Zusammenfassung
Kutane Malignitäten sind eine der häufigsten dermatologischen Diagnosen und können je nach Subtyp mit einer hohen Mortalität und Morbidität einhergehen. Die initiale Diagnostik erfolgt üblicherweise mittels makroskopischer und dermatoskopischer Beurteilung, wobei insbesondere melanozytäre Läsionen teilweise sehr herausfordernd sein können. Die künstliche Intelligenz (KI) ist eine Unterform des maschinellen Lernens und ermöglicht es, Muster aus riesigen, standardisierten Datenmengen zu lernen und diese erfolgreich auf neue Daten anzuwenden. Dabei können auch Muster erkannt werden, die sich der menschlichen Wahrnehmung entziehen. Die Daten in der Dermatoonkologie, die neben klinischen und dermatoskopischen Bildern auch 3‑D-Ganzkörperaufnahmen, histologische Schnitte und Ausgaben von neuen Bildgebungssystemen wie der Line-Field-Technik bei der optischen Kohärenztomographie umfassen, eigenen sich daher sehr gut für die Analyse mittels KI. Während es für die Analyse von klinischen/dermatoskopischen Bildern für die Klassifikation von melanozytären Läsionen bereits marktreife Systeme gibt, steckt die KI-Forschung für andere Fragestellungen noch in den Kinderschuhen. Insbesondere für die Analyse von hochstandardisierten Daten wie 3‑D-Ganzkörperaufnahmen, histopathologischen Schnitten und Daten der neuen Bildgebungsverfahren ist das Potenzial der KI-Analysen noch nicht vollständig ausgeschöpft. Der Einsatz von KI in der Dermatoonkologie muss jedoch immer kritisch und vor dessen technischem Hintergrund erfolgen. Hürden und Limitationen sind insbesondere die Notwendigkeit standardisierter Daten und die Spezifität der Algorithmen auf die Fragestellung, für welche sie trainiert wurden.
Abstract
Cutaneous malignancies are one of the most common dermatological diagnoses and—depending on subtype—are associated with high mortality and morbidity. Initial assessment includes macroscopic and dermoscopic analysis of the lesion, where melanocytic lesions can be especially challenging. Artificial intelligence (AI) is a subtype of machine learning and enables patterns to be learnt on huge, standardized datasets, with subsequent successful application of the algorithm to new data. AI algorithms can also learn patterns which elude human perception. Data in dermato-oncology include not only clinical and dermoscopic images, but also 3D full-body scans, histological slides, and the output of new imaging modalities such as line-field optical coherence tomography, all of which are well suited for AI analysis. While for AI-based assessment of melanocytic lesions on clinical/dermoscopic images there are already market-ready applications, AI research in other domains of dermato-oncology is still in its infancy. Particularly for the analysis of highly standardized data like 3D full-body scans, histological slides, and data from new imaging modalities, the full potential of AI analysis not yet exhausted. AI in dermato-oncology must be used critically and against its technical background. Limitations and obstacles include the need for standardized data and the specificity of AI algorithms for the question they have been trained to answer.
Literatur
Schadendorf D et al (2015) Melanoma. Nat Rev Dis Primer 1:1–20
Therapie-Handbuch – Dermatologie und Allergologie. (Elsevier, 2022).
