Journals / Batı Karadeniz Tıp Dergisi / 2026 / Cilt: 10 Sayı: 2

Artificial intelligence-assisted analysis of clinical and dermoscopic skin lesion images: A comparative study with dermatologist evaluation

Klinik ve dermoskopik deri lezyon görüntülerinin yapay zekâ destekli analizi: Dermatolog değerlendirmesi ile karşılaştırmalı bir çalışmaKlinik ve dermoskopik deri lezyon görüntülerinin yapay zekâ destekli analizi: Dermatolog değerlendirmesi ile karşılaştırmalı bir çalışma

Seçil Soylu(Afyonkarahisar Health Sciences University, Faculty of Medicine, Department of Dermatology and Venereology, Afyonkarahisar, Türkiye)0000-0001-5293-5201
Rimsha Shadid(University of Central Punjab, Department of Biotechnology, Lahore, Pakistan)0009-0005-2884-8788
Pages
347–363
Publication Date
2026-08-31
🏛️Bibliographic metadata and full text access for this article are provided via DergiPark.
DergiPark Record ↗
✨

AI Summary & Key Takeaways

AI Insight
⚡ TL;DR(Too Long; Didn't Read)

Results: The AI ensemble achieved an overall seven-class accuracy of 94.2%, with an overall sensitivity of 93.8% (95% CI: 91.4-95.7) and specificity of 94.5% (95% CI: 92.3-96.2).

📌 3 Key Takeaways

🎯
Objective & Hypothesis:Aim: Although dermoscopy has improved diagnostic accuracy for melanocytic lesions and melanoma in particular, inter-observer variability remains a clinical concern.
📊
Key Findings & Data:Material and Methods: A dataset comprising more than 25,000 clinical and dermoscopic images obtained from the ISIC Archive and HAM10000 databases was used.
💡
Conclusion & Impact:Results: The AI ensemble achieved an overall seven-class accuracy of 94.2%, with an overall sensitivity of 93.8% (95% CI: 91.4-95.7) and specificity of 94.5% (95% CI: 92.3-96.2).
🔬Design:Modeling & Simulation
📍Scope:Comprehensive Sample
⚙️Variables:Yapay zeka · dermoskopi · deri tümörleri

Abstract

Aim: Although dermoscopy has improved diagnostic accuracy for melanocytic lesions and melanoma in particular, inter-observer variability remains a clinical concern. This study evaluates the potential of an artificial intelligence (AI) ensemble to support dermatologists in a seven-class skin lesion classification task (including melanoma, basal cell carcinoma, actinic keratosis/intraepithelial carcinoma (AKIEC) and several benign lesions), using clinical and dermoscopic images. The comparison is intentionally restricted to an image-only, standardized classification setting and is not intended to represent full dermatologic diagnostic practice. Material and Methods: A dataset comprising more than 25,000 clinical and dermoscopic images obtained from the ISIC Archive and HAM10000 databases was used. A four-model ensemble deep learning architecture (ResNet-50, VGG-16, InceptionV3, and EfficientNet-B4) was developed using weighted voting aggregation. Dermatologists evaluated only the images and did not have access to patient age, sex, Fitzpatrick skin type, lesion site, duration, evolution, symptoms, prior treatments or palpation findings; the reader arm therefore reflects standardized image classification, not routine clinical practice. Model performance was compared with the dermatology panel on the same held-out image set using standard diagnostic metrics (sensitivity, specificity, F1, AUC-ROC). Grad-CAM was used to visualize model decision regions. Results: The AI ensemble achieved an overall seven-class accuracy of 94.2%, with an overall sensitivity of 93.8% (95% CI: 91.4-95.7) and specificity of 94.5% (95% CI: 92.3-96.2). The melanoma-specific sensitivity was 94.1%. On the same image-only test set, twelve board-certified dermatologists reached a mean accuracy of 88.5% (95% CI: 86.1-90.9). Ensemble modeling outperformed individual networks, and Grad-CAM outputs consistently mapped onto clinically relevant lesion regions. Conclusion: These findings should not be interpreted as AI being clinically superior to dermatologists, but as a comparative evaluation of standardized image-classification performance under an image-only setting. AI-assisted systems may serve as supportive tools in dermatology by assisting early detection, optimizing triage in high-volume settings and aiding teledermatology decision-making, complementing — not replacing — clinical expertise.

