EFOMP - European School for Medical Physics Experts (ESMPE) Statistics edition 2020
Start Date: Apr 23rd, 2020 09:00
End Date: Apr 25th, 2020 14:00
Location: Athens, Greece
EFOMP’s European School for Medical Physics Expert (ESMPE) has been established to fulfil one of the missions of EFOMP: to promote education and training programmes for Medical Physicists in Europe.
The 'Statistics in Medical Physics' school will cover the methods of inferential statistics most frequently used in the medical field, the statistical methods used in radiomics, the treatment of errors and uncertainties in radiation dosimetry. Much in the course is relevant to the work of the medical physicist, whether they are working in imaging, therapy or protection, as well as giving a good grounding in the basic principles of statistics.
What skills will you learn?
The school opens with consideration of how to design an experiment and analyse the resulting data. For a medical physicist an important question is often how one evaluates a diagnostic test. This is addressed by a session dealing with the trade-off between sensitivity and specificity utilising ROC analysis. If you want to examine the source of differences between two or more variables in an experiment, then you will need to use applied regression analysis. In the afternoon the lecturers will talk about the application of analysis of variance and of covariance together with worked examples from medical physics. If the dependent variable is binary (e.g. alive/dead, yes/no) then logistic regression is needed and a discussion of this, together with examples from visual grading experiments, completes the day.
If your experiment involves two or more possible variables, then data analysis requires the application of multiple linear regression This, together with worked examples, forms the start of the second day. How you assess the success of treatment may be addressed by the use of survival analysis and I will be talking about how to create and interpret survival curves.
The school finishes with a series of presentations on radiomics, dosimetry and radiotherapy. In radiomics the question of feature evaluation and quality by statistical methods is explored. Quality is again a theme in dosimetry, looking at how uncertainties and the related quality of measurements are handled using statistical techniques. In radiotherapy the problem of agreement such as in dose distributions and volume estimations are investigated.
Registration: EFOMP website