Research of textural features for the diagnostics of nephrological diseases using ultrasound images
Andrey Gaidel, S. N. Larionova, Alexander G. Khramov · Вестник Самарского государственного аэрокосмического университета им. академика С.П. Королёва (национального исследовательского университета) · 2014
A method of automated diagnostics of kidney diseases using ultrasound images is proposed. The efficiency of different groups of information features of such images for the task of recognition is analyzed. According to the data of a number of experiments on real data the group of two Haralicks features showed the best result. The estimation of probability of wrong recognition for that group was 0.06. Spectral correlation features also demonstated high efficiency, the estimation of the latter for them being 0.10.