The Melanoma Skin Cancer Detection Using Convolutional Neural Network

Abstract

The most common cause of death for persons in the modern world is melanoma skin cancer. Melanoma is an aggressive kind of skin cancer that typically develops on parts of the body exposed to sunlight, UV radiation, dust, pollution, and microbes. A study found that 79% of humans who do have melanoma skin cancer in its early stages are unaware of having it. When it is finally identified, it may have progressed further into the skin and may have impacted other parts of the body, making treatment extremely difficult and it also makes the survival rate for humans very low. As a result, melanoma skin cancer kills the majority of its victims. If melanoma is identified or recognized in the early stages, then it could have been cured easily and it also has a bigger survival rate for humans. There is an automated system that has been designed with the compilation of data sets with a variety of diagnoses. This automated system will help to detect in early stages of melanoma. Convolutional neural network technology was initially employed in our automated system to classify data. But as a result, our accuracy was poor. Then, we used support vector machines (SVMs) to analyze the entire dataset after categorizing and segmenting the data into zones, using Tensor Flow libraries to implement the entire model. After using the model, we achieve a 96% accuracy rate.