Classification of normal and medical renal disease using B-mode ultrasound images
In the present work, a computer-aided diagnostic (CAD) system is proposed for the classification of normal and medical renal disease (MRD) using B-mode ultrasound images. Nineteen ultrasoundimages consisting of 11 normal and 8 MRD images are used. Regions of interest (ROIs) are marked by the radiologist in the parenchyma region of kidney. Texture features have been extracted by different methods including first order statistics, gradient, moment invariant, GLCM, RLM and Laws features.
The optimal feature sets are obtained using DEFS. Exhaustive experiments are carried out with different feature sets. An average classification accuracy and standard deviation of 85.8±3.1 has been obtained using gradient and GLCM features together with SVM classifier. The promising results show that the proposed CAD system design could assist the radiologists for the diagnosis of medical renal disease.