AI-Assisted Line-Field Confocal Optical Coherence Tomography to Detect Subclinical Basal Cell Carcinoma
Moritz Ronicke, Laura Dürr, Michael W. Höner, Léonie Staats, Michael Erdmann, Carola BerkingImportance
Basal cell carcinoma (BCC) is the most common type of skin cancer worldwide. Particularly on the face, its locally destructive growth leads to functional and aesthetic impairment. Early diagnosis enables less invasive treatment. Line-field confocal optical coherence tomography (LC-OCT) is an effective noninvasive imaging method for the diagnosis of BCC.
Objectives
To investigate whether LC-OCT paired with artificial intelligence (AI)–assisted image recognition can identify subclinical BCCs through systematic screening of inconspicuous facial skin in patients at high risk of developing BCC.
Design, Setting, and Participants
This prospective cross-sectional feasibility study was conducted from April 2025 to October 2025 at the Department of Dermatology, Uniklinikum Erlangen, in Erlangen, Germany. Consecutive patients receiving inpatient care who had at least 2 established BCC risk factors were enrolled in the study.
Exposure
After the exclusion of lesions macroscopically suggestive of BCC, inconspicuous facial skin was systematically examined using AI-assisted LC-OCT.
Main Outcome and Measure
The prespecified main outcome was positive predictive value (PPV) of AI-assisted LC-OCT for histologically confirmed subclinical BCC.
Results
Among 150 patients (mean [SD] age, 72.9 [10.3] years; 57 [38.0%] female; 93 [62.0%] male), 17 subclinical BCCs were identified in 14 patients (9.3%). Of 18 lesions diagnosed as BCC with AI-assisted LC-OCT, 15 were histologically confirmed, corresponding to a PPV of 83.3% (95% CI, 58.1%-96.4%). One lesion had a false-positive result (actinic keratosis); 2 patients declined biopsy. In a sensitivity analysis including the 2 nonbiopsied lesions that were confirmed with expert analysis, PPV was 94.4% (95% CI, 72.7%-99.9%). The distribution of BCC subtypes included 13 cases of superficial BCCs (76.5%), followed by 3 cases of nodular BCCs (17.6%), and 1 case of infiltrative BCC (5.9%). Subtype classification was accurate in 13 of 15 histologically confirmed cases (86.7% [95% CI, 59.5%-98.3%]).
Conclusions and Relevance
In this cross-sectional feasibility study, the findings suggest that systematic screening with AI-assisted LC-OCT may be a feasible approach for BCC detection at a preclinical stage in populations at high risk for BCC. Further prospective studies are needed to assess the sensitivity and cost-effectiveness of this method, and whether early detection translates to tangible clinical benefit before implementation in routine care can be recommended.