Original article
The dermoscopic inverse approach significantly improves the accuracy of human readers for lentigo maligna diagnosis

https://doi.org/10.1016/j.jaad.2020.06.085Get rights and content

Background

A recently introduced dermoscopic method for the diagnosis of early lentigo maligna (LM) is based on the absence of prevalent patterns of pigmented actinic keratosis and solar lentigo/flat seborrheic keratosis. We term this the inverse approach.

Objective

To determine whether training on the inverse approach increases the diagnostic accuracy of readers compared to classic pattern analysis.

Methods

We used clinical and dermoscopic images of histopathologically diagnosed LMs, pigmented actinic keratoses, and solar lentigo/flat seborrheic keratoses. Participants in a dermoscopy masterclass classified the lesions at baseline and after training on pattern analysis and the inverse approach. We compared their diagnostic performance among the 3 timepoints and to that of a trained convolutional neural network.

Results

The mean sensitivity for LM without training was 51.5%; after training on pattern analysis, it increased to 56.7%; and after learning the inverse approach, it increased to 83.6%. The mean proportions of correct answers at the 3 timepoints were 62.1%, 65.5, and 78.5%. The percentages of readers outperforming the convolutional neural network were 6.4%, 15.4%, and 53.9%, respectively.

Limitations

The experimental setting and the inclusion of histopathologically diagnosed lesions only.

Conclusions

The inverse approach, added to the classic pattern analysis, significantly improves the sensitivity of human readers for early LM diagnosis.

Section snippets

Methods

This diagnostic study was held during a 3-day dermoscopy masterclass, using a data set of facial pigmented macules, histopathologically diagnosed as LM, PAK, or SL/SK. The participants were asked to classify the lesions at 3 different timepoints, using a voting system with manual devices. One screen was placed in front of every 3 participants, and a video wall projection was visible to all of them. The study was conducted by using appropriately anonymized data sets and, therefore, ethics

Results

The male-to-female ratio of the 78 participants was 1:3.4, and the mean age was 41.1 years, ranging from 24 to 74 years. The main results are shown in Table II.

Discussion

Our study shows that the inverse approach significantly improves the ability of clinicians to accurately classify flat pigmented facial lesions. The improvement is most pronounced in the sensitivity for LM, which is the most relevant diagnostic measure from an outcome perspective. Although our study was not conducted in a clinical setting, the remarkable improvement in all diagnostic measures strongly suggests that the application of the inverse approach could significantly facilitate the

References (34)

  • H.A. Haenssle et al.

    Man against machine reloaded: performance of a market-approved convolutional neural network in classifying a broad spectrum of skin lesions in comparison with 96 dermatologists working under less artificial conditions

    Ann Oncol

    (2020)
  • J.K. Winkler et al.

    Melanoma recognition by a deep learning convolutional neural network-Performance in different melanoma subtypes and localisations

    Eur J Cancer

    (2020)
  • C. Garbe et al.

    European consensus-based interdisciplinary guideline for melanoma. Part 1: diagnostics—update 2019

    Eur J Cancer

    (2020)
  • L.M. Cohen

    Lentigo maligna and lentigo maligna melanoma

    J Am Acad Dermatol

    (1995)
  • B.N. Akay et al.

    Dermatoscopy of flat pigmented facial lesions: diagnostic challenge between pigmented actinic keratosis and lentigo maligna

    Br J Dermatol

    (2010)
  • A. Lallas et al.

    Management rules to detect melanoma

    Dermatology

    (2013)
  • Z. Hongyan

    Dermatoscopy of pigmented actinic keratosis—a striking similarity to lentigo maligna

    Int J Dermatol

    (2007)
  • Cited by (0)

    Funding sources: None.

    Conflicts of interest: None disclosed.

    IRB approval status: Not applicable.

    Reprints not available from the authors.

    View full text