Improving Skin Cancer Classification Based on Features Fusion and Selection
| Autor: | Youssef Filali, My Abdelouahed Sabri, Abdellah Aarab |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
integumentary system
Computer science business.industry Melanoma ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Image processing Feature selection Pattern recognition Texture (music) medicine.disease ComputingMethodologies_PATTERNRECOGNITION Region of interest Component (UML) medicine Segmentation Artificial intelligence Skin cancer business ComputingMethodologies_COMPUTERGRAPHICS |
| Zdroj: | Embedded Systems and Artificial Intelligence ISBN: 9789811509469 |
| DOI: | 10.1007/978-981-15-0947-6_36 |
| Popis: | Recently, skin cancer has been rapidly increasing in terms of the number of melanoma cases due to the skin exposure to the sun. Melanoma is the deadliest skin cancer in the world. It is necessary to use a computer-aided diagnostic to help and facilitate the early detection of the skin cancer. In this paper, the proposed approach uses a fusion of shape, texture and color features that contains the lesion to classify the skin cancer. An image decomposition using the multi-scale is used, which gives two components: object and texture components. The object component will be used in the segmentation to identify the region of interest. The features are then extracted from the texture component, the shape of the lesion and color that contain the lesion. After combining all the features, a feature selection is moderate to keep only the best one. The classification is performed using the support vector machine classifier to classify skin cancer. The accuracy of our proposed approach is 92%, showing the effectiveness of our system. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|