Fast Non-local Means Algorithm Research Articles

Quantitative analysis of fast non-local means algorithm smoothing factor impact on lung segmentation accuracy in computed tomography images

In computed tomography (CT), noise inevitably occurs during image acquisition and degrades segmentation accuracy. Thus, in this study, we compared and evaluated quantitative factors to derive the optimized smoothing factor of the fast non-local means (FNLM) algorithm, which can improve the accuracy of lung segmentation. Gaussian noise with a standard deviation of 0.05 was added to the lung phantom image and clinically acquired lung image, respectively, and the smoothing factor of the FNLM algorithm was modeled and applied from 0.03 to 0.02 in 0.01 intervals. The segmentation algorithm is modeled based on a region-growing approach. For evaluating segmentation performance, the root mean square error (RMSE), peak signal-to-noise ratio (PSNR) and dice score (DSC) were used. The quantitative evaluation demonstrated that the RMSE, PSNR and DSC exhibited the greatest improvement when the smoothing factor was set as 0.10 for both the phantom and clinically acquired lung images. In the lung phantom images, the RMSE, PSNR, and DSC displayed improvements of 8.40, 1.33, and 1.33 times compared to the noisy images. In the clinically acquired lung images, the RMSE, PSNR, and DSC exhibited enhancements of 9.74, 1.35, 1.37 times respectively, compared to the noisy images. We confirmed that lung segmentation performance was most effectively improved when the smoothing factor of the FNLM algorithm was set to 0.10. Moreover, we demonstrated that adjusting the appropriate smoothing factor in CT images can improve lung segmentation performance.

Optimization of smoothing factor for fast non-local means algorithm in high pitch based low-dose computed tomography images with tin-filter

In this study, we applied the fast non-local means (FNLM) algorithm to high-pitch low-dose computed tomography (CT) images using a tin-filter. This study was conducted via a simulation study using a Shepp-Logan phantom and an experimental approach using American association of physicists in medicine (AAPM) phantom images. To optimize the smoothing factor of the FNLM algorithm, the values were adjusted from 0.01 to 0.05 at intervals of 0.001 and applied to the CT image. The smoothing factor of the optimal FNLM algorithm, derived through various quantitative evaluations, was 0.03. According to the results, when the optimized FNLM algorithm was applied to the acquired CT image with a low-dose protocol based on a high pitch condition, the estimated coefficient of variation, contrast-to-noise ratio, and signal-to-noise ratio were improved by 8.7, 6.7, and 8.0 times, respectively, compared to the general CT scanning condition (none). In addition, compared with the noise filtering methods mainly used in the past, both simulation and real experimental results proved that the application efficiency of the FNLM algorithm was high in the obtained low-dose-based CT images. In summary, the applicability of the FNLM algorithm in CT images obtained under low-dose conditions with high pitch using a tin-filter was demonstrated, and the derived research results are expected to be actively utilized in the diagnostic medical field to help reduce exposure doses and improve image quality.

An Artificial Intelligence Multiprocessing Scheme for the Diagnosis of Osteosarcoma MRI Images.

Osteosarcoma is the most common malignant osteosarcoma, and most developing countries face great challenges in the diagnosis due to the lack of medical resources. Magnetic resonance imaging (MRI) has always been an important tool for the detection of osteosarcoma, but it is a time-consuming and labor-intensive task for doctors to manually identify MRI images. It is highly subjective and prone to misdiagnosis. Existing computer-aided diagnosis methods of osteosarcoma MRI images focus only on accuracy, ignoring the lack of computing resources in developing countries. In addition, the large amount of redundant and noisy data generated during imaging should also be considered. To alleviate the inefficiency of osteosarcoma diagnosis faced by developing countries, this paper proposed an artificial intelligence multiprocessing scheme for pre-screening, noise reduction, and segmentation of osteosarcoma MRI images. For pre-screening, we propose the Slide Block Filter to remove useless images. Next, we introduced a fast non-local means algorithm using integral images to denoise noisy images. We then segmented the filtered and denoised MRI images using a U-shaped network (ETUNet) embedded with a transformer layer, which enhances the functionality and robustness of the traditional U-shaped architecture. Finally, we further optimized the segmented tumor boundaries using conditional random fields. This paper conducted experiments on more than 70,000 MRI images of osteosarcoma from three hospitals in China. The experimental results show that our proposed methods have good results and better performance in pre-screening, noise reduction, and segmentation.

Application of Fast Non-Local Means Algorithm for Noise Reduction Using Separable Color Channels in Light Microscopy Images.

The purpose of this study is to evaluate the various control parameters of a modeled fast non-local means (FNLM) noise reduction algorithm which can separate color channels in light microscopy (LM) images. To achieve this objective, the tendency of image characteristics with changes in parameters, such as smoothing factors and kernel and search window sizes for the FNLM algorithm, was analyzed. To quantitatively assess image characteristics, the coefficient of variation (COV), blind/referenceless image spatial quality evaluator (BRISQUE), and natural image quality evaluator (NIQE) were employed. When high smoothing factors and large search window sizes were applied, excellent COV and unsatisfactory BRISQUE and NIQE results were obtained. In addition, all three evaluation parameters improved as the kernel size increased. However, the kernel and search window sizes of the FNLM algorithm were shown to be dependent on the image processing time (time resolution). In conclusion, this work has demonstrated that the FNLM algorithm can effectively reduce noise in LM images, and parameter optimization is important to achieve the algorithm’s appropriate application.

