Artificial Intelligence

From dermoscopedia
Main PageDigital dermoscopyArtificial Intelligence
(0 votes)
 Author(s): N. N.
Description This chapter explains how artificial intelligence works in dermoscopy
Author(s) N. N.
Responsible author N. N.→ send e-mail
Status open
Status update August 25, 2023
Status by Ralph P. Braun

Introduction to computer assisted diagnosis

The early diagnosis of melanoma remains challenging.[1] Meta-analysis has demonstrated that the use of dermoscopy by trained users improves diagnostic accuracy for melanoma compared to naked-eye examination alone, mainly by increasing sensitivity.[2] [3] [4] As non-physicians detect the majority of melanomas[5] and population-based melanoma screening by clinicians is not currently recommended [6] outside of Germany, there is interest in the development of dermoscopy image analysis algorithms to aid laypersons or non-dermatology physicians in melanoma detection.[7] [8] [9] [10] [11]

There have been many efforts to train algorithms to diagnose melanoma.[12] [13] [14]

A recent landmark paper demonstrates the potential of ‘deep learning’ algorithms to aid in melanoma diagnosis.[15] In their study, the researchers used deep convolutional neural networks (CNNs) which they trained in a large dataset of clinical images and then compared against 21 dermatologists for the binary classification of keratinocyte carcinomas versus benign seborrheic keratoses; and malignant melanomas versus benign nevi, with the CNN achieving performance on par with all tested experts across both tasks, demonstrating an artificial intelligence capable of classifying skin cancer with a level of competence comparable to dermatologists. While the potential for computer assisted diagnosis is great, the lack of a large public dataset of validated skin images has limited the ability to directly compare the diagnostic performance of automated image analysis algorithms across studies and against clinicians. The ISIC Archive (link available on dermoscopedia menu bar) was created to address this need and has hosted consecutive international ‘Challenges’ that have engaged the broader computer science community.

There are many potential barriers to the implementation of automated diagnosis in clinical care (e.g., vetting of the accuracy of the algorithms, potential for overdiagnosis and false reassurance, acceptance by the medical community) but it is logical to expect that some form of computer assisted triage or diagnosis is forthcoming in the near future. There are already hundreds of ‘apps’ for mobile devices that are being marketed as educational or assistive devices for melanoma early detection.[16] The medicolegal and social landscape around the adoption of these apps is rapidly changing.

Artificial intelligence (AI) algorithms have been developed in order to improve diagnostic accuracy of melanoma. They have been recently shown to be reliable tools, classifying melanoma with a level of competence comparable to a dermatologist [17].

In a recent study on 1550 images of skin lesions, more than half of the melanoma diagnoses were either in situ or <1 mm deep, indicating that the studied algorithm could play a role in detecting thin or early-stage lesions. Of course, diagnostic specificity varied with the camera used to photograph the lesions [17].

Automated diagnosis

Neural networks

Machine learning techniques, and neural networks in particular, for diagnosing melanoma have been in focus of research for more than two decades in dermatology[18][10]. A systematic review as early as 2009 has shown that automated classifiers can have a comparable diagnostic odds ratio to physicians[13]. Systems reviewed in this case commonly relied on manual or semi-automatic feature extraction of digital images[19]. Unprecedented image classification accuracy through so-called deep convolutional neural networks (CNN), which don't need a feature extraction step, has regained interest in automated analyses in the medical domain.

Comparison to human accuracy

Several studies found human level performance of such state-of-the-art CNNs to dermatologists for melanoma diagnosis[15][20][21][22][23] or multi-class accuracy[20]. Similar rates were found for classification of close-up images[15][24][25][24][26].

A more current endeavour was comparing multi-class accuracy of the ISIC 2018 challenge winners to human readers in a world wide and open online game. This study [27] showed for the first time that even average machine learning algorithms can show a higher accuracy in a multi-class setting encompassing all relevant pigmented lesion types.


CNNs need a large number of images during training, are thus highly dependant on datasets providing those - and are currently limited to detecting diagnoses covered in them. At the moment the largest publicly available datasets are:

The future central hub for any public dataset is probably the ISIC-Archive, where already datasets from Memorial Sloan Kettering Cancer Center, the SONIC study[29] and others are available in a structured way.


