Deep learning outperformed 11 pathologists in the classification of histopathological melanoma images

Abstract

Background: The diagnosis of most cancers is made by a board-certified pathologist based on a tissue biopsy under the microscope. Recent research reveals a high discordance between individual pathologists. For melanoma, the literature reports on 25e26% of discordance for classifying a benign nevus versus malignant melanoma. A recent study indicated the potential of deep learning to lower these discordances. However, the performance of deep learning in classifying histopathologic melanoma images was never compared directly to human experts. The aim of this study is to perform such a first direct comparison. Methods: A total of 695 lesions were classified by an expert histopathologist in accordance with current guidelines (350 nevi/345 melanoma). Only the haematoxylin & eosin (H&E) slides

Precision medicine for human cancers with Notch signaling dysregulation (Review)

“…Whole-genome sequencing, as well as whole-exome sequencing plus transcriptome analysis, is applied for the exploration of unknown cancer drivers, and the development of novel therapeutics for known but intractable targets with the aid of human intelligence, cognitive computing and artificial intelligence in basic and translational oncology (205)(206)(207)(208). Moreover, artificial intelligence is also applied for computer-aided diagnostic approaches (209,210), such as chest computed tomography (211), dermoscopy (212), gastrointestinal endoscopy (213), mammography (214) and histopathological diagnosis (215)(216)(217)(218). To avoid the lack of transparency associated with black box artificial intelligence based on deep learning technologies, the development of explainable artificial intelligence is necessary (219).…”

Reply to the letter to the editor: ‘Deep learning outperformed 11 pathologists in the classification of histopathological melanoma images’

“…The authors view their work in this new field as what it is, the first two pilot studies implementing deep learning into the histopathologic analysis of melanoma. They are no definitive studies, and the limitations have been extensively discussed in both manuscripts [1,2]. While we agree that the data are preliminary data and acknowledge this in both manuscripts, we argue that our studies are important as they reveal the potential of deep learning in this field and introduce a new application.…”

Machine Learning in Dermatology: Current Applications, Opportunities, and Limitations

“…Hekler and colleagues claimed that a CNN which they tested outperformed 11 pathologists in the classification of histopathological melanoma images [73]. In this study, 695 lesions were classified by one expert histopathologist using tissue slides stained with hematoxylin and eosin.…”

“…Chief physicians had the highest mean specificity of 69.2% and a mean sensitivity of 73.3%. At the same mean specificity of 69.2%, the CNN had a mean sensitivity of 84.5% [73].…”