Development of computer aided diagnosis system (CAD) for detection of kidney ultrasound images
Nurul Aimi Shaharuddin, Wan Mahani Hafizah Wan Mahmud · 2017
Ultrasound machines are undergoing intense research and rapid development in the world of medical imaging. Ultrasound machines with different technical specifications produce different qualities of ultrasound images. As the analysis and interpretation of ultrasound machines depend on the expertise of sonographers and physicians, human errors and mistakes may occur. The objective of this research is to develop a computer-aided diagnosis (CAD) based on kidney ultrasound images, which can be used regardless of the qualities of ultrasound machines. The software was developed using MATLAB. Meanwhile, the images were enhanced by histogram equalization, filtered by Wiener filtering, segmented by users, and classified using the artificial neural network (ANN). Performance evaluation of the system showed that the percentage of correctly-identified kidney and non-kidney images from ultrasound machines A, B, C, and D were (94.3%) and (92.2%) respectively. For the ultrasound machines other than A, B, C, and D, the percentage of correctly-identified images were 72.2 % (kidney) and 62.4% (non-kidney).