Efficient Speckle Reduction Techniques in Ultrasound Video Quality Enhancement – A Comparative Evaluation
01 Sep 2017-
TL;DR: In the present work seven different filters are listed and experimental work is carried out on LEE and Hybrid Median filter (HMF), the experimental result of PSNR and MSE values are considered for evaluation purpose.
Abstract: Amongst many imaging modalities available now a days, Ultrasound is the most widely used because of the reasons like radiation free, real time, non-invasive and non-ionizing During patient’s care, medical images, videos & other physiological signals, become an integral part of diagnostic and treatment phases Ultrasound is more advantageous and cheaper than other modalities like MRI, CT, PET Speckle noise present in US hampers the quality of images which in turn considerably increases the difficulty in medical visual inspection The diagnostic accuracy will be more if image is less noisy, this necessitates the use of efficient despeckling filter which will take care of edge detection, loss of information and improve visual evaluation MSE assess the image quality, if the value is high then image quality is poor PSNR is another quality metrics, high value indicates the better denoising algorithm In the present work seven different filters are listed and experimental work is carried out on LEE and Hybrid Median filter (HMF) The experimental result of PSNR and MSE values are considered for evaluation purpose
Citations
More filters
01 Feb 2021
TL;DR: A comparative analysis of well established despeckling filters compiled for speckle suppression and feature enhancement shows that feature enhanced Fast Non-Local Means (FNLM) filter outperforms other state-