Figure 2.

Deep learning is a subset of machine learning based on deep neural networks, where multiple sequentially ordered layers allow for the network to form increasingly complex conclusions. The input layer includes pixels from the photograph, while each hidden layer extracts certain features from the photo to reach a conclusion in the output layer. In this example, the first hidden layer might detect the edges of the lesion, and deeper hidden layers might recognize that certain edge patterns are consistent with a neoplasm. The deepest layer would classify the lesion into the output category of squamous cell carcinoma. Created with BioRender.com.