Exploring Population Metadata

Summary

Some metadata may come from the DICOM headers, patient history, and image labels. Once we have all of a dataset's metadata stored in a single place, we'll then want to explore data features.

Histograms

Histograms help us look at distributions of single variables. Sometimes we only want to look at distributions within a single class of our data.

Scatterplots

Scatterplots are useful for assessing relationships between two variables.

Pearson Correlation Coefficient

Pearson Correlation Coefficient measures how two variables are linearly related. The value ranges from -1 to 1. A value of 1 or -1 means the two variables are perfectly linearly related. A value of 0 implies there is no linear relationship between the two variables.

Co-Occurrence Matrices

Co-Occurrence Matrices are useful for assessing how frequently different classifications co-occur together.