Automated technique for carotid plaque characterisation and classification using RDWT in ultrasound images

In this manuscript, a novel method of rational-dilation wavelet transform (RDWT) is proposed for carotid plaque characterisation and classification in Ultrasound images. RDWT is mainly utilised for image acquisition, pre-processing, feature extraction and ensemble classification in automated plaque classification. Here, the transition bands are constructed from the transition function. The statistical features, viz mean, standard deviation, skewness, Renyi entropy, energy are extracted from the sub-bands of RDWT. The Salp Swarm Algorithm (SSA) is mainly used for selecting the optimum features. In this for selecting optimum features using SSA algorithm two conditions are satisfied such as, in the first approach, mean, standard deviation, skewness are selected and then utilised for converting the continual version of salp swarm algorithm to binary. Subsequently, the crossover operator is utilised to select the Renyi entropy, energy features including transfer functions for replacing the average operator and enhancing the characteristics of research method. Plaque Classification uses K Nearest Neighbour (KNN), Probabilistic Neural Network (PNN) and Support Vector Machine (SVM) classifiers. Experimental outcomes show the efficiency of the proposed method depending on accuracy, specificity and sensitivity. The proposed method attains accuracy of 93%, sensitivity of 90% and specificity of 94% when likened to the existing techniques, such as KNN, PNN and SVM.