Özet

Amaç: Dermoskopi melanositik lezyonların ve melanomun tanısında tanı doğruluğunu artırmış olsa da gözlemciler arası tutarlılık hâlâ bir sorun olarak kalmaktadır. Bu çalışma, yalnızca görüntü tabanlı bir standart sınıflandırma senaryosunda, yapay zekâ (YZ) modelinin klinik ve dermoskopik görüntüler kullanarak yedi sınıflı deri lezyonu sınıflandırmasında (melanom, bazal hücreli karsinom, aktinik keratoz/intraepitelyal karsinom (AKIEC) ve çeşitli benign lezyonlar) dermatologları destekleme potansiyelini değerlendirmektedir. Amaç, yapay zekânın dermatologların yerini alması değil, standart görüntü sınıflandırma performansı açısından karar destek aracı olarak katkısının araştırılmasıdır. Gereç ve Yöntemler: Çalışmada, ISIC Arşivi ve HAM10000 veri setlerinden elde edilen 25.000’den fazla klinik ve dermoskopik görüntü kullanılmıştır. Dört farklı konvolüsyonel sinir ağı mimarisinden (ResNet-50, VGG-16, InceptionV3 ve EfficientNet-B4) oluşan ensemble derin öğrenme modeli, ağırlıklı oylama yöntemi ile entegre edilmiştir. Dermatologlar yalnızca görüntü üzerinden değerlendirme yapmış; hasta yaşı, cinsiyeti, Fitzpatrick deri tipi, lezyon lokalizasyonu, süresi, semptomlar ve palpasyon bulguları gibi klinik bilgiler kendilerine sunulmamıştır. Bu nedenle karşılaştırma, gerçek klinik tanı sürecini değil, standart görüntü sınıflandırma performansını yansıtmaktadır. Model performansı, aynı görüntü kümesi üzerinde dermatoloji uzmanlarının değerlendirmeleri ile karşılaştırılmış; duyarlılık ve özgüllük başta olmak üzere temel tanısal performans ölçütleri analiz edilmiştir. Bulgular: Yapay zekâ ensemble modelinin yedi sınıflı genel doğruluğu %94,2, genel duyarlılığı %93,8 (%95 GA: 91,4-95,7) ve genel özgüllüğü %94,5 (%95 GA: 92,3-96,2) olarak bulunmuştur. Melanom sınıfına özgü duyarlılık ise %94,1 olarak saptanmıştır. Sadece görüntü tabanlı bu değerlendirmede, on iki dermatoloğun ortalama doğruluğu %88,5 (%95 GA: 86,1-90,9) olarak saptanmıştır. Ensemble yaklaşımının, tekil modellerin performansına kıyasla daha yüksek doğruluk sağladığı görülmüştür. Grad-CAM görselleştirme teknikleri ile modelin karar odaklarının dermatolojik olarak yorumlanabilir olduğu gösterilmiştir. Sonuç: Bu çalışmanın sonuçları, yapay zekânın dermatologlardan üstün olduğu şeklinde değil, klinik bilgi ve fizik muayeneden bağımsız, standardize görüntü sınıflandırma senaryosundaki karşılaştırmalı performansı olarak yorumlanmalıdır. Yapay zekâ destekli sistemler; erken tanının desteklenmesi, yüksek hasta yükü bulunan merkezlerde triyajın optimize edilmesi ve teledermatoloji uygulamalarında karar desteği sağlanması açısından değerli olabilir. Klinik uzmanlığın yerini almak yerine tamamlayıcı bir araç olarak konumlandırılmalıdır.

🔖

Cite This Article

Soylu, S., Shadid, R. (2026). Artificial intelligence-assisted analysis of clinical and dermoscopic skin lesion images: A comparative study with dermatologist evaluation. Batı Karadeniz Tıp Dergisi, Cilt: 10 Sayı: 2, 347–363. https://doi.org/10.29058/mjwbs.1890283

📄 Full Text PDF Viewer

📄

PDF görüntüleyici hazırlanıyor...

References (38)

