OBJECTIVE
To determine whether digital technologies can be used for cytological diagnosis of uterine cervical diseases to further apply in cervical cancer screening programs.
SUBJECTS AND METHODS
A cytological study was conducted to examine the cervical and cervical canal (CC) scrapes from 1775 patients aged 20 to 60 years who had gone to a gynecologist during their preventive examination in 2019—2020. The materials were examined using conventional cytology (CC), liquid-based cytology (LBC), digital cytology (DC) using digitized liquid smears, and automated analysis of scanned liquid specimens (DC with artificial intelligence (AI)). The investigators assessed the following indicators: adequacy of material; visibility of cytology specimens; morphological identity; detectability, pattern of pathology; comparability of results; study time; and labor intensity.
RESULTS
The adequacy of the material examined by CC and LBC was the same and amounted to 87%. The visibility of cytology specimens by LBC was 2% worse than that by CC; the visibility by DC was 4% better than that by microscopy. The detectability of pathology by CC and LBC was identical (10 and 11%, respectively); the detectability by DC with AI was 18%. The pattern of the detected pathological changes by DC slightly differed from that by CC, but corresponded to that by LBC. Comparison of the results of these techniques showed that in 25% of cases, DC with AI revealed pathology where CC and LBC were absent. The time spent on a study by DC with AI was 2—6 times less than that by CC and LBC and averaged 2 to 8 min. CC was least labor-intensive and cheap; LBC and DC with AI were most objective, functional, standardized, and promising.
CONCLUSION
The findings will assist in identifying the optimal method for cervical cytological diagnosis and lay a foundation for cervical cancer telescreening.