Statistics for nuclear medicine. Introduction, descriptive statistics, and graphic displays.

Because the field of statistics has become so important in medicine, it is well worth the time of a physician in nuclear medicine to become acquainted with the lan guage of statistics, the elementary concepts, and a few of the more commonly used procedures. For this purpose we are presenting these essentials in a series of six short, nontechnical papers. Obviously, what can be accom plished in the brief space allotted each of them is limited. However, the reader can expect to gain an understanding of what statistics is, an ability to understand the statis tical aspects of much of the medical literature, a feel for when it will be necessary to consult with a statistician, and—for those occasions—an ability to communicate effectively with him. Unfortunately, such an elementary acquaintance as weoffer may lead a reader to overestimate his statistical capabilities and fail to consult a statistician in under taking a research effort. We do not believethat this series of papers, or any review of statistics at the introductory level, will enable anyone to proceed without professional assistance in medical research requiring statistical ex pertise. Medical research studies may be classified into two broad categories. Descriptive studies are intended to describe the characteristics of only the study group, using observations obtained from every member of the group. Inferential studies, on the other hand, are designed to enable the investigator to use observations from selected individuals (a sample) to form conclusions about the larger group (population) from which they were drawn. First, we will deal with descriptive studies, focusing on summary statistics (such as the mean and median) and graphic techniques (such as histograms and scatter diagrams). We then describe how one may estimate characteristics of the population from characteristics of