Skin Cancer Detection Using Deep Learning Technique
Dilli Hemalatha, Kaile Nishi Latha, P. Latha
Abstract
One of the most deadly forms of cancer is skin cancer. Due to its relatively high difficult and expensive diagnosis, as well as human subjectivity and challenges, dermatological diseases are one of the most pressing medical issues of the 21st century. When it comes to lethal diseases like melanoma, early detection is crucial for assessing the likelihood of recovery. Skin cancer is brought on by DNA breaks in skin cells that are not repaired, which results in genetic faults or mutations. Since skin cancer is more treatable in its early stages and has a tendency to gradually spread to other body parts, it is best detected at an early stage. Early detection of skin cancer symptoms is necessary due to the rising incidence of cases, high death rate, and expensive medical treatments. Given the severity of these difficulties, researchers have created a number of early. The automatic classification of melanoma images utilizing machine learning and computing vision methods has been the subject of numerous investigations. The segmentation of the skin lesion and the features identified for the classification approach have a significant impact on the classification performance of traditional machine learning and computer vision detection techniques for skin cancer, despite the fact that these studies yield encouraging results. Lesion characteristics including symmetry, color, size, shape, and others are used to distinguish between benign and malignant skin cancer. Experts believe the use of automated approaches will aid in early diagnosis, particularly when a batch of photos has a variety of diagnosis. In-depth investigation of deep learning techniques for skin cancer early detection is provided in this study. The three stages of our model creation are data collection and augmentation, model construction, and prediction. To improve the structure and increase accuracy to 84%, we used image processing tools in conjunction with the Inception V3 method.