AI-assisted imaging found hidden basal cell carcinomas on normal-looking faces

Design

Prospective cross-sectional feasibility study, single centre

Population

150 inpatients with at least two basal cell carcinoma risk factors

Primary outcome

Positive predictive value for histologically confirmed subclinical BCC

Effect

PPV 83.3% (95% CI 58.6% to 96.4%); 9.3% of patients had subclinical BCC

A German feasibility study scanned clinically unremarkable facial skin in 150 high-risk inpatients using line-field confocal optical coherence tomography with AI-assisted image reading.

It found 17 subclinical basal cell carcinomas in 14 patients (9.3%). Of 18 lesions called basal cell carcinoma, 15 were confirmed on histology, a positive predictive value of 83.3% (95% CI 58.6% to 96.4%). Most were superficial.

This shows the technique can find early tumours, but not whether finding them earlier improves outcomes or justifies the cost. The authors are clear it is not ready for routine screening.

• Examine the whole face carefully in patients with multiple risk factors for basal cell carcinoma
• Use dermoscopy on subtle pink or shiny facial patches
• Do not refer for imaging-based screening outside research settings yet
• Reinforce sun protection for high-risk patients at every visit

Why it matters

High-risk patients may carry more early basal cell carcinomas than a visual examination reveals.

Don't overread it

A feasibility study in 150 inpatients; sensitivity, cost-effectiveness and clinical benefit are unknown.

The statistics, in plain English

A positive predictive value of 83% means about five in six imaging-diagnosed lesions were truly cancer. The study could not measure how many cancers were missed.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.