Saturday 17 March 2018

The medical community has access to a wide range of medical tests and imaging studies. However, for diagnosis and radiation therapy treatment planning, standard tests, typically governed by outdated funding guidelines, lead to decisions in treatment approaches that are based on ‘population average’ data leading to a one-size fits all approach to diagnosis and treatment. Radiomics is a rapidly growing field that leverages quantitative features in medical images that can be incorporated into diagnostic, prognostic and predictive models to aid decision making and personalised patient care. These models are built by accumulating masses of imaging data and correlating with ground truth information (such as tissue specimens) or treatment outcomes. Going well beyond identifying the location of a tumour for targeted radiotherapy, radiomics has the ability to identify biological features and heterogeneity within the tumour to facilitate a voxelwise approach to treatment. In this presentation we will look at the basic principles of machine learning techniques that are used for deriving the mathematical models, some clinical applications, limitations of current approaches and future opportunities. Empathy in action: testing, treating, caring Julie McCrossin, Sandra Turner Invited Speakers