  1. Dinnes J, Deeks JJ, Chuchu N, Ferrante di Ruffano L, Matin RN, Thomson DR, et al. Dermoscopy, with and without visual inspection, for diagnosing melanoma in adults. Cochrane Database Syst Rev. 2018;12(12):CD011902. https://doi.org/10.1002/14651858.CD011902
  2. Brinker TJ, Hekler A, Enk AH, Berking C, Haferkamp S, Hauschild A, et al. Deep learning outperformed 136 of 157 dermatologists in a head-to-head dermoscopic melanoma image classification task. Eur J Cancer. 2019;113:47-54. https://doi.org/10.1016/j.ejca.2019.04.001
  3. Esteva A, Kuprel B, Novoa RA, Ko J, Swetter SM, Blau HM, et al. Dermatologist-level classification of skin cancer with deep neural networks. Nature. 2017;542(7639):115-118. https://doi.org/10.1038/nature21056
  4. Haenssle HA, Fink C, Schneiderbauer R, Toberer F, Buhl T, Blum A, et al. Man against machine: diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists. Ann Oncol. 2018;29(8):1836-1842. https://doi.org/10.1093/annonc/mdy166
  5. Chen JY, Fernandez K, Fadadu RP, Reddy R, Wei ML. Skin cancer diagnosis by lesion, physician, and examination type: a systematic review and meta-analysis. JAMA Dermatol. 2025;161(2):135-146. https://doi.org/10.1001/jamadermatol.2024.4382
  6. Tschandl P, Rosendahl C, Kittler H. The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions. Sci Data. 2018;5:180161. https://doi.org/10.1038/sdata.2018.161
  7. Rotemberg V, Kurtansky N, Betz-Stablein B, Caffery L, Chousakos E, Codella N, et al. A patient-centric dataset of images and metadata for identifying melanomas using clinical context. Sci Data. 2021;8:34. https://doi.org/10.1038/s41597-021-00815-z
  8. Barata C, Celebi ME, Marques JS. A survey of feature extraction in dermoscopy image analysis of skin cancer. IEEE J Biomed Health Inform. 2019;23(3):1096-1109. https://doi.org/10.1109/JBHI.2018.2845939
  9. Braun RP, Rabinovitz HS, Oliviero M, Kopf AW, Saurat JH. Pattern analysis: a two-step procedure for the dermoscopic diagnosis of melanoma. Clin Dermatol. 2002;20(3):236-239. https://doi.org/10.1016/S0738-081X(02)00216-X
  10. Zalaudek I, Argenziano G, Soyer HP, Corona R, Sera F, Blum A, et al. Three-point checklist of dermoscopy: an open internet study. Br J Dermatol. 2006;154(3):431-437. https://doi.org/10.1111/j.1365-2133.2005.06983.x
  11. Argenziano G, Soyer HP, Chimenti S, Talamini R, Corona R, Sera F, et al. Dermoscopy of pigmented skin lesions: results of a consensus meeting via the Internet. J Am Acad Dermatol. 2003;48(5):679-693. https://doi.org/10.1067/mjd.2003.281
  12. Marghoob AA, Swindle LD, Moricz CZM, Sanchez Negron FA, Slue B, Halpern AC. Instruments and new technologies for the in vivo diagnosis of melanoma. J Am Acad Dermatol. 2003;49(5):777-797. https://doi.org/10.1016/S0190-9622(03)02470-8
  13. Celebi ME, Kingravi HA, Uddin B, Iyatomi H, Aslandogan YA, Stoecker WV, et al. A methodological approach to the classification of dermoscopy images. Comput Med Imaging Graph. 2007;31(6):362-373. https://doi.org/10.1016/j.compmedimag.2007.01.003
  14. Codella NCF, Nguyen QB, Pankanti S, Gutman D, Helba B, Halpern A, et al. Deep learning ensembles for melanoma recognition in dermoscopy images. IBM J Res Dev. 2017;61(4/5):1-28. https://doi.org/10.1147/JRD.2017.2708299
  15. Fujisawa Y, Otomo Y, Ogata Y, Nakamura Y, Fujita R, Ishitsuka Y, et al. Deep-learning-based, computer-aided classifier developed with a small dataset of clinical images surpasses board-certified dermatologists in skin tumour diagnosis. Br J Dermatol. 2019;180(2):373-381. https://doi.org/10.1111/bjd.16924
  16. Han SS, Park GH, Lim W, Kim MS, Na JI, Park I, et al. Classification of the clinical images for benign and malignant cutaneous tumors using a deep learning algorithm. J Invest Dermatol. 2018;138(7):1529-1538. https://doi.org/10.1016/j.jid.2018.01.028
  17. Binder A, Bach S, Montavon G, Müller KR, Samek W. Analyzing and validating neural networks predictions for medical imaging. In: IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2017;1204-1212.
  18. Tan M, Le QV. EfficientNet: Rethinking model scaling for convolutional neural networks. In: Proc Int Conf Mach Learn; 2019;6105-6114.
  19. Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas. 1960;20(1):37-46. https://doi.org/10.1177/001316446002000104
