Automatic initialization for active contour model in breast cancer detection utilizing the integration of ultrasonography image modalities

Breast cancer is the most common category of cancers in woman around the world. Ultrasonography imaging modalities (US) are highly recommended for breast cancer examining due to their sensitivity, specificity, cost-effective, accessibility, portability, comfort, as well as non-invasive tool. In addition, an integration of a conventional US and its adjunct modalities which is Power Doppler has been proved for reducing fault-positive in breast cancer diagnosis. The combination of conventional US and Power Doppler also increases sensitivity and specificity in a woman who has dense breast. Hence, the integration of US-based imaging modalities would also benefit in medical image processing such as a breast cancer segmentation. An affective method for medical image segmentation, active contour model has been widely used for decades. A crucial stage that affect the performance of the model is initialization. This paper proposed a novel method, an automatic initialization for parametric active contour model. It is significant difference from previous methods that focus on vector flow analysis. Our method estimates an appropriate initial contour by utilizing the integration of conventional US and Power Doppler. Examples and comparison with a state-of-the-art method for automatic initialization are demonstrated showing better performance for initialization.