Özet:

Skin cancer begins with the uncontrolled multiplication of malignant tumors. It is a common type of cancer worldwide. It is difficult to be examined and diagnosed with the naked eye by specialists. Therefore, computer-backed diagnosis systems can help diagnose the doctors. These systems are widely using deep nerve networks, which are a kind of artificial intelligence today. In the study that includes many deep nerve networks, medical images are used as data input. Depending on the network architecture, these systems draw the properties into their own layers. In this study, using the pre-trained deep nerve network VGG16 first obtained properties related to images from the network layers. The size of these properties that contain a high amount of data was later reduced. Thus, the qualifications that will the best success in classification have been obtained. Numeric data obtained using the data enhancement algorithm has been enhanced and a 96% classification accuracy and 100% AUC success has been achieved in the CNN type deep nerve network.

Anahtar Kelimeler: