Classification of uterine fibroid from ultrasound images using wavelet transform

Abstract

Uterine myoma and adenomyoma are the most common benign tumors of the uterus. Ultrasound Imaging is the widely used method in the diagnosis of both the disease conditions; however the diagnosis strongly depends on the physician’s expertise and ultrasound system quality. These drawbacks have motivated the development of computer aided applications for the quantitative analysis of ultrasound images to assist the physician in the accurate diagnosis. In this work, statistical texture based features of uterine myoma and adenomyoma of ultrasound images are extracted using wavelet transform and the effectiveness of the selected features are analysed using various classifiers. The energy feature proved to be the best feature in the classification of uterine myoma and adenomyoma with the classification accuracy of about 70%.