Abstract: This invention teaches a method to identify cellular abnormalities which are associated with disease states. The method utilizes infrared (IR) spectra of cell samples which are dried on an infrared-transparent matrix and scanned at the frequency range from 3000-950 cm−1. The identification of samples is based on establishing a reference using a representative set of spectra of normal and/or diseased specimens. During the reference assembly process, multivariate techniques such as Principal Component Analysis (PCA) and/or Partial Least Squares (PLS) are used. PCA and PLS reduce the data based on maximum variations between the spectra, and generate clusters in a multidimensional space representing the different populations. The utilization of Mahalinobis distances, or linear regression (e.g., Principle Component Regression on the reduced data from PCA) form the basis for the discrimination.