Ultrasound Image Denoising with Multi-shape Patches Aggregation Based Non-local Means

• Published in: 2011 International Conference on Intelligent Computation and Bio-Medical Instrumentation pp. 35 - 38
• Main Authors: Wanjun Chen, Mingyue Ding, Yalin Miao, Lei Luo
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.12.2011
• Subjects: aggregation; Biomedical imaging; Noise reduction; Non-Local Means; risk estimation; Shape; Signal to noise ratio; Speckle; Ultrasonic imaging; Ultrasound image despeckling
• ISBN: 9781457711527; 1457711524
• DOI: 10.1109/ICBMI.2011.87

The Non-Local Means (NLM) filter uses the redundancy of information in the image to remove noise, this scheme gives some of the best results among other powerful methods such as wavelet based approaches or diffusion techniques. Though simple to implement and efficient in practice, the classical NLM suffers from ringing artifacts around edges when using square patches, due to an abrupt lack of redundancy of the image. This paper presents an extended NLM based on Multi-Shape Patches Aggregation (NLM-MSPA) to overcome this problem, and uses it to remove medical ultrasound images corrupted by multiplicative speckle noise. We have incorporated a preprocessing step to make the speckle noise much closer to the real additive white Gaussian noise, hence more amenable to a denoising algorithm such as NLM-MSPA. Results on real images and artificially speckled images show that the proposed scheme outperforms several classical methods chosen for comparison in its ability to reduce speckle and preserve edge details.