Philipp Tschandl1,2, Cliff Rosendahl3,4, Bengu Nisa Akay5, Giuseppe Argenziano6, Andreas Blum7, Ralph P Braun8, Horacio Cabo9, Jean-Yves Gourhant10, Jürgen Kreusch11, Aimilios Lallas12, Jan Lapins13, Ashfaq Marghoob14, Scott Menzies15, Nina Maria Neuber2, John Paoli16, Harold S Rabinovitz17, Christoph Rinner18, Alon Scope19, H Peter Soyer20, Christoph Sinz2, Luc Thomas21, Iris Zalaudek22, Harald Kittler2. 1. School of Computing Science, Simon Fraser University, Burnaby, British Columbia, Canada. 2. Vienna Dermatologic Imaging Research Group, Department of Dermatology, Medical University of Vienna, Vienna, Austria. 3. School of Medicine, The University of Queensland, Brisbane, Queensland, Australia. 4. School of Medicine, Tehran University of Medical Sciences, Tehran, Iran. 5. Department of Dermatology, Ankara University Faculty of Medicine, Ankara, Turkey. 6. Dermatology Unit, University of Campania, Naples, Italy. 7. Public, Private and Teaching Practice of Dermatology, Konstanz, Germany. 8. Department of Dermatology, University Hospital Zürich, Zürich, Switzerland. 9. Department of Dermatology, Instituto de Investigaciones Médicas ALanari, University of Buenos Aires, Buenos Aires, Argentina. 10. Centre de Dermatologie, Nemours, France. 11. private practice, Lübeck, Germany. 12. First Department of Dermatology, Aristotle University, Thessaloniki, Greece. 13. Department of Dermatology, Karolinska University Hospital and Karolinska Institutet, Stockholm, Sweden. 14. Dermatology Service, Memorial Sloan Kettering Cancer Center, Hauppauge, New York. 15. Sydney Melanoma Diagnostic Centre and Discipline of Dermatology, University of Sydney, Sydney, Australia. 16. Department of Dermatology, Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden. 17. Skin and Cancer Associates, Plantation, Florida. 18. Center for Medical Statistics, Informatics and Intelligent Systems, Medical University of Vienna, Vienna, Austria. 19. Medical Screening Institute, Chaim Sheba Medical Center, Sackler School of Medicine, Tel Aviv University, Tel Aviv, Israel. 20. Dermatology Research Centre, The University of Queensland, The University of Queensland Diamantina Institute, Brisbane, Australia. 21. Department of Dermatology, Centre Hospitalier Lyon Sud, Lyon 1 University, Lyons Cancer Research Center, Lyon, France. 22. Dermatology Clinic, Maggiore Hospital, University of Trieste, Trieste, Italy.
Abstract
Importance: Convolutional neural networks (CNNs) achieve expert-level accuracy in the diagnosis of pigmented melanocytic lesions. However, the most common types of skin cancer are nonpigmented and nonmelanocytic, and are more difficult to diagnose. Objective: To compare the accuracy of a CNN-based classifier with that of physicians with different levels of experience. Design, Setting, and Participants: A CNN-based classification model was trained on 7895 dermoscopic and 5829 close-up images of lesions excised at a primary skin cancer clinic between January 1, 2008, and July 13, 2017, for a combined evaluation of both imaging methods. The combined CNN (cCNN) was tested on a set of 2072 unknown cases and compared with results from 95 human raters who were medical personnel, including 62 board-certified dermatologists, with different experience in dermoscopy. Main Outcomes and Measures: The proportions of correct specific diagnoses and the accuracy to differentiate between benign and malignant lesions measured as an area under the receiver operating characteristic curve served as main outcome measures. Results: Among 95 human raters (51.6% female; mean age, 43.4 years; 95% CI, 41.0-45.7 years), the participants were divided into 3 groups (according to years of experience with dermoscopy): beginner raters (<3 years), intermediate raters (3-10 years), or expert raters (>10 years). The area under the receiver operating characteristic curve of the trained cCNN was higher than human ratings (0.742; 95% CI, 0.729-0.755 vs 0.695; 95% CI, 0.676-0.713; P < .001). The specificity was fixed at the mean level of human raters (51.3%), and therefore the sensitivity of the cCNN (80.5%; 95% CI, 79.0%-82.1%) was higher than that of human raters (77.6%; 95% CI, 74.7%-80.5%). The cCNN achieved a higher percentage of correct specific diagnoses compared with human raters (37.6%; 95% CI, 36.6%-38.4% vs 33.5%; 95% CI, 31.5%-35.6%; P = .001) but not compared with experts (37.3%; 95% CI, 35.7%-38.8% vs 40.0%; 95% CI, 37.0%-43.0%; P = .18). Conclusions and Relevance: Neural networks are able to classify dermoscopic and close-up images of nonpigmented lesions as accurately as human experts in an experimental setting.