Keung EZ, Gershenwald JE (2018) The eighth edition American Joint Committee on Cancer (AJCC) melanoma staging system: implications for melanoma treatment and care. Expert Rev Anticancer Ther 18:775–784
Mintz Y, Brodie R (2019) Introduction to artificial intelligence in medicine. Minim Invasive Ther Allied Technol 28:73–81
Dai Z, Liu H, Le QV, Tan M (2021) Coatnet: marrying convolution and attention for all data sizes. ArXiv210604803 Cs
Hosny A, Parmar C, Quackenbush J, Schwartz LH, Aerts HJWL (2018) Artificial intelligence in radiology. Nat Rev Cancer 18:500–510
Winkler JK, Haenssle HA (2022) Bilderkennung mittels künstlicher Intelligenz in der Hautkrebsdiagnostik. Dermatol 73:838–844
Efimenko M, Ignatev A, Koshechkin K (2020) Review of medical image recognition technologies to detect melanomas using neural networks. BMC Bioinform 21:270
Fabbrocini G et al (2011) Teledermatology: from prevention to diagnosis of nonmelanoma and melanoma skin cancer. Int J Telemed Appl. https://doi.org/10.1155/2011/125762
Ali A‑RA, Deserno TM (2012) A systematic review of automated melanoma detection in dermatoscopic images and its ground truth data. Med Imaging 2012: Image Percept Obs Perform Technol Assess 8318:421–431 (SPIE, 2012). https://doi.org/10.1117/12.912389
Carli P et al (2003) Pattern analysis, not simplified algorithms, is the most reliable method for teaching dermoscopy for melanoma diagnosis to residents in dermatology. Br J Dermatol 148:981–984
Dolianitis C, Kelly J, Wolfe R, Simpson P (2005) Comparative performance of 4 dermoscopic algorithms by nonexperts for the diagnosis of melanocytic lesions. Arch Dermatol 141:1008–1014
Brinker TJ et al (2018) Skin cancer classification using convolutional neural networks: systematic review. J Med Internet Res 20:e11936
Esteva A et al (2017) Dermatologist-level classification of skin cancer with deep neural networks. Nature 542:115–118
Haenssle HA et al (2018) Man against machine: diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists. Ann Oncol 29:1836–1842
Haenssle HA et al (2021) Skin lesions of face and scalp—Classification by a market-approved convolutional neural network in comparison with 64 dermatologists. Eur J Cancer 144:192–199
Jaworek-Korjakowska J, Kłeczek P (2016) Automatic classification of specific melanocytic lesions using artificial intelligence. Biomed Res Int. https://doi.org/10.1155/2016/8934242
Liu Y et al (2020) A deep learning system for differential diagnosis of skin diseases. Nat Med 26:900–908
Höhn J et al (2021) Integrating patient data into skin cancer classification using convolutional neural networks: systematic review. J Med Internet Res 23:e20708
Winkler JK et al (2019) Association between surgical skin markings in dermoscopic images and diagnostic performance of a deep learning convolutional neural network for melanoma recognition. JAMA Dermatol 155:1135–1141
Winkler JK et al (2021) Association between different scale bars in dermoscopic images and diagnostic performance of a market-approved deep learning convolutional neural network for melanoma recognition. Eur J Cancer 145:146–154
Sitaru S, Kaczmarczyk R, Erdmann M, Biedermann T, Zink A (2022) 3D whole body scans in dermatology—a new era in clinical practice and research? Dermatologie 73:575–579
Rutjes C, Torrano J, Soyer HP (2022) A 3D total-body photography research network: the Australian experiment. Hautarzt. https://doi.org/10.1007/s00105-021-04938-7
Freeman K et al (2020) Algorithm based smartphone apps to assess risk of skin cancer in adults: systematic review of diagnostic accuracy studies. BMJ 368:m127
Matin RN, Dinnes J (2021) AI-based smartphone apps for risk assessment of skin cancer need more evaluation and better regulation. Br J Cancer 124:1749–1750
Schmidle P, Braun SA (2022) Digitalisierung in der Dermatopathologie. Dermatologie 73:845–852
Le’Clerc Arrastia J et al (2021) Deeply supervised unet for semantic segmentation to assist dermatopathological assessment of basal cell carcinoma. J Imaging 7:71
