A-068 Validation of an Automated Digital and AI-Assisted Workflow for Ova and Parasite Examination

Abstract

Traditional ova and parasite (O&P) examination by manual microscopy is labor-intensive, subjective and highly dependent on technologist expertise. Much of this expertise resides in an aging workforce, making it difficult to standardize and pass on to newer technologists. Automation of slide preparation combined with digital pathology and artificial intelligence (AI) has the potential to improve efficiency, standardization, and consistency while preserving diagnostic accuracy. We validated an automated workflow that encompasses the complete O&P examination, including both direct wet mount and permanently stained components, as an integrated diagnostic process. Direct wet mounts and trichrome-stained slides were prepared from stool specimens preserved in Para-Pak® Ultra Ecofix™. The automated workflow for the slides was the following: the slides were processed using the Leica XL ST5010 Automated Slide Stainer, scanned with the Pramana Spectra HD-2 scanner, and analyzed using Techcyte AI software. The Techcyte AI system classifies objects and presents them for technologist review. Slides classified as positive or flagged for review were confirmed by manual microscopy. Slides classified as negative were released without further review. Accuracy was assessed using 45 unique positive specimens (tested in triplicate) and 48 negative specimens, including reference material from external laboratories. Precision, reportable range, analytical sensitivity (limit of detection comparison), specificity, pooled specimen performance, and quality control procedures were evaluated. Overall concordance of the automated workflow with manual microscopy was 91%. Specifically, we had a sensitivity of 97% and specificity of 75%. No major discrepancies (missed organism classes) were identified. Within-run precision demonstrated 87% agreement, consistent with known stool heterogeneity, while between-run precision showed 100% agreement across 54 slides. Limit-of-detection comparison studies demonstrated 97% overall concordance, with Techcyte detecting organisms at dilution factors equal to or greater than manual microscopy. Analytical specificity was 100% in specimens containing artifacts. This validation supports implementation of an automated workflow of the complete O&P examination combining automated staining, digital scanning, and AI-assisted classification. The approach maintains high sensitivity while also reducing the manual review burden of negative cases while preserving confirmatory microscopy for positive cases. This system offers a scalable, reproducible alternative to traditional O&P examination in clinical laboratories.