Review On Skin Cancer Detection Using Combined Decision of Deep Learners

Information science and engineering

Cancer is a fatal illness that develops when body cells grow out of control. Cancer has been dubbed the most significant public health issue due to the high number of deaths it causes each year. Any area of the human body, which may be made up of trillions of cells, might develop cancer. Skin cancer, which arises in the epidermis, is one of the most common types of cancer. In the past, protein sequences and several imaging modalities have been utilised to diagnose skin cancer using machine learning approaches. The disadvantage of machine learning techniques is that they necessitate human-engineered features, which is a very time-consuming and difficult process. By offering the capability of automatic feature extraction, deep learning partially resolved this problem. This study uses the ISIC public dataset to detect skin cancer using convolution-based deep neural networks. Cancer detection is a delicate matter that is prone to mistakes if it is not promptly and precisely identified. Each machine learning model's ability to identify cancer is constrained. It is anticipated that the decision made by the group of learners will be more accurate than that of the individual learners. To make a better choice, the ensemble learning technique takes advantage of the diversity of learners. Therefore, by combining each learner's decision for delicate topics like cancer diagnosis, the prediction accuracy can be improved. This article develops an ensemble of deep learners for skin cancer detection employing learners from VGG, Caps Net, and Res Net. The findings demonstrate that, in terms of sensitivity, accuracy, specificity, F-score, and precision, the combined judgement of deep learners is better than the finding of individual learners. The study's experimental findings offer strong justification for its application in the identification of various diseases.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.