Extraction of Breast Tumor Boundary in Ultrasonic Image Based on Improved Variational Level Set Algorithm / 航天医学与医学工程

ABSTRACT

Objective To propose an extraction algorithm for extraction of breast tumor boundary in its ultrasonic image. Methods A novel Global-Local Energy function was proposed in the variational Level Set formulation for the curve evolution. By enhancing the intensity of local information, the breast tumor boundary was obtained from ultrasonic images accurately. Results Experiments on a synthetic ultrasound image showed that this method was more accurate at segmenting the object from the background than the Chan-Vese algorithm based on the global optimization. Experiments on 103 ultrasonic breast tumor images also showed that the boundaries were extracted accurately with this method without any pre-processing step. Conclusion This proposed method is expected for the fast and accurate boundary extraction of ultrasonic breast tumor boundary from ultrasonic image.