Automated scoring of Sperm Hy-Liter™-stained spermatozoa by the MetaSystems Metafer image analysis software system in sexual assault specimens

Abstract

The MetaSystems Metafer image analysis software system was purchased three years ago in the hope of developing a routine approach in the RCMP Forensic Laboratories to automate the scoring of human spermatozoa in sexual assault exhibits. This would enhance case throughput, increase assay sensitivity and standardize the search for spermatozoa. The development of appropriate classifiers was challenging but essential to teach the software system to specifically recognize human spermatozoa fluorescently stained using the Sperm Hy-Liter™ kit (Independent Forensics). Optimized classifiers were tested/validated using a diverse set of slides prepared from mock sexual assault samples containing a limited or a large number of spermatozoa (fecal swabs, vaginal swabs, all mixed with different semen dilutions in addition to urine, blood and yeasts for a subset of those swabs). The performance of Metafer was recorded with respect to false positive counts, false negative counts and time required for the detection of spermatozoa in each sample. Automated spermatozoa counts were further compared to manual spermatozoa scoring in addition to comparing the time spent executing the identification. An excellent concordance was noted between automated and manual counts. The results of this study indicate that automated scoring of fluorescently stained spermatozoa in mock sexual assault exhibits can be carried out reliably and reproducibly using well-developed classifiers for the MetaSystems Metafer image analysis software system. The automated scoring of spermatozoa combining Sperm Hy-Liter™/MetaSystems Metafer will be tested on a large number of sexual assault cases as part of a pilot project within an operational setting.