Improvement of Ultrasound Image Quality Using Non-Local Means Noise-Reduction Approach for Precise Quality Control and Accurate Diagnosis of Thyroid Nodules

This study aimed to improve the quality of ultrasound images by modeling an algorithm using a non-local means (NLM) noise-reduction approach to achieve precise quality control and accurate diagnosis of thyroid nodules. An ATS-539 multipurpose phantom was used to scan the dynamic range and gray-scale measurement regions, which are most closely related to the noise level. A convex-type 3.5-MHz frequency probe is used for scanning according to ATS regulations. In addition, ultrasound images of human thyroid nodules were obtained using a linear probe. An algorithm based on the NLM noise-reduction approach was modeled based on the intensity and relative distance of adjacent pixels in the image, and conventional filtering methods for image quality improvement were designed as a comparison group. When the NLM algorithm was applied to the image, the contrast-to-noise ratio and coefficient of variation values improved by 28.62% and 19.54 times, respectively, compared with those of the noisy images. In addition, the image improvement efficiency of the NLM algorithm was superior to that of conventional filtering methods. Finally, the applicability of the NLM algorithm to human thyroid images using a high-frequency linear probe was validated. We demonstrated the efficiency of the proposed algorithm in ultrasound images and the possibility of capturing improved images in the dynamic range and gray-scale region for quality control parameters

Computer vision · Dynamic range · Image (mathematics · Image noise · Image quality · Imaging phantom · Noise (video · Noise reduction · Nuclear medicine · Pixel · Radiology · Reduction (mathematics · Thyroid · Thyroid nodules · Ultrasound · Computer Science · Mathematics · Medicine · Photoacoustic and Ultrasonic Imaging · Radiation Dose and Imaging · Ultrasound and Hyperthermia Applications · Artificial Intelligence