managed international initiatives such as the “UK-Latin American Neutrino Initiative” (2018–2019). Abstract: Assessing the robustness of Radiomic features for generalisable prediction outcomes Radiomic image features are becoming a promising non-invasive method to obtain quantitative measurements for tumour classification and therapy response assessment in oncological research. These features include characteristics related to the distribution of the pixel values, as well as shape measurements and analysis of first and higher order textures, and convey information from imaging data that may not be visible to the human eye, thus providing valuable quantitative, reader-independent information. However, despite its increasingly established application and potential in precision oncology, there is still a need for standardisation criteria, as well as further validation of feature robustness with respect to imaging acquisition protocols, reconstruction parameters and patient demographics. In this talk, I will discuss the studies that we have carried out in our group in a variety of cancers (hepatic, renal and ovarian), in order to establish the robustness and reliability of CT-derived radiomic features. I will also present use-cases in which the impact of these studies has been further analysed in terms of generalisability of prediction outcomes in real-world datasets. Finally, I will mention the efforts to make the extraction of these features, and predictions based on them, accessible in tools used by radiologists. © HUEBL Dr. Lorena Escudero Sanchez, PhD Senior Research Associate, University of Cambridge Borysiewicz Interdisciplinary Fellow, University of Cambridge Turing Fellow, The Alan Turing Institute

RkJQdWJsaXNoZXIy NTA3MTM=