Refine
Has Fulltext
- yes (2)
Is part of the Bibliography
- yes (2)
Document Type
- Journal article (2)
Language
- English (2)
Keywords
- Background Epithelioid haemangioma (1)
- artificial intelligence (1)
- deep learning (1)
- melanoma (1)
- protein kinase pathway (1)
- skin cancer (1)
- somatic mutations (1)
Background
Epithelioid haemangioma (EH) arising from the skin is a benign vascular tumour with marked inflammatory cell infiltration, which exhibits a high tendency to persist and frequently recurs after resection. So far, the underlying pathogenesis is largely elusive.
Objectives
To identify genetic alterations by next-generation sequencing and/or droplet digital polymerase chain reaction (ddPCR) in cutaneous EH.
Methods
DNA and RNA from an EH lesion of an index patient were subjected to whole-genome and RNA sequencing. Multiplex PCR-based panel sequencing of genomic DNA isolated from archival formalin-fixed paraffin-embedded tissue of 18 patients with cutaneous EH was performed. ddPCR was used to confirm mutations.
Results
We identified somatic mutations in genes of the mitogen-activated protein kinase (MAPK) pathway (MAP2K1 and KRAS) in cutaneous EH biopsies. By ddPCR we could confirm the recurrent presence of activating, low-frequency mutations affecting MAP2K1. In total, nine out of 18 patients analysed showed activating MAPK pathway mutations, which were mutually exclusive. Comparative analysis of tissue areas enriched for lymphatic infiltrate or aberrant endothelial cells, respectively, revealed an association of these mutations with the presence of endothelial cells.
Conclusions
Taken together, our data suggest that EH shows somatic mutations in genes of the MAPK pathway which might contribute to the formation of this benign tumour.
Background
Melanoma is the most dangerous type of skin cancer but is curable if detected early. Recent publications demonstrated that artificial intelligence is capable in classifying images of benign nevi and melanoma with dermatologist-level precision. However, a statistically significant improvement compared with dermatologist classification has not been reported to date.
Methods
For this comparative study, 4204 biopsy-proven images of melanoma and nevi (1:1) were used for the training of a convolutional neural network (CNN). New techniques of deep learning were integrated. For the experiment, an additional 804 biopsy-proven dermoscopic images of melanoma and nevi (1:1) were randomly presented to dermatologists of nine German university hospitals, who evaluated the quality of each image and stated their recommended treatment (19,296 recommendations in total). Three McNemar's tests comparing the results of the CNN's test runs in terms of sensitivity, specificity and overall correctness were predefined as the main outcomes.
Findings
The respective sensitivity and specificity of lesion classification by the dermatologists were 67.2% (95% confidence interval [CI]: 62.6%–71.7%) and 62.2% (95% CI: 57.6%–66.9%). In comparison, the trained CNN achieved a higher sensitivity of 82.3% (95% CI: 78.3%–85.7%) and a higher specificity of 77.9% (95% CI: 73.8%–81.8%). The three McNemar's tests in 2 × 2 tables all reached a significance level of p < 0.001. This significance level was sustained for both subgroups.
Interpretation
For the first time, automated dermoscopic melanoma image classification was shown to be significantly superior to both junior and board-certified dermatologists (p < 0.001).