Neural Network Diagnosis System for 3-Dimensional Ultrasonography with Gabor Filter Aided Speckle Decorrelation

In this paper, a neural network diagnosis system for 3-dimensional ultrasonography using Gabor filters to eliminate the speckle noise is introduced. Since the importance of the relationship between surface features and internal architecture of the breast tumor, we applied our system using three dimensional (3D) inter-pixel correlation instead of 2D features. Furthermore, Gabor filters provide a multi-resolution representation of texture, which increasing the capabilities of ultrasound technology in the differential diagnosis of solid breast tumors. Our experiments show that the performance of the proposed diagnostic method is effective.