An evaluation of a number of techniques for decreasing the computational complexity of texture feature extraction through an application to ultrasonic image analysis

Texture feature extraction has been proved to be a fundamental process in medical image analysis. Therefore, the reduction of its computational time and storage requirements should be an aim of continuous research. This paper investigates a number of techniques in the direction of the above goal. They are all based on the space efficient co-occurrence trees in the spatial grey level dependence method (SGLDM). The techniques are applied to a number of ultrasonic images, giving lower bound results on their time performance. A comparison with the co-occurrence matrix approach is performed. Finally, their usefulness in a real clinical application is discussed.