Although published accuracy rates for neural network classifiers - similar to studies >10 years ago - seem promising, researchers are raising concerns whether neural network classification can live up to its current promise to "solve" automated melanoma recognition[30][31]. This is due to many reasons, some exemplarily shown below:

  • Instability: Current neural networks are highly sensitive to small image perturbations invisible to the human eye, shown by adversarial attacks against medical images. But also more evident differences in images can have an impact: E.g. if only melanoma images in the training dataset are photographed with skin markings or rulers, a network may also "diagnose" images of angiomas as melanomas if depicted alongside rulers and skin markings[31].
  • Explainability: Although every parameter of a neural network is known, explainability of those systems is impeded by the vast number of parameters. One approach for intuitively explaining neural network decisions is by simply providing similar images with "content based image retrieval" (CBIR). A free non-commercial example application can be used at
  • Evaluation: Classic evaluation metrics of studies and machine learning challenges show comparability to human accuracy, but it is debateable if they provide valuable insight for performance and clinical applicability. Also, a recent Meta-Analysis has shown significant bias of metrics connected to known test datasets[32]
  • Real world application: While recent studies show users accept automated classifiers in principle[33], more in depth early studies showed that only non-experts, and only in ambiguous cases, would change their decision based on an automated system[34]. It is completely unclear whether published metrics accurately reflect this population and cases, and whether automated decisions in those cases can provide beneficial and safe outcomes. An underestimated danger is also that an automated system can of course not diagnose a melanoma that is not being photographed at all, which was in fact the case in a prospective clinica trial in Vienna[35].

Device-based automated diagnostics

Apart from image analysis and classification of dermatoscopic images, several other techniques try to solve melanoma diagnosis (list non-exhaustive):

  • Dermatofluoroscopy[36]
  • Multispectral image analysis[37]
  • Raman spectroscopy [38]
  • Electrical Impedance Spectroscopy[39]
  • Genomic analysis through adhesive tape[40]

They all try to provide a binary output with little to no user interaction. Albeit accuracy rates have been reported high for some of those, it is still unclear if and how they will provide meaningful input in clinical practice.


