Classification of renal diseases using first order and higher order statistics
Komal Sharma, Jitendra Virmani · International Conference on Computing for Sustainable Global Development · 2016
In the present work, a computer assisted classification system has been proposed for classification of renal ultrasound images into (a) normal, (b) medical renal disease (MRD) and (c) cyst classes. The work has been carried out on 35(11 normal, 8 MRD and 16 cyst) renal ultrasoundimages. In case of normal and MRD image classes the regions of interest have been cropped from the cortex region of the kidney and from regions inside renal cyst for cyst image class. The performance of (a) First order statistical texture descriptors, (b) Higher order gray level run length matrix statistical texture features have been extensively evaluated by using SVM classifier. The result of the study indicates that concatenation of first order statistics and grey level run length statistics yield the maximum overall classification accuracy of 75.3% with individual classification accuracy of 68.3%, 74.5% and 95.6% for normal, MRD and cyst classes respectively.