Diagnosis of Renal Calculus Disease in Medical Ultrasound Images

Abstract

Ultrasound imaging is used as the primary imaging modality for diagnosis of renal calculus. Speckle noise and shadows present in ultrasound images makes the identification of kidney stones very complex and challenging. Therefore despeckling of ultrasound images is carried out as a preprocessing step. The preprocessing of kidney ultrasound images consists of denoising using wavelet thresholding method Wavelet decomposition is performed on despeckled images and wavelet energy features are extracted for different wavelet families. Further, these features are used by feed-forward, backpropagation artificial neural network for classification of kidney ultrasound image as renal calculi image or normal image. Experimental results demonstrate that the approach is suitable and effective.