GPU-Based Block-Wise Nonlocal Means Denoising for 3D Ultrasound Images

Table 2

Time comparisons of CPU single-thread, CPU multithread and GPU of ultrasound carotid artery denoising.

Similarity window	Stp_S	Processing time (s)
		CPU single-thread	CPU multithread	GPU
3×3×3	2	1865.31	637.23	33.58
5×5×5	3	2039.22	841.77	57.22
7×7×7	4	2076.39	933.12	90.64
9×9×9	5	2129.09	962.11	127.96
11×11×11	6	2165.79	970.43	356.19