A Concatenated Model with Deep Learning Technique for Skin Lesion Detection

Unusual development of skin cells tends to the skin lesion problem. Early diagnosis may help to avoid some types of cancers like melanoma and focal cell carcinoma. The detection and classification of malignant growth in the skin at the beginning stage are challenging tasks for researchers, and medical professionals. For further improvement of the deep learning models, a concatenated model is proposed with a deep learning technique that uses an integration of non-statistical and statistical features. This offered model utilizes the Convolutional neural network (CNN) model to mine non-statistical image features and Multilayer Perceptron (MLP) to process statistical features as handcrafted features. It is shown that the accuracy of the proposed model is improved to 98.1%.