Big data classification is a challenging assignment for the knowledge processing of relevant information in various domains. Many advanced machine learning (ML) and metaheuristic algorithms have been introduced to pro...
详细信息
Most cancers are not fatal if diagnosed early and treated with medication. But if cancer is not diagnosed early and people do not start treatment, their lives may be in danger. Oral cancer is no exception to this rule...
详细信息
Most cancers are not fatal if diagnosed early and treated with medication. But if cancer is not diagnosed early and people do not start treatment, their lives may be in danger. Oral cancer is no exception to this rule, and as mentioned, the lives of people can be saved if diagnosed in time. Recently, using deep learning in the early diagnosis of different types of diseases has increased. The results of using such methods show good outcomes for the accurate diagnosis of diseases. Therefore, in this paper, a new methodology, using deep-learning, based on a metaheuristic approach has been designed to provide an accurate cancer diagnosis tool. In this study, we first use three preprocessing techniques, including Gamma correction, noise reduction, and data augmentation for enhancing the quality of the raw images and increasing their numbers to provide enough data during con-volutional neural network training. The network weights are then optimally selected by ISSA (an upgraded type of squirrelsearchalgorithm) to provide higher accuracy. The designed method is then applied to a standard benchmark database, termed "Oral Cancer (Lips and Tongue) images dataset" and a comparison between its results and some techniques in the literature indicates the proposed method's higher operation in diagnosing oral cancer.
暂无评论