Importance: Convolutional neural networks (CNNs) achieve expert-level accuracy in the diagnosis of pigmented melanocytic lesions. However, the most common types of skin cancer are nonpigmented and nonmelanocytic, and are more difficult to diagnose. Objective: To compare the accuracy of a CNN-based classifier with that of physicians with different levels of experience. Design, Setting, and Participants: A CNN-based classification model was trained on 7895 dermoscopic and 5829 close-up images of lesions excised at a primary skin cancer clinic between January 1, 2008, and July 13, 2017, for a combined evaluation of both imaging methods. The combined CNN (cCNN) was tested on a set of 2072 unknown cases and compared with results from 95 human raters who were medical personnel, including 62 board-certified dermatologists, with different experience in dermoscopy. Main Outcomes and Measures: The proportions of correct specific diagnoses and the accuracy to differentiate between benign and malignant lesions measured as an area under the receiver operating characteristic curve served as main outcome measures. Results: Among 95 human raters (51.6% female; mean age, 43.4 years; 95% CI, 41.0-45.7 years), the participants were divided into 3 groups (according to years of experience with dermoscopy): beginner raters (<3 years), intermediate raters (3-10 years), or expert raters (>10 years). The area under the receiver operating characteristic curve of the trained cCNN was higher than human ratings (0.742; 95% CI, 0.729-0.755 vs 0.695; 95% CI, 0.676-0.713; P < .001). The specificity was fixed at the mean level of human raters (51.3%), and therefore the sensitivity of the cCNN (80.5%; 95% CI, 79.0%-82.1%) was higher than that of human raters (77.6%; 95% CI, 74.7%-80.5%). The cCNN achieved a higher percentage of correct specific diagnoses compared with human raters (37.6%; 95% CI, 36.6%-38.4% vs 33.5%; 95% CI, 31.5%-35.6%; P = .001) but not compared with experts (37.3%; 95% CI, 35.7%-38.8% vs 40.0%; 95% CI, 37.0%-43.0%; P = .18). Conclusions and Relevance: Neural networks are able to classify dermoscopic and close-up images of nonpigmented lesions as accurately as human experts in an experimental setting.
Authors: Scott W Menzies; Leanne Bischof; Hugues Talbot; Alex Gutenev; Michelle Avramidis; Livian Wong; Sing Kai Lo; Geoffrey Mackellar; Victor Skladnev; William McCarthy; John Kelly; Brad Cranney; Peter Lye; Harold Rabinovitz; Margaret Oliviero; Andreas Blum; Alexandra Varol; Alexandra Virol; Brian De'Ambrosis; Roderick McCleod; Hiroshi Koga; Caron Grin; Ralph Braun; Robert Johr Journal: Arch Dermatol Date: 2005-11
Authors: Christoph Sinz; Philipp Tschandl; Cliff Rosendahl; Bengu Nisa Akay; Giuseppe Argenziano; Andreas Blum; Ralph P Braun; Horacio Cabo; Jean-Yves Gourhant; Juergen Kreusch; Aimilios Lallas; Jan Lapins; Ashfaq A Marghoob; Scott W Menzies; John Paoli; Harold S Rabinovitz; Christoph Rinner; Alon Scope; H Peter Soyer; Luc Thomas; Iris Zalaudek; Harald Kittler Journal: J Am Acad Dermatol Date: 2017-09-20 Impact factor: 11.527
Authors: Andre Esteva; Brett Kuprel; Roberto A Novoa; Justin Ko; Susan M Swetter; Helen M Blau; Sebastian Thrun Journal: Nature Date: 2017-01-25 Impact factor: 49.962
Authors: W Wachsman; V Morhenn; T Palmer; L Walls; T Hata; J Zalla; R Scheinberg; H Sofen; S Mraz; K Gross; H Rabinovitz; D Polsky; S Chang Journal: Br J Dermatol Date: 2011-03-25 Impact factor: 9.302
Authors: J Malvehy; A Hauschild; C Curiel-Lewandrowski; P Mohr; R Hofmann-Wellenhof; R Motley; C Berking; D Grossman; J Paoli; C Loquai; J Olah; U Reinhold; H Wenger; T Dirschka; S Davis; C Henderson; H Rabinovitz; J Welzel; D Schadendorf; U Birgersson Journal: Br J Dermatol Date: 2014-10-19 Impact factor: 9.302
Authors: Michael A Marchetti; Konstantinos Liopyris; Stephen W Dusza; Noel C F Codella; David A Gutman; Brian Helba; Aadi Kalloo; Allan C Halpern Journal: J Am Acad Dermatol Date: 2019-07-12 Impact factor: 11.527
Authors: Claire M Felmingham; Nikki R Adler; Zongyuan Ge; Rachael L Morton; Monika Janda; Victoria J Mar Journal: Am J Clin Dermatol Date: 2021-03 Impact factor: 7.403
Authors: Alican Bozkurt; Kivanc Kose; Jaume Coll-Font; Christi Alessi-Fox; Dana H Brooks; Jennifer G Dy; Milind Rajadhyaksha Journal: Sci Rep Date: 2021-06-15 Impact factor: 4.379