A Quantitative Approach to the Analysis of Human Hair for Forensic Examinations

Hair is frequent trace evidence found in crime scenes due to its high transfer potential due to natural shed. They can provide investigative information in various cases, including homicide, sexual assault, missing persons, and human trafficking. Physical examination of hair involves the analysis of microscopic features, such as color and pigmentation, commonly using bright field microscopy. Microscopical techniques are non-destructive, with minimum to no sample preparation that can be followed by chemical and biological analysis. Although individualization is not possible via microscopical methods, it can be used to include or exclude potential donors. However, previous microscopical techniques lack quantitative and statistical approaches. This research uses quantitative and statistical approaches to increase discrimination power microscopical examinations. The inter- and intra-sample variance of hair features is further explored. Statistical methods, such as ANOVA, are used to analyze variable importance and classification models. Random Forest (RF) and Convolution Neural Networks (CNNs) were used to discriminate between sample populations and individuals. Additionally, the effect of hair dyes is explored. Non-destructive dye extraction methods were developed, and chemical analysis using Direct Analysis in Real Time Mass Spectrometry (DART-MS) was used. The chemical profile of the extracted dyes compared to the original dye and dye extracts. Heats maps and Pearson correlations were used to obtain similarity scores between dyes. This project aims to increase the value of microscopical examination and develop non-destructive chemical analysis methods for hair dyes.