Computer-Aided Diagnosis Applied to Ultrasound Image of Calculi Using Gradient Vector Flow Model

Abstract

Medical image segmentation makes use of deformation models for its analysis. Ultrasound (US) imaging allows faster, inexpensive and accurate procedures due to its real capabilities. The accurate detection of objects from US images plays an important rule in many applications. In the present paper, computer aided diagnosis applied to Ultrasound images of calculi is developed and its objective is to provide a tool for physician to diagnose the calculi from the ultrasound test. The developed system consists of three stages of image analysis. The first stage is based on segmentation using an active contour in order to segment the calculi from the image. In the second stage, from the segmented calculi, geometrical and texture features have been extracted. Finally, Artificial Neural Network (ANN) has been designed and extracted features from previous stage are fed to neural network to classify the calculi. The proposed method has been implemented and experimentally evaluated on various ultrasound calculi images acquired from different patients. The classification of calculi is based on size, small, medium and large. From the size of calculi, the Physician can decide whether surgery or extra corporal shock wave lithography (ESWL) to be performed or using medicine, it can be dissolved.