A reproducible method to transcranial B - MODE ultrasound images analysis based on echogenicity evaluation in selectable ROI

The presented paper demonstrates how to detect pathological issues in transcranial B-MODE ultrasound images. We developed an algorithm based on binary thresholding with subsequential computing of area inside ROI which represents an area in which we detect the issue. We work with a collection of images acquisted from 3 different ultrasound machines. We detect echogenic area in substantia nigra which is primary feature to Morbus Parkinson and also raphe nucleus echogenicity to detection of other neurological diseases. All achieved results were verified by an erudite neurosonologist and based on statistics such as correlation and kappa analysis. Average correlation between observers r > 0.88, level of agreement kappa > 0.82. So, we proved that developed algorithm is highly reproducible and also could be used for different cases, not only in neurology due to principle of B-MODE imaging. The algorithm has been implemented in MATLAB with Image Processing Toolbox. Achieved results will verified for 3 different ultrasound machines Keywords— B-MODE, Parkinson, echogenicity, hyperechogenic SN, substantia nigra parkinson