  1. Marghoob & Scope: The complexity of diagnosing melanoma. J. Invest. Dermatol. 2009;129:11-3. PMID: 19078984. DOI.
  2. Kittler et al.: Diagnostic accuracy of dermoscopy. Lancet Oncol. 2002;3:159-65. PMID: 11902502.
  3. Bafounta et al.: Is dermoscopy (epiluminescence microscopy) useful for the diagnosis of melanoma? Results of a meta-analysis using techniques adapted to the evaluation of diagnostic tests. Arch Dermatol 2001;137:1343-50. PMID: 11594860.
  4. Vestergaard et al.: Dermoscopy compared with naked eye examination for the diagnosis of primary melanoma: a meta-analysis of studies performed in a clinical setting. Br. J. Dermatol. 2008;159:669-76. PMID: 18616769. DOI.
  5. Brady et al.: Patterns of detection in patients with cutaneous melanoma. Cancer 2000;89:342-7. PMID: 10918164.
  6. Bibbins-Domingo et al.: Screening for Skin Cancer: US Preventive Services Task Force Recommendation Statement. JAMA 2016;316:429-35. PMID: 27458948. DOI.
  7. Ferris et al.: Computer-aided classification of melanocytic lesions using dermoscopic images. J. Am. Acad. Dermatol. 2015;73:769-76. PMID: 26386631. DOI.
  8. Zortea et al.: Performance of a dermoscopy-based computer vision system for the diagnosis of pigmented skin lesions compared with visual evaluation by experienced dermatologists. Artif Intell Med 2014;60:13-26. PMID: 24382424. DOI.
  9. Blum et al.: Digital image analysis for diagnosis of cutaneous melanoma. Development of a highly effective computer algorithm based on analysis of 837 melanocytic lesions. Br. J. Dermatol. 2004;151:1029-38. PMID: 15541081. DOI.
  10. 10.0 10.1 Menzies et al.: The performance of SolarScan: an automated dermoscopy image analysis instrument for the diagnosis of primary melanoma. Arch Dermatol 2005;141:1388-96. PMID: 16301386. DOI.
  11. Rubegni et al.: Evaluation of cutaneous melanoma thickness by digital dermoscopy analysis: a retrospective study. Melanoma Res. 2010;20:212-7. PMID: 20375922. DOI.
  12. Masood & Al-Jumaily: Computer aided diagnostic support system for skin cancer: a review of techniques and algorithms. Int J Biomed Imaging 2013;2013:323268. PMID: 24575126. DOI.
  13. 13.0 13.1 Rajpara et al.: Systematic review of dermoscopy and digital dermoscopy/ artificial intelligence for the diagnosis of melanoma. Br. J. Dermatol. 2009;161:591-604. PMID: 19302072. DOI.
  14. Maglogiannis & Doukas: Overview of advanced computer vision systems for skin lesions characterization. IEEE Trans Inf Technol Biomed 2009;13:721-33. PMID: 19304487. DOI.
  15. 15.0 15.1 15.2 Esteva et al.: Dermatologist-level classification of skin cancer with deep neural networks. Nature 2017;542:115-118. PMID: 28117445. DOI.
  16. March et al.: Practical application of new technologies for melanoma diagnosis: Part I. Noninvasive approaches. J. Am. Acad. Dermatol. 2015;72:929-41; quiz 941-2. PMID: 25980998. DOI.
  17. 17.0 17.1 Phillips et al.: Assessment of Accuracy of an Artificial Intelligence Algorithm to Detect Melanoma in Images of Skin Lesions. JAMA Netw Open 2019;2:e1913436. PMID: 31617929. DOI.
  18. Binder et al.: Application of an artificial neural network in epiluminescence microscopy pattern analysis of pigmented skin lesions: a pilot study. Br. J. Dermatol. 1994;130:460-5. PMID: 8186110.
  19. Rubegni et al.: Automated diagnosis of pigmented skin lesions. Int. J. Cancer 2002;101:576-80. PMID: 12237900. DOI.
  20. 20.0 20.1 Tschandl et al.: A pretrained neural network shows similar diagnostic accuracy to medical students in categorizing dermatoscopic images after comparable training conditions. Br. J. Dermatol. 2017;177:867-869. PMID: 28569993. DOI.
  21. Marchetti et al.: Results of the 2016 International Skin Imaging Collaboration International Symposium on Biomedical Imaging challenge: Comparison of the accuracy of computer algorithms to dermatologists for the diagnosis of melanoma from dermoscopic images. J. Am. Acad. Dermatol. 2018;78:270-277.e1. PMID: 28969863. DOI.
  22. Haenssle 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;. PMID: 29846502. DOI.
  23. Yu et al.: Acral melanoma detection using a convolutional neural network for dermoscopy images. PLoS ONE 2018;13:e0193321. PMID: 29513718. DOI.
  24. 24.0 24.1 Han et al.: Classification of the Clinical Images for Benign and Malignant Cutaneous Tumors Using a Deep Learning Algorithm. J. Invest. Dermatol. 2018;. PMID: 29428356. DOI.
  25. Fujisawa et al.: Deep learning-based, computer-aided classifier developed with a small dataset of clinical images surpasses board-certified dermatologists in skin tumor diagnosis. Br. J. Dermatol. 2018;. PMID: 29901853. DOI.
  26. Tschandl et al.: Expert-Level Diagnosis of Nonpigmented Skin Cancer by Combined Convolutional Neural Networks. JAMA Dermatol 2019;155:58-65. PMID: 30484822. DOI.
  27. Tschandl et al.: Comparison of the accuracy of human readers versus machine-learning algorithms for pigmented skin lesion classification: an open, web-based, international, diagnostic study. Lancet Oncol. 2019;20:938-947. PMID: 31201137. DOI.
  28. Tschandl et al.: The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions. Sci Data 2018;5:180161. PMID: 30106392. DOI.
  29. Scope et al.: Dermoscopic patterns and subclinical melanocytic nests in normal-appearing skin. Br. J. Dermatol. 2009;160:1318-21. PMID: 19416274. DOI.
  30. Marghoob et al.: Comment on: Screening for malignant melanoma-a critical assessment in historical perspective. Dermatol Pract Concept 2018;8:73-74. PMID: 29785321. DOI.
  31. 31.0 31.1 Navarrete-Dechent et al.: Automated Dermatological Diagnosis: Hype or Reality?. J. Invest. Dermatol. 2018;. PMID: 29864435. DOI.
  32. Dick et al.: Accuracy of Computer-Aided Diagnosis of Melanoma: A Meta-analysis. JAMA Dermatol 2019;. PMID: 31215969. DOI.
  33. Fink et al.: Patient acceptance and trust in automated computer-assisted diagnosis of melanoma with dermatofluoroscopy. J Dtsch Dermatol Ges 2018;. PMID: 29927518. DOI.
  34. Dreiseitl et al.: Applying a decision support system in clinical practice: results from melanoma diagnosis. AMIA Annu Symp Proc 2007;191-5. PMID: 18693824.
  35. Dreiseitl et al.: Computer versus human diagnosis of melanoma: evaluation of the feasibility of an automated diagnostic system in a prospective clinical trial. Melanoma Res. 2009;19:180-4. PMID: 19369900. DOI.
  36. Forschner et al.: Diagnostic accuracy of dermatofluoroscopy in cutaneous melanoma detection: results of a prospective multicentre clinical study in 476 pigmented lesions. Br. J. Dermatol. 2018;. PMID: 29569229. DOI.
  37. Hauschild et al.: To excise or not: impact of MelaFind on German dermatologists' decisions to biopsy atypical lesions. J Dtsch Dermatol Ges 2014;12:606-14. PMID: 24944011. DOI.
  38. Lui et al.: Real-time Raman spectroscopy for in vivo skin cancer diagnosis. Cancer Res. 2012;72:2491-500. PMID: 22434431. DOI.
  39. Malvehy et al.: Clinical performance of the Nevisense system in cutaneous melanoma detection: an international, multicentre, prospective and blinded clinical trial on efficacy and safety. Br. J. Dermatol. 2014;171:1099-107. PMID: 24841846. DOI.
  40. Wachsman et al.: Noninvasive genomic detection of melanoma. Br. J. Dermatol. 2011;164:797-806. PMID: 21294715. DOI.
Cookies help us deliver our services. By using our services, you agree to our use of cookies.