[BGCS 2026] Clinical Evaluation of an AI-powered Digital Cytology Platform for Automated Cervical Cancer Screening
miLab™ CER-assisted review improved sensitivity for detecting cervical abnormalities versus conventional microscopy while maintaining comparable specificity.
BGCS 2026
|June 25, 2026
|Mijin Kim, et al.
![[BGCS 2026] Clinical Evaluation of an AI-powered Digital Cytology Platform for Automated Cervical Cancer Screening](/_next/image?url=%2Fimages%2Fthumbnails%2Fclinical-evaluation-of-an-ai-powered-digital-cytology-platform-for-automated-cervical-cancer-screening-poster.png&w=3840&q=75&dpl=dpl_5NYdPN23Yv9z4xTNbbArpDeYoZfa)
Unauthorized downloading or distribution of this poster is strictly prohibited.
Abstract
This study evaluated the diagnostic performance of miLab™ CER, an all-in-one platform integrating automated staining, imaging, and AI-assisted analysis, for the detection of cervical abnormalities based on the Bethesda System (TBS). A total of 150 liquid-based cytology (LBC) slides were independently reviewed by three external cytologists using both miLab™ CER-assisted review and conventional light microscopy, with a 2-week washout period between assessments. Ground truth diagnoses were established by consensus among three expert cytologists. The results demonstrate the potential of miLab™ CER to standardize cervical cancer screening, reduce manual screening workload, and improve the detection of cervical abnormalities—particularly high-grade lesions—while maintaining specificity comparable to conventional microscopy.
| miLab™ CER | Microscopy | Difference | |
|---|---|---|---|
| Low-Risk+ (ASCUS+) | |||
| Sensitivity | 92.44% | 83.11% | +9.33 pp |
| Specificity | 92.00% | 94.67% | -2.67 pp |
| High-Risk+ (ASC-H+) | |||
| Sensitivity | 82.46% | 52.63% | +29.83 pp |
| Specificity | 98.47% | 99.24% | -0.77 pp |
Table 1: Clinical Performance Comparison: miLab™ CER vs. Microscopy