A reproducible application to B-MODE transcranial ultrasound images based on echogenicity evaluation analysis in selectable area of interest

The goal of this presented paper is to show and explain how to measure echogenicity level in B-MODE ultrasound images. We present a method which we use in practice in our developed application in MATLAB. The application is usable in medical practice to echogenicity analysis in B-images. The core of the application is based on echogenicity level analysis in selectable area of interest which can be select manually or by semi-automatic algorithm. We use the application in case of transcranial B-MODE images but it is usable in another cases in which are used B-images. The application is based on measuring of echogenicity level by binary tresholding algorithm and pattern recognition inside area of interest and subsequently is computed real echogenic area for all levels of intensity from 0 to 255 as 8bit depth. The results were verified by an experienced neurosonologist and also by independent statistical analysis by 2 independent observers. The method was proved for detection pathology in substantia nigra to parkinsonism, for raphe nuclei and also to monitoring of iron in post-mortem brains. The results proved that the application is reliable and helpful in medical practice and is more accurate then visual diagnostics of an experienced neurosonologist. We tested correlation analysis, ROC and kappa analysis to objective evaluation of the reached results.