Introduction: Residual protein contamination on reusable surgical instruments remains a significant patient safety risk in central sterile supply departments. Inadequate cleaning of it undermines sterilization itself: residual organic material can shield microorganisms from the sterilizing agent, becoming a document contributing cause of healthcare-associated infections (HAIs). The WHO Global Action Plan on infection prevention and control (2024-2030) targets HAI-free care by 2030, calling for new technologies to strengthen and objectify prevention practices. Yet cleaning verification, the checkpoint before sterilization, still relies on subjective visual inspection, unvalidated on metallic surfaces. This study addresses that gap with an automated OPA fluorescence detection system integrated with semantic segmentation.
Methods: The station integrates a UV-A excitation source, a motorized rotating platform, an atomizer and a peristaltic pump for reagent application, and image acquisition via raspberry pi HQ camera. An esp32 microcontroller coordinates home calibration, synchronized reagent dispensing a controlled 120s OPA reaction interval, UV-A activation and triggered image capture. Captured are processed using u-nets architecture with MobileNetV2 encoder pre-trained via transfer learning, balancing accuracy and efficiency for embedded deployment.
Results and discussion: Automated control system was implemented and verified with rotation, reagent dispensing, timing and image capture subsystems operating in the designed sequence, eliminating operator dependent variability inherent to manual inspection. An initial image dataset of surgical blade instruments has been collected to support segmentation model development with the U-net architecture selected to enable automated identification of contamination regions beyond manual assessment. The system's biochemical detection stage is currently advancing through validation testing.
Conclusions: This work demonstrates the feasibility of an automated, low-cost optical station combining OPA fluorescence with U-net segmentation, advancing pre-sterilization cleaning verification towards quantitative, AI assisted standards.