PROJECT TITLE :

Phase Asymmetry Ultrasound Despeckling With Fractional Anisotropic Diffusion and Total Variation

ABSTRACT:

To remove tissue-dependent complex speckle noises from ultrasound images while still keeping various edge properties, we present a new ultrasonic speckle filtering approach. Edge significance is determined by the phase congruence-based edge significance measure called phase asymmetry (PAS), which accepts 0 in non-edge smooth regions and 1 at the idea step edge, as well as intermediate values at slowly variable ramp edges. Despeckling performance with ramp edge preservation and reduced staircase effect can be achieved by using the PAS metric in designing weighting coefficients to maintain a balance between fractional-order anisotropic diffusion and total variation (TV) filters, and we propose a new fractional TV framework for this purpose. A new fractional-order diffusion coefficient for diffusion filtering is then designed based on the PAS metric. It's also worth noting that the PAS metric-based adaptive fractional order diffusion filters have been developed in place of the more traditional fixed fractional-order diffusion filters. To obtain the final denoised image, the gradient descent method minimises the proposed fractional TV model. For both visual evaluation and quantitative indices, the proposed technique beats other state-of-the-art ultrasound despeckling filters in terms of both speckle reduction and retention of features. There are three best scores on feature similarity indexes (0.867, 0.844) and the greatest breast ultrasound segmentation accuracy in terms of mean and median dice similarity coefficients (96.25% and 96.25%).