Automating Forensic Analysis

Project description

Detection, collection and analysis of DNA from crime scenes is of vital importance for the criminal justice system. Touch DNA samples make up the majority of biological evidence encountered in forensic investigations. For example, in 2020 over 60% of all DNA samples analysed in Forensic Science South Australia were touch DNA. Touch DNA is latent, and thus not visible to a naked eye. However, once stained cells can be made to fluoresce allowing for analysis. Manual counting of cells is time consuming and susceptible to subjective bias. Cell counting software is available, but has many settings that can be tricky to set up for people unfamiliar with image processing. Furthermore, questions remain as to if the software is accurately counting cells and if anything that fluoresces is, in fact, a component of touch DNA. This is because touch deposits contain a variety of different types of cellular and non-cellular material, all of which contain different quantities of DNA. Not all of which are of equal relevance to the investigation. This project will solve a number of issues: 1) Optimise software to automatically identify fibres and other artefacts as well as set analysis thresholds for an automated and more accurate count. 2) Gather data, based on individuals depositing their own touch deposits on a surface, to teach the software what is and what is not a cell based on fluorescence intensity (i.e.  related to DNA amounts) and the shape and size of the cells. 3) Determine if it is possible to distinguish between different types of touch cells.