Multiframe-based Adaptive Despeckling Algorithm for Ultrasound B-mode Imaging with Superior Edge and Texture

TL;DR

This paper introduces a multiframe adaptive despeckling algorithm for ultrasound B-mode imaging that enhances image quality by effectively removing speckle noise while preserving edges and textures, outperforming traditional methods.

Contribution

The paper presents a novel multiframe-based adaptive despeckling algorithm based on a signal generation model, improving ultrasound image quality without artifacts.

Findings

Achieved 8.55-15.91 dB SNR and PSNR improvements in simulation data.

Attained 13.24-32.85 NIQE and BRISQUE improvements in in-vivo data.

Visual results show superior texture and resolution compared to commercial scanners.

Abstract

Removing speckle noise from medical ultrasound images while preserving image features without introducing artifact and distortion is a major challenge in ultrasound image restoration. In this paper, we propose a multiframe-based adaptive despeckling (MADS) algorithm to reconstruct a high-resolution B-mode image from raw radio-frequency (RF) data that is based on a multiple input single output (MISO) model. As a prior step to despeckling, the speckle pattern in each frame is estimated using a novel multiframe-based adaptive approach for ultrasonic speckle noise estimation (MSNE) based on a single input multiple output (SIMO) modeling of consecutive deconvolved ultrasound image frames. The elegance of the proposed despeckling algorithm is that it addresses the despeckling problem by completely following the signal generation model unlike conventional ad-hoc smoothening or filtering based approaches, and therefore, it is likely to maximally preserve the image features. As deconvolution is a necessary pre-processing step to despeckling, we describe here a 2-D extension of the SIMO model-based 1-D deconvolution method. Finally, a complete framework for the generation of high-resolution ultrasound B-mode image has been also established in this paper. The results show 8.55-15.91 dB, 8.24-14.94 dB improvement in terms of SNR and PSNR, respectively, for simulation data and 2.22-3.17, 13.24-32.85 improvement in terms of NIQE and BRISQUE, respectively, for in-vivo data compared to the traditional despeckling algorithms. Visual comparison shows superior texture, resolution, details of B-mode images offered by our method compared to those by a commercial scanner, and hence, it may significantly improve the diagnostic quality of ultrasound images.