EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Perifollicular Crescentic Pigmentation Improves Recognition of Flat Pigmented Facial Basal Cell Carcinoma: A Dermatoscopic Training and Artificial Intelligence Study

Ahmet Taha Aydemir, Handan Merve Erol Mart, Philipp Tschandl, Aylin Okçu Heper, Bengu Nisa Akay

Journal of the American Academy of Dermatology · 2026-11

DOI: 10.1016/j.jaad.2026.11.032 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of the American Academy of Dermatology, 2026-11. Verfasst von Ahmet Taha Aydemir und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Dermatology, Yozgat City Hospital, Yozgat, Türkiye. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: ERC.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

BACKGROUND: Flat pigmented basal cell carcinoma (BCC) of the face is an underrecognized diagnostic pitfall that may mimic other facial pigmented lesions dermatoscopically. OBJECTIVE: To characterize perifollicular crescentic pigmentation (PCP) as a novel dermatoscopic criterion of flat facial pigmented BCC, assess its educational impact, and compare its diagnostic utility with contemporary artificial intelligence (AI) models. METHODS: A retrospective dataset of 120 flat facial pigmented lesions (60 BCCs, 20 lentigo maligna, 20 pigmented actinic keratoses, and 20 solar lentigines/seborrheic keratoses) was assembled. Twenty-six dermatology residents completed a diagnostic test before and one week after a structured educational session on PCP recognition. ADAE, ResNet34, PanDerm, and MedSigLIP were evaluated on the same dataset. RESULTS: PCP was identified in 63.3% of flat facial BCCs. Interobserver agreement was substantial (κ=0.780; 95% CI: 0.706-0.857). Following education, resident sensitivity increased from 73.7% to 88.0% (P < .001), and overall accuracy from 85.8% to 90.4% (P = .001). AI sensitivities ranged from 26.7% to 70.0%. PanDerm sensitivity increased to 96.7% after linear probing on the study dataset. LIMITATIONS: Retrospective single-center design and potential recall bias. CONCLUSION: PCP is a reproducible and readily teachable dermatoscopic criterion that significantly improves the recognition of flat pigmented facial BCC.

Schlagwörter

Stichwörter der Autor:innen

artificial intelligencebasal cell carcinomadermatoscopydermatoscopy trainingfacial pigmented lesionslentigo malignaperifollicular crescentic pigmentation

Autor:innen und Institutionen

  • Department of Dermatology, Yozgat City Hospital, Yozgat, Türkiye.
  • Department of Dermatology, Faculty of Medicine, Ankara University, Ankara, Türkiye.
  • Department of Dermatology, Medical University of Vienna, Vienna, Austria.
  • Department of Pathology, Faculty of Medicine, Ankara University, Ankara, Türkiye.
  • Department of Dermatology, Faculty of Medicine, Ankara University, Ankara, Türkiye. Electronic address: nisaakay@gmail.com.

Förderung

  • ERC

Zitieren

APA

Aydemir, A. T., Erol Mart, H. M., Tschandl, P., Heper, A. O., & Akay, B. N. (2026). Perifollicular Crescentic Pigmentation Improves Recognition of Flat Pigmented Facial Basal Cell Carcinoma: A Dermatoscopic Training and Artificial Intelligence Study. Journal of the American Academy of Dermatology. https://doi.org/10.1016/j.jaad.2026.11.032

Vancouver

Aydemir AT, Erol Mart HM, Tschandl P, Heper AO, Akay BN. Perifollicular Crescentic Pigmentation Improves Recognition of Flat Pigmented Facial Basal Cell Carcinoma: A Dermatoscopic Training and Artificial Intelligence Study. Journal of the American Academy of Dermatology. 2026. doi:10.1016/j.jaad.2026.11.032

BibTeX
@article{Aydemir2026,
  author = {Aydemir, Ahmet Taha and Erol Mart, Handan Merve and Tschandl, Philipp and Heper, Aylin Okçu and Akay, Bengu Nisa},
  title = {Perifollicular Crescentic Pigmentation Improves Recognition of Flat Pigmented Facial Basal Cell Carcinoma: A Dermatoscopic Training and Artificial Intelligence Study},
  journal = {Journal of the American Academy of Dermatology},
  year = {2026},
  doi = {10.1016/j.jaad.2026.11.032}
}
Kostenlos starten →