Caracterização tecidual ultrassonográfica renal (CATUS): Análise comparativa entre tons de cinza de imagens renais após padronização de sua aquisição – Proposição de um método diagnóstico auxiliado por computador (CAD)

Abstract

The evaluation of the renal ecodensity through ultrasound remains the same since its advent. As the ultrasound images are obtained in 256 shades of gray and the human being capacity of perceiving it is limited to 32 shades, it is made usage of intelligent computer systems of Computer-Assisted Diagnosis (CAD) kind which assist the expert in the analysis of the intense variability of gray in the images. Following this sense, an investigation was performed using 24 patients, pondering four types of renal sonographic images according to their acquisition and processing, obtained by five specialists and two types of ultrasound equipment. A software prototype was developed whose objective was to determine the gray scale median (GSM) and different intervals of gray tones representing virtual histological tissues, helping the identification of the variability of the renal sonographic image and verifying the possibility of a CAD that helps in a possible diagnosis based on the gray-scale analysis. The results showed that the GSM of renal images of the same patient, obtained by two different types of devices, are statistically different. Four types of images were analyzed, from a totally random one (without any standardization in their acquisition), to a standardized one (fixed the gain, TGC and dynamic range), and these two going through a normalization process already described in the literature and comparing them with the images acquired by three different physicians. It was found that the random image, according to the Wilcoxon- Mann-Whitney test, differs statistically among the professionals, even passing through the normalization process. The standardized image, after passing through the normalization process, presents equivalent results among the three physicians, not having statistical difference. It was verified that the gray scale analysis, to be used in a CAD and compared with different professionals, must be acquired by a same type of ultrasound machine and using an image in which physical aspects are standardized, and after that, passing through a process of standardization.