Derivative domain fitting: a new method for resolving a mixture of normal distributions in the presence of a contaminating background

Cytometry. 1993;14(5):510-8. doi: 10.1002/cyto.990140510.

Abstract

Derivative domain least squares analysis is a new method for resolving multiple peaks superimposed on a slowly varying continuum into separate normal (Gaussian) distributions without developing a functional approximation for the continuum. The method is based on fitting the first derivative of the data with the first derivative of the sum of a series of normal distributions. A functional approximation for the continuum is not necessary as long as the first derivative of the continuum is approximately zero (i.e., the continuum varies slowly compared to the normal distributions).

MeSH terms

  • Background Radiation*
  • Chromosomes / ultrastructure
  • Flow Cytometry / methods*
  • Flow Cytometry / standards
  • Karyotyping
  • Normal Distribution*
  • Statistics as Topic / methods*