Comparison Feed Forward Back Propagation Networks (FFBPNs) with Support Vector Machine (SVM) for Diagnosis Skin Cancer Based on Images

Skin cancer is a type of malignancy responsible for 70 percent of overall skin cancer-related death worldwide. The purpose of the research is to use AI to detect skin cancer of all types more quickly and improve the efficiency of diagnostic radiology.The method used in this paper is an artificial neural network implemented for the detection of skin cancer and the watershed segmentation method for segmentation. The features extracted are shape and Gray-Level Co-Occurrence Matrix. The extracted feature is used for classification. The classifiers are Support Vector Machine and Feedforward Back Propagation applied in a Matlab environment and an image processing technique on a set of photographs that were collected from several websites, including the Kaggle web. The implementation of code for the detection of skin cancer by using data as 100 images 50 no cancer and 50 is cancer, the result shows a successful implementation for the detection of cancer in FFBP classifier a 45 and 2 is bad detection, as well as in SVM classifier 49 with 1 is bad diagnostic. The Conclusion shows SVM classifier provided results for the skin lesions classification produced 98% accuracy and the accuracy of the FFBP of 96 %. The conclusion of this study is helping people with skin cancer undergo a CT scan. The scan is tested using a computer trained to analyze CT scan data.