Pattern Recognition Based Skin Lesion Stage Analysis Using IoT

Skin diseases are considered the emerging cause of high mortality throughout the world. To reduce the rate of death, that arise due to skin is minimized with early detection and cure. The lesions in the skin need to be evaluated significantly to increase the detection rate in the early stage. NowadaysComputer-Aided Diagnosis (CAD) is considered as the effective technique to detect lesions in the skin with the use of pattern recognition technique. Through CAD system skin lesions are classified into different classes. In other hand, medical environment uses the Internet of Things (IoT) for the monitoring and detection of diseases in patients. Hence, this paper constructed a framework of Dragon Pattern Optimization Stacked Classifier (DPOSC) automated model for the early detection of skin lesions. The DPOSC automated model is interconnected with the IoT devices implanted in the patient body for the classification of lesions as malignant, benign and normal in the early stage. The DPOSC model perform the image pre-processing with PCA (Principal component analysis) followed by the segmentation. The pattern recognition approach is developed with estimation of the features in the skin images. The technique uses quantitative analysis with the coding process to extract the pattern in the skin lesion and utilized for the automated detection system. The modifier dragon optimization model is implemented for the extraction and optimization of features in the skin lesion images. Upon the extracted features classification is performed with the stacked classification model. The DPOSC model effectively detect the lesions in the skin images and linked with the IoT environment for the detection of skin cancer in the early diagnosis. The developed automated integrated with IoT is emerging innovative technique for the detection of cancer in early stages. The accuracy of the improved DPOSCis 0.5% superior to SVM and NN, 2.4% more improved than GA-NN, PSO-NN and DA-NN. The precision of the developed DPOSC is 11.6% superior to SVM and NN, 5.4% higher than GA-NN, 1.4% larger than PSO-NN and 1.6% superior to DA-NN.