This prospective cross-sectional feasibility study evaluated whether AI-assisted line-field confocal optical coherence tomography (LC-OCT) could detect subclinical basal cell carcinoma (BCC) by systematically screening inconspicuous facial skin in 150 high-risk inpatients (≥2 BCC risk factors) at a single German center from April–October 2025.
AI-assisted LC-OCT identified subclinical BCC in 14 of 150 patients (9.3%), with a PPV of 83.3% (95% CI, 58.1%–96.4%) for histologically confirmed BCC; sensitivity analysis including 2 non-biopsied expert-confirmed lesions raised PPV to 94.4% (95% CI, 72.7%–99.9%). Subtype classification was accurate in 13 of 15 confirmed cases (86.7%).
- Sensitivity cannot be calculated — there is no reference standard for truly missed subclinical BCCs, so false-negative rate is unknown. - Small single-center sample (n=150) limits generalizability. - 2 patients declined biopsy, introducing uncertainty in the primary PPV estimate.
AI-assisted LC-OCT shows promise for detecting subclinical BCC on the face in high-risk patients with a PPV exceeding 80%, but sensitivity and cost-effectiveness data are lacking — routine implementation is premature until larger prospective studies confirm clinical benefit.
Explore related topics
