Skin cancer classification via convolutional neural networks: systematic review of studies involving human experts. (October 2021)

• Record Type: Journal Article
• Authors: Haggenmüller, Sarah; Maron, Roman C.; Hekler, Achim; Utikal, Jochen S.; Barata, Catarina; Barnhill, Raymond L.; Beltraminelli, Helmut; Berking, Carola; Betz-Stablein, Brigid; Blum, Andreas; Braun, Stephan A.; Carr, Richard; Combalia, Marc; Fernandez-Figueras, Maria-Teresa; Ferrara, Gerardo; Fraitag, Sylvie; French, Lars E.; Gellrich, Frank F.; Ghoreschi, Kamran; Goebeler, Matthias; Guitera, Pascale; Haenssle, Holger A.; Haferkamp, Sebastian; Heinzerling, Lucie; Heppt, Markus V.; Hilke, Franz J.; Hobelsberger, Sarah; Krahl, Dieter; Kutzner, Heinz; Lallas, Aimilios; Liopyris, Konstantinos; Llamas-Velasco, Mar; Malvehy, Josep; Meier, Friedegund; Müller, Cornelia S.L.; Navarini, Alexander A.; Navarrete-Dechent, Cristián; Perasole, Antonio; Poch, Gabriela; Podlipnik, Sebastian; Requena, Luis; Rotemberg, Veronica M.; Saggini, Andrea; Sangueza, Omar P.; Santonja, Carlos; Schadendorf, Dirk; Schilling, Bastian; Schlaak, Max; Schlager, Justin G.; Sergon, Mildred; Sondermann, Wiebke; Soyer, H. Peter; Starz, Hans; Stolz, Wilhelm; Vale, Esmeralda; Weyers, Wolfgang; Zink, Alexander; Krieghoff-Henning, Eva; Kather, Jakob N.; von Kalle, Christof; Lipka, Daniel B.; Fröhling, Stefan; Hauschild, Axel; Kittler, Harald; Brinker, Titus J.;
• Is Part Of: European journal of cancer. Volume 156(2021)
• Page Start: 202
• Page End: 216
• Publication Date: 2021-10
• Subjects: Skin cancer classification -- Digital biomarkers -- Convolutional neural network(s) -- Artificial intelligence -- Machine learning -- Deep learning -- Dermatology -- Malignant melanoma; Cancer -- Periodicals; Neoplasms -- Periodicals; Cancer -- Périodiques; Cancer; Tumors; Electronic journals; Periodicals; Electronic journals; 616.994
• DOI: 10.1016/j.ejca.2021.06.049
• Languages: English
• ISSNs: 0959-8049

Abstract

Abstract: Background: Multiple studies have compared the performance of artificial intelligence (AI)–based models for automated skin cancer classification to human experts, thus setting the cornerstone for a successful translation of AI-based tools into clinicopathological practice. Objective: The objective of the study was to systematically analyse the current state of research on reader studies involving melanoma and to assess their potential clinical relevance by evaluating three main aspects: test set characteristics (holdout/out-of-distribution data set, composition), test setting (experimental/clinical, inclusion of metadata) and representativeness of participating clinicians. Methods: PubMed, Medline and ScienceDirect were screened for peer-reviewed studies published between 2017 and 2021 and dealing with AI-based skin cancer classification involving melanoma. The search terms skin cancer classification, deep learning, convolutional neural network (CNN), melanoma (detection), digital biomarkers, histopathology and whole slide imaging were combined. Based on the search results, only studies that considered direct comparison of AI results with clinicians and had a diagnostic classification as their main objective were included. Results: A total of 19 reader studies fulfilled the inclusion criteria. Of these, 11 CNN-based approaches addressed the classification of dermoscopic images; 6 concentrated on the classification of clinical images, whereas 2 dermatopathological studies utilised digitised histopathological whole slide images. Conclusions: All 19 included studies demonstrated superior or at least equivalent performance of CNN-based classifiers compared with clinicians. However, almost all studies were conducted in highly artificial settings based exclusively on single images of the suspicious lesions. Moreover, test sets mainly consisted of holdout images and did not represent the full range of patient populations and melanoma subtypes encountered in clinical practice. Highlights: We analysed studies comparing artificial intelligence (AI)–based skin cancer classifiers with clinicians. All 19 included studies showed superior or equivalent performance of AI. Studies were often carried out in artificial settings, with little external testing. To enhance clinical relevance of reader studies, less artificial settings are vital. Future comparisons should be validated with the use of external test sets.