Figueroa-Silva O, Pastur Romay LA, Viruez Roca RD, Rojas MDS-AY, Suárez-Peñaranda JM (2022) Machine learning techniques in predicting BRAF mutation status in cutaneous melanoma from clinical and histopathologic features. Appl Immunohistochem Mol Morphol 30:674–680
Mindpeak News | BreastIHC CE-IVD certified. https://mindpeak.ai//post/breastihc-ce-ivd-certified. Zugegriffen: 04.01.2023
Lee K, Lee SH (2020) Artificial intelligence-driven oncology clinical decision support system for multidisciplinary teams. Sensors 20:E4693
Gilmore SJ (2018) Automated decision support in melanocytic lesion management. PLoS ONE 13:e203459
Liu W, Zhu Y, Lin C, Liu L, Li G (2022) An online prognostic application for melanoma based on machine learning and statistics. J Plast Reconstr Aesthetic Surg 75:3853–3858
Johannet P et al (2021) Using machine learning algorithms to predict immunotherapy response in patients with advanced melanoma. Clin Cancer Res 27:131–140
Indini A, Di Guardo L, Cimminiello C, De Braud F, Del Vecchio M (2019) Artificial intelligence estimates the importance of baseline factors in predicting response to anti-PD1 in metastatic melanoma. Am J Clin Oncol 42:643–648
Kong J et al (2022) Network-based machine learning approach to predict immunotherapy response in cancer patients. Nat Commun 13:3703
Liu M, Xu Y (2022) Gene identification and potential drug therapy for drug-resistant melanoma with bioinformatics and deep learning technology. Dis Markers. https://doi.org/10.1155/2022/2461055
Lewinson RT et al (2021) Machine learning for prediction of cutaneous adverse events in patients receiving anti-PD‑1 immunotherapy. J Am Acad Dermatol 84:183–185
Lambin P et al (2017) Decision support systems for personalized and participative radiation oncology. Adv Drug Deliv Rev 109:131–153
Pellacani G, Pepe P, Casari A, Longo C (2014) Reflectance confocal microscopy as a second-level examination in skin oncology improves diagnostic accuracy and saves unnecessary excisions: a longitudinal prospective study. Br J Dermatol 171:1044–1051
Dinnes J et al (2018) Reflectance confocal microscopy for diagnosing cutaneous melanoma in adults. Cochrane Database Syst Rev. https://doi.org/10.1002/14651858.CD013190
Sies K, Winkler JK, Zieger M, Kaatz M, Haenssle HA (2020) Neue optische Untersuchungsverfahren für die Diagnostik von Hautkrankheiten. Hautarzt 71:101–108
Schuh S, Ruini C, Sattler E, Welzel J (2021) Konfokale Line-Field-OCT. Hautarzt 72:1039–1047
Perez-Anker J et al (2022) Morphological evaluation of melanocytic lesions with three-dimensional line-field confocal optical coherence tomography: correlation with histopathology and reflectance confocal microscopy. A pilot study. Clin Exp Dermatol. https://doi.org/10.1111/ced.15383
Chou H‑Y, Huang S‑L, Tjiu J‑W, Chen HH (2021) Dermal epidermal junction detection for full-field optical coherence tomography data of human skin by deep learning. Comput Med Imaging Graph 87:101833
Fischman S et al (2022) Non-invasive scoring of cellular atypia in keratinocyte cancers in 3D LC-OCT images using Deep Learning. Sci Rep 12:481
Warnat-Herresthal S et al (2021) Swarm Learning for decentralized and confidential clinical machine learning. Nature 594:265–270
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Interessenkonflikt
S. Sitaru und A. Zink geben an, dass kein Interessenkonflikt besteht.
Für diesen Beitrag wurden von den Autor/-innen keine Studien an Menschen oder Tieren durchgeführt. Für die aufgeführten Studien gelten die jeweils dort angegebenen ethischen Richtlinien.
Additional information
Redaktion
Hans-Joachim Schulze, Münster
Rights and permissions
About this article
Cite this article
Sitaru, S., Zink, A. Digitalisierung in der Dermatoonkologie: künstliche Intelligenz zur Diagnostik. best practice onkologie 18, 20–26 (2023). https://doi.org/10.1007/s11654-022-00461-w
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11654-022-00461-w