  20. McNemar Q. Note on the sampling error of the difference between correlated proportions or percentages. Psychometrika. 1947;12(2):153-157. https://doi.org/10.1007/BF02295996
  21. Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. Paper presented at: 3rd International Conference on Learning Representations, ICLR, San Diego, 2015.
  22. Salinas MP, Sepúlveda J, Hidalgo L, Peirano D, Morel M, Uribe P, et al. A systematic review and meta-analysis of artificial intelligence versus clinicians for skin cancer diagnosis. npj Digit Med. 2024;7:125. https://doi.org/10.1038/s41746-024-01103-x
  23. Laiouar-Pedari S, Kühn A, Wies C, Nogueira Garcia C, Winterstein JT, Heinlein L, et al. Prospective evidence on artificial intelligence-assisted melanoma diagnostics: a systematic review and meta-analysis. JAMA Dermatol. 2026;162(5):478-487. https://doi.org/10.1001/jamadermatol.2026.0217
  24. Ricci Lara MA, Frutos EL, Rodríguez Kowalczuk MV, Ferraresso MG, Benítez SE, Mazzuoccolo LD. External validation of an artificial intelligence ensemble for skin cancer detection: enhancing diagnostic performance on dermoscopic images. J Invest Dermatol. 2025;145(11):2885-2888.e2. https://doi.org/10.1016/j.jid.2025.04.021
  25. Thwin SM, Park HS. Skin lesion classification using a deep ensemble model. Appl Sci. 2024;14(13):5599. https://doi.org/10.3390/app14135599
  26. Pacal I, Alaftekin M, Zengul FD. Enhancing skin cancer diagnosis using swin transformer with hybrid shifted window-based multi-head self-attention and SwiGLU-based MLP. J Imaging Inform Med. 2024;37(6):3174-3192. https://doi.org/10.1007/s10278-024-01140-8
  27. Selvaraju RR, Cogswell M, Das A, Vedantam R, Parikh D, Batra D. Grad-CAM: Visual explanations from deep networks via gradient-based localization. Int J Comput Vis. 2020;128:336-359. https://doi.org/10.1007/s11263-019-01228-7
  28. Kawahara J, Daneshvar S, Argenziano G, Hamarneh G. Seven-point checklist and skin lesion classification using multitask multimodal neural nets. IEEE J Biomed Health Inform. 2019;23(2):538-546. https://doi.org/10.1109/JBHI.2018.2824327
  29. Yu L, Chen H, Dou Q, Qin J, Heng PA. Automated melanoma recognition in dermoscopy images via very deep residual networks. IEEE Trans Med Imaging. 2017;36(4):994-1004. https://doi.org/10.1109/TMI.2016.2642839
  30. Codella N, Cai J, Abedini M, Garnavi R, Halpern A, Smith JR. Deep Learning, Sparse Coding, and SVM for Melanoma Recognition in Dermoscopy Images. In: Zhou L, Wang L, Wang Q, Shi Y (eds). Machine Learning in Medical Imaging. MLMI 2015. Lecture Notes in Computer Science 2015; volume 9352. Springer, Cham. https://doi.org/10.1007/978-3-319-24888-2_15
  31. Marchetti MA, Codella NCF, Dusza SW, Gutman D, Helba B, Kalloo A, et al. Results of the 2016 ISIC challenge: comparison of computer algorithms to dermatologists. J Am Acad Dermatol. 2018;78(2):270-277. https://doi.org/10.1016/j.jaad.2017.08.016
  32. Kittler H, Marghoob AA, Argenziano G, Carrera C, Curiel-Lewandrowski C, Hofmann-Wellenhof R, et al. Standardization of terminology in dermoscopy. J Am Acad Dermatol. 2016;74(6):1093-1106. https://doi.org/10.1016/j.jaad.2015.12.038
  33. Blum A, Rassner G, Garbe C. Digital image analysis for diagnosis of cutaneous melanoma. Br J Dermatol. 2004;151(5):1029-1038. https://doi.org/10.1111/j.1365-2133.2004.06210.x
  34. Carrera C, Marchetti MA, Dusza SW, Argenziano G, Braun RP, Halpern AC, et al. Validity and reliability of dermoscopic criteria. JAMA Dermatol. 2016;152(7):798-806. https://doi.org/10.1001/jamadermatol.2016.0624
  35. Johr R. Dermoscopy: Alternative melanocytic algorithms. Clin Dermatol. 2002;20(3):240-247. https://doi.org/10.1016/S0738-081X(02)00236-5
  36. Menzies SW, Ingvar C, Crotty KA, McCarthy WH. Frequency and morphologic characteristics of invasive melanomas lacking specific features. Arch Dermatol. 1996;132(10):1178-1182. https://doi.org/10.1001/archderm.1996.03890340038007
  37. Kreusch J. Vascular patterns in skin tumors. Clin Dermatol. 2002;20(3):248-254. https://doi.org/10.1016/S0738-081X(02)00227-4
  38. Puig S, Malvehy J. Monitoring patients with multiple nevi. Dermatol Clin. 2013;31(4):565-577. https://doi.org/10.1016/j.det.2013.06.004