Ultrasound B-MODE image processing as a MATLAB software tool and as an experimental solution on ARM platform

The paper is focused on image processing of B-images from diagnostic ultrasound, their processing and analysis with own developed algorithm based on binary thresholding which is useful for images in grayscale such as B-images. The presented algorithm has been created as MATLAB-based application and verified its function for 2 different structures displayed in midbrain - substantia nigra and raphe nucleus. The processing and analysis is based on measuring of their echogenicity level from the principle of B-MODE. The results also was verified and rated by an experienced sonographer. The second part is focused on implementation of the algorithm on embedded system ARM Cortex-M4 which allows to create an independent hardware computing unit connectable to an ultrasound machine. Sonographers could evaluate pathological issues directly with no loading images to a computer. We selected ARM platform due to its performance, low consumption and hardware scalability. Communication from MATLAB is realized via serial port RS232 and loading the source code into microcontroller of ARM CPU. All reached results have been investigated and verified by the experienced sonographer with statistical analysis of reliability and reproducibility (correlation, kappa and ROC). This analysis verified successful reproducibility of the algorithm in clinical practice to early diagnostic.