Quantitative evaluation of the image quality using the fast nonlocal means denoising approach in diffusion-weighted magnetic resonance imaging with high b-value

Brain images acquired using the diffusion-weighted imaging (DWI) method indicate that the diagnostic efficiency of infarction improves and the noise increases as the b-value increases. In this study, we designed a fast nonlocal means (FNLM) noise reduction algorithm and evaluated its effectiveness for de-noising brain images with high b-values. The designed algorithm uses an approach that measures the similarity of local parts in an image, calculates weights based on the result, and uses the principle of reducing processing time using a simplification of the calculation. To demonstrate the effectiveness of the algorithm, we compared the qualities of the images obtained using FNLM with those obtained using previously developed algorithms with noise reduction performance and no-reference image-quality assessment parameters. The results of applying the FNLM noise reduction algorithm to DWI images obtained at high b-values indicated superior quantitative characteristics. In particular, the signal-to-noise ratio, coefficient of variation, and blind/referenceless image spatial quality evaluator (BRISQUE) results using the proposed FNLM algorithm were approximately 1.84, 1.44, and 1.21 times better than those of the noisy image, respectively. In conclusion, our results verified that the FNLM approach achieves higher noise reduction efficiency in diffusion-weighted magnetic resonance imaging.

Experimental study of the fast non-local means noise reduction algorithm using the Hoffman 3D brain phantom in nuclear medicine SPECT image

Gamma imaging of the brain in nuclear medicine usually contains a noise distribution owing to attenuation and scattering effects. This study aimed at improving the quality of the image of the brain acquired by single photon emission computed tomography (SPECT) by applying the fast non-local means (FNLM) algorithm, which an excellent noise reduction technique, using a Hoffman 3D brain phantom filled with 99mTc. Gaussian noise was added to the acquired image using MATLAB and the image quality evaluated using conventional noise reduction filters and the proposed FNLM algorithm. The contrast to noise ratio (CNR), peak signal to noise ratio, and root mean square error were used for quantitative evaluation. Comparing the image qualities obtained using various filters and algorithms, the CNR results of the images with Gaussian and median filters, and FNLM algorithm were found to surpass those of the noise image by 1.06, 1.22, and 1.41 times, respectively. In addition, similarity analysis indicated that the image evaluated using the FNLM algorithm showed excellent quality. Thus, our results demonstrated that the FNLM algorithm provided the greatest improvement in image quality based on quantitative evaluation results in the SPECT system.

Feasibility of fast non-local means noise reduction algorithm in magnetic resonance imaging using 1.5 and 3.0 T with diffusion-weighted image technique

Magnetic resonance imaging (MRI) has many advantages and has developed various pulse sequences. In particular, the diffusion weighted image (DWI) technique is widely used because it can acquire images quickly during examination of stroke, through a proper adjustment of the diffusion-weighted gradient b-value. However, a setting with inappropriate b-value causes loss of image signal that increases the influence of noise. Therefore, in this study, we quantitatively evaluated image quality after applying a variety of algorithms to the image acquired by changing the b-value and the main magnetic field in the MRI device. To acquire the image, the phantom was self-produced with an acrylic panel and chicken breast. Wiener filter, total variation (TV), and our proposed fast non-local means (FNLM) noise reduction algorithms were applied to the image. Consequently, the signal intensity at a 3.0 T magnetic field increased by a factor 4.8 compared to a 1.5 T magnetic field. Moreover, the signal-to-noise ratio and contrast-to-noise ratio were highest with the FNLM algorithm, and the values increased by factors of 9.5 and 9.9 with a 1.5 T magnetic field and by factors of 9.9 and 5.0 with a 3.0 T magnetic field compared to the noise image, respectively. The result of time resolution, the Wiener filter appeared the finest value, but had no significant difference compared to FNLM algorithm. In conclusion, our results confirmed that the proposed FNLM noise reduction algorithm can acquire both improved image quality and high processing time in MRI imaging with the DWI technique.

Fast non-local means noise reduction algorithm with acceleration function for improvement of image quality in gamma camera system: A phantom study

Abstract Gamma-ray images generally suffer from a lot of noise because of low photon detection in the gamma camera system. The purpose of this study is to improve the image quality in gamma-ray images using a gamma camera system with a fast nonlocal means (FNLM) noise reduction algorithm with an acceleration function. The designed FNLM algorithm is based on local region considerations, including the Euclidean distance in the gamma-ray image and use of the encoded information. To evaluate the noise characteristics, the normalized noise power spectrum (NNPS), contrast-to-noise ratio (CNR), and coefficient of variation (COV) were used. According to the NNPS result, the lowest values can be obtained using the FNLM noise reduction algorithm. In addition, when the conventional methods and the FNLM noise reduction algorithm were compared, the average CNR and COV using the proposed algorithm were approximately 2.23 and 7.95 times better than those of the noisy image, respectively. In particular, the image-processing time of the FNLM noise reduction algorithm can achieve the fastest time compared with conventional noise reduction methods. The results of the image qualities related to noise characteristics demonstrated the superiority of the proposed FNLM noise reduction algorithm in a gamma camera system.

