Melanoma Skin Cancer Detection by using Ensemble Model

Abstract

Skin cancer is the abnormal development of skin cells that occurs most commonly on sun exposed skin. This type of cancer can also develop on parts of your skin that aren't often exposed to the sun. A convolutional neural network is a type of deep neural network that is most typically used to evaluate visual imagery in deep learning. The main goal of this model is to detect skin cancer at an early stage so that it may be treated. This model is designed to distinguish between the seven forms of skin cancer. The major goal of this model is to increase the accuracy of existing models, with a particular focus on melanoma skin cancer detection because it is the worst type of skin cancer and failing to detect it early can result in death. If one notice a change in the skin, it could be a new growth, a scar that refuses to heal, or an uneven mole. There is an ABCDE rule for melanoma detection that helps dermatologists as well as the general public identify it. In this model there are numerous models, including Inception, Dense Net, and others. These are pre-trained models, and by fine-tuning the top layers, as well as all of the layers, to see how accurate they are. Finally, there is a combination of the two best models into an ensemble model to improve the system's accuracy and range even further.