Speckle noise reduction in breast ultrasound image

Speckle noise is an important factor that reduces the resolution and contrast of the ultrasound images. Speckle noise occurs with random fluctuating in medical ultrasound imaging system. Speckle noise occurs especially in soft tissue organs. Properly reflected and partly scattered sound beams form the structure of the usual grainy ultrasound image. However, this reduces the contrast and sharpness of the images. Therefore, difficulties in diagnosis due to poor image quality are realized. Today, the methods used to eliminate the speckle noise in ultrasound images are grouped under four main headings. These methods are the filter-based methods, wavelet-based methods, region growing based methods and model-based method, respectively. In this thesis, an approach is proposed for speckle reduction using 2-b homogeneity and directional average filters. We have conducted experiments on numerous clinic breast ultrasound images. The experimental results are evaluated using several criterions. The obtained results demonstrate that the proposed approach can reduce the speckle noise effectively without blurring the edges and damaging the textual information. Key Words: Speckle Reduction, Homogeneity, Ultrasound İmage, Model-Based Method