Résumé : Actinic keratoses (AKs) commonly arise on sun-damaged skin. Visible lesions are often associated with subclinical lesions on surrounding skin, giving rise to field cancerization. To avoid multiple biopsies to diagnose subclinical/early invasive lesions, there is an increasing interest in non-invasive diagnostic tools, such as high-definition optical coherence tomography (HD-OCT). We previously developed a HD-OCT-based diagnostic algorithm for the discrimination of AK from squamous cell carcinoma (SCC) and normal skin. The aim of this study was to test the applicability of HD-OCT for non-invasive discrimination of AK from SCC and normal skin using this algorithm. Three-dimensional (3D) HD-OCT images of histopathologically proven AKs and SCCs and images of normal skin were collected. All images were shown in a random sequence to three independent observers with different experience in HD-OCT, blinded to the clinical and histopathological data and with different experience with HD-OCT. Observers classified each image as AK, SCC or normal skin based on the diagnostic algorithm. A total of 106 (38 AKs, 16 SCCs and 52 normal skin sites) HD-OCT images from 71 patients were included. Sensitivity and specificity for the most experienced observer were 81.6% and 92.6% for AK diagnosis and 93.8% and 98.9% for SCC diagnosis. A moderate interobserver agreement was demonstrated. HD-OCT represents a promising technology for the non-invasive diagnosis of AKs. Thanks to its high potential in discriminating SCC from AK, HD-OCT could be used as a relevant tool for second-level examination, increasing diagnostic confidence and sparing patients unnecessary excisions.