The aim of this book is to present statistical problems and methods in a friendly way to radiologists, emphasizing statistical issues and methods most frequently used in radiological studies (e.g., nonparametric tests, analysis of intra- and interobserver reproducibility, comparison of sensitivity and specificity among different imaging modality, difference between clinical and screening application of diagnostic tests, ect.). The tests will be presented starting from a radiological "problem" and all examples of statistical methods applications will be "radiological".
Many radiological manuscripts are burdened by limitations relating to study design, definition of enrollment criteria and standard of reference, lack of preliminary estimation of sample size, and incorrect use of statistical tests. These limitations sometimes oblige the reviewer to recommend rejection of a manuscript more frequently than would be the case if rejection were based solely on inherent errors in technical performance or interpretation of imaging modalities.
An underlying aim of this book is to facilitate the interaction between radiologists and statisticians by explaining the basics of statistical methods as applied to medical imaging. The intention is to help develop a way of thinking beyond merely how to perform calculations and get p values. Thus, radiologists will be able to solve small statistical problems and, more importantly, to assume the right perspective during the crucial phase when a study is planned and undertaken.