Optical processing for least mean-squares analysis of biomedical patterns

An optical method for the analysis of coherent light scattering patterns from samples of whole blood or urine is being investigated. The purpose of such analysis is the customary one of inferring counts of the numbers of particulates in each of several chosen classes. The linearity of expected light intensity with respect to numbers of scatterers permits their least mean-squares estimation using a generalized matrix inverse. Implementation of this inversion is by means of a mask operating directly on the light patterns to give immediate minimum-variance linear unbiased estimates.