Classification of breast masses in ultrasound images using self-adaptive differential evolution extreme learning machine and rough set feature selection.

A method using rough set feature selection and extreme learning machine (ELM) whose learning strategy and hidden node parameters are optimized by self-adaptive differential evolution (SaDE) algorithm for classification of breast masses is investigated. A pathologically proven database of 140 breast ultrasound images, including 80 benign and 60 malignant, is used for this study. A fast nonlocal means algorithm is applied for speckle noise removal, and multiresolution analysis of undecimated discrete wavelet transform is used for accurate segmentation of breast lesions. A total of 34 features, including 29 textural and five morphological, are applied to a [Formula: see text]-fold cross-validation scheme, in which more relevant features are selected by quick-reduct algorithm, and the breast masses are discriminated into benign or malignant using SaDE-ELM classifier. The diagnosis accuracy of the system is assessed using parameters, such as accuracy (Ac), sensitivity (Se), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV), Matthew's correlation coefficient (MCC), and area ([Formula: see text]) under receiver operating characteristics curve. The performance of the proposed system is also compared with other classifiers, such as support vector machine and ELM. The results indicated that the proposed SaDE algorithm has superior performance with [Formula: see text], [Formula: see text], [Formula: see text], [Formula: see text], [Formula: see text], [Formula: see text], and [Formula: see text] compared to other classifiers.

Fast non-local means algorithm based on Krawtchouk moments

Non-local means (NLM) method is a state-of-the-art denoising algorithm, which replaces each pixel with a weighted average of all the pixels in the image. However, the huge computational complexity makes it impractical for real applications. Thus, a fast non-local means algorithm based on Krawtchouk moments is proposed to improve the denoising performance and reduce the computing time. Krawtchouk moments of each image patch are calculated and used in the subsequent similarity measure in order to perform a weighted averaging. Instead of computing the Euclidean distance of two image patches, the similarity measure is obtained by low-order Krawtchouk moments, which can reduce a lot of computational complexity. Since Krawtchouk moments can extract local features and have a good antinoise ability, they can classify the useful information out of noise and provide an accurate similarity measure. Detailed experiments demonstrate that the proposed method outperforms the original NLM method and other moment-based methods according to a comprehensive consideration on subjective visual quality, method noise, peak signal to noise ratio (PSNR), structural similarity (SSIM) index and computing time. Most importantly, the proposed method is around 35 times faster than the original NLM method.

Learning multiple linear mappings for efficient single image super-resolution.

Example learning-based superresolution (SR) algorithms show promise for restoring a high-resolution (HR) image from a single low-resolution (LR) input. The most popular approaches, however, are either time- or space-intensive, which limits their practical applications in many resource-limited settings. In this paper, we propose a novel computationally efficient single image SR method that learns multiple linear mappings (MLM) to directly transform LR feature subspaces into HR subspaces. In particular, we first partition the large nonlinear feature space of LR images into a cluster of linear subspaces. Multiple LR subdictionaries are then learned, followed by inferring the corresponding HR subdictionaries based on the assumption that the LR-HR features share the same representation coefficients. We establish MLM from the input LR features to the desired HR outputs in order to achieve fast yet stable SR recovery. Furthermore, in order to suppress displeasing artifacts generated by the MLM-based method, we apply a fast nonlocal means algorithm to construct a simple yet effective similarity-based regularization term for SR enhancement. Experimental results indicate that our approach is both quantitatively and qualitatively superior to other application-oriented SR methods, while maintaining relatively low time and space complexity.

A Robust and Fast Non-Local Means Algorithm for Image Denoising

In the paper, we propose a robust and fast image denoising method. The approach integrates both 'Non-Local means algorithm and Laplacian Pyramid. Given an image to be denoised, we first decompose it into Laplacian pyramid. Exploiting the redundancy property of Laplacian pyramid, we then perform non-local means on every level image of Laplacian pyramid. Essentially, we use the similarity of image features in Laplacian pyramid to act as weight to denoise image. Since the features extracted in Laplacian pyramid are localized in spatial position and scale, they are much more able to describe image, and computing the similarity between them is more reasonable and more robust. Also, based on the efficient Summed Square Image (SSI) scheme and Fast Fourier Transform (FFT), we present an accelerating algorithm to break the bottleneck of non-local means algorithm -similarity computation of compare windows. After speedup, our algorithm is fifty times faster than original non-local means algorithm. Experiments demonstrated the effectiveness of our algorithm.