Dari Mikroskop ke Machine Learning: Tinjauan Literatur tentang Teknologi Pemeriksaan Spermatozoa pada Kasus Persetubuhan (2019–2024)

Abstract

Background:   Forensic examination of semen in cases of sexual intercourse (e.g., sexual assault) aims to detect the presence of spermatozoa or male-specific biomarkers in vaginal swab samples from victims. Conventional methods, including microscopic examination with staining and chemical tests (acid phosphatase and prostate-specific antigen/PSA), are routinely used but have limitations in sensitivity and specificity. In the past five years, advancements have emerged in semen detection technology, including Loop-mediated Isothermal Amplification (LAMP), Y-STR (Y chromosome short tandem repeat) analysis, sperm-specific aptamers, and Convolutional Neural Networks (CNN) for automated identification. Objective: This review article aims to summarize current literature (2019–2024) regarding spermatozoa detection methods in vaginal swab samples, covering both conventional and innovative techniques. Methods: Literature was searched through international and national databases using relevant keywords including “forensic sperm detection,” “vaginal swab semen,” “LAMP for sperm,” “Y-STR sexual assault,” “aptamer sperm detection,” and “CNN sperm microscope” within the publication range of 2019–2024. Results: Recent studies confirm that microscopy (histological staining) and chemical testing (acid phosphatase and PSA) remain the initial standard for semen examination. However, both are presumptive and susceptible to false-positive or false-negative results. LAMP and Y-screen methods demonstrated high sensitivity in detecting male-specific DNA, comparable to Y-STR profiling. Fluorescent DNA aptamers selectively bind to sperm cells, enhancing detection speed and accuracy. The application of CNN to colored microscopic images has enabled automated “sperm hunting” with accuracy exceeding 90%, significantly expediting laboratory analysis. Conclusion:  Conventional methods such as microscopy and chemical assays remain essential in forensic semen analysis, but show limited reliability in low-sperm samples or cases involving azoospermic perpetrators. Emerging techniques based on DNA, biochemical markers, and artificial intelligence offer improvements in sensitivity, specificity, and detection efficiency. Integration of multi-method approaches is recommended to optimize outcomes. Further standardization and research are needed to facilitate widespread implementation of these novel technologies in forensic laboratories.