The stability of anatomical and centroid reference points in cephalometric analysis.

Cephalometric records of 60 fetuses were combined on a coordinate reference grid to measure the statistical spread of the outlines. When centroid instead of anatomical points were used for superimposition of records, the spread of the outlines was reduced dramatically, being statistically significant (p less than 0.001). The centroid is a mean point being less variable and preferable to anatomical points commonly used for comparing successive cephalometric records.