Statistical Concepts in Modern Pathology Practice

Abstract

As pathologists, we have long used data to cluster diseases into different classes. For example, we use the morphologic characteristics of a tissue to classify it into benign or malignant and even further sub-classify it into meaningful disease categories. We are facing an exponential growth in diagnostic and prognostic information available to us, the size of the data available to us makes it next to impossible for humans to be able to use most of this information in a meaningful way. Thus, computer algorithms and statistical models have been developed that can deal with this so called ‘big data’. Some of these algorithms are clustering algorithms which either summarize the data or classify and categorize the data.

In this chapter, we will briefly introduce two of the most common clustering approaches used in classifying and clustering patients: K-means clustering and hierarchical clustering.