Differentiation of human hair by colour and diameter using light microscopy,digital imaging and statistical analysis

This research introduces and evaluates a novel method that offers the potential of providing objective criteria to forensic microscopical hair comparisons. The method combines hair diameter with numeric characterisations of red, green, and blue colour content as determined with the use of digital imaging at defined locations of the hair. Thirty hairs were collected from each of twenty participants, all with naturally coloured brown hair. The hairs were examined with an Olympus BX53® polarising light microscope and digital images were viewed with an Olympus DP72® camera under 400× magnification. Using Olympus cellSens^? Entry software, hair diameter was measured at 1000, 1500 and 2000 m from the base of the root. The Olympus cellSens^? Entry software uses a red, green and blue (RGB) colour model to quantitatively define the colour of each pixel on an image based on its composition of these three principal colour components. This software was used to collect numerical characterisations of hair colour at each distance interval. The diameter and colour values for each hair were compared using discriminant analysis (DA) and principal component analysis. Although a large amount of intrapersonal variation was observed, the degree of interpersonal variation was greater and enabled the statistical model to differentiate between the hair samples from each participant. The DA model achieved sample reclassification with an error rate of 7.33%. A validation study was conducted on a subset of hair samples from which 18 of the 20 were correctly assigned to the participant from whom they originated. These results support the potential of this method to provide an objective addition to current microscopical hair comparison practices.