Abstract
OBJECTIVES: Expanding the virtual microscopy (VM) slide collection for nongynecological cytology is important to teaching. In a mixed-methods approach, this study evaluated VM's performance and user experience to determine its feasibility and usage in an educational setting.
METHODS: From September through October 2022, the students reviewed 44 whole-slide imaged nongynecological slides. The concordance rate with reference diagnosis was compared with that from light microscopy (LM) from 4 months earlier. In addition to assessing the overall performance from VM, imaged urinary cytology's accuracy was reviewed for both urinary and nonurinary cytology. Finally, the students' weekly feedback logs were analyzed to gain insights for improving the digital screening experience.
RESULTS: The overall nongynecological diagnostic accuracy was significant between the 2 screening platforms (P < .001), favoring LM over the VM platform. Light microscopy also performed better than VM in urine cytology cases, with 84.2% concordance against reference diagnosis, compared with 61.1% for the VM platform (P = .03). As for the accuracy of nonurinary cases, its glass slide (LM) agreement with the reference diagnosis was also superior at 84.8%, compared with 58.8% for VM (P = .03). Finally, the overarching theme discerned from reviewing the user logs was concern over image quality, which was mentioned 76 times.
CONCLUSIONS: The VM results were poorer compared with LM in our validation. Its use seems promising, but more focus is needed to improve the VM screening platform.