Computer-aided diagnosis of carotid atherosclerosis using laws' texture features and a hybrid trained neural network

Objective diagnosis of carotid atherosclerosis and classification into symptomatic or asymptomatic is crucial in planning optimal treatment of atheromatous plaque. The Computer-Aided Diagnostic (CAD) system described in this paper can analyze B-mode ultrasound images of the carotid artery and classify them into Symptomatic . (S) or Asymptomatic (A). Images from 54 S and 54 A plaques were fed to the CAD system, which consists of three modules: the feature extraction module, where texture features are estimated based on Laws' texture energy, the dimensionality reduction module, where the number of features is reduced using ANOVA statistics, and the classifier module with a Neural Network (NN) trained via a novel hybrid method in order to recognize the type of atheromatous plaques. The hybrid training method uses Genetic Algorithms (GA's) to locate a starting point close to the optimal solution, and then the back-propagation (BP) algorithm with adaptive learning rate and momentum to refine the NN configuration with local search. The hybrid method is able to select the most robust features, to adjust automatically the NN architecture, and to optimize the classification performance. The proposed CAD system has achieved a total classification performance of 99%.