Classification of kidney lesions using Gabor wavelet texture features

In the present work, characterization of primary benign and primary malignant focal kidney lesions has been carried out using Gabor wavelet texture features. The angiomyolipomas (AMLs) i.e., primary benign kidney lesions and renal cell carcinomas (RCCs) i.e., primary malignant kidney lesions have been considered in the present study. The Gabor features have been computed from region of interest (ROIs) extracted from each image. The total dataset consists of 101ROIs (23 AML ROIs and 78 RCC ROIs) that have been extracted from within the lesions (ignoring the renal sinus part). For the classification task, support vector machine (SVM) has been used. It is observed that the proposed system yields an overall classification accuracy of 90.0 %for differential analysis between primary benign and primary malignant kidney lesions.