Annotation of skin image using learned feature

A dermatoscope, comprising:

• an image capture device for acquiring a dermatoscopic image in digital form;
• an image processor for applying a first and a second convolutional neural network to the acquired dermatoscopic image and identifying a region of lesion skin within the acquired dermatoscopic image based on results of the application of the first and second convolutional neural networks;
• and
• a display device for displaying the acquired dermatoscopic image with the identified region of lesion highlighted therein,
• wherein the first convolutional neural network is trained according to an interior of regions of lesion skin within a plurality of training dermatoscopic images, and
• wherein the second convolutional neural network is trained according to a boundary between the regions of lesion skin and regions of normal skin within the plurality of training dermatoscopic images.

A method for annotation of skin images, comprising:

• receiving a first convolutional neural network that has been trained according to an interior of a region of lesion skin using each of a plurality of dermatoscopic training images;
• receiving a second convolutional neural network that has been trained according to a boundary between the region of lesion skin and the region of normal skin in each of the plurality of dermatoscopic training images;
• acquiring a clinical dermatoscopic image;
• and
• using the received first and second convolutional neural networks to identify a region of lesion skin within the acquired clinical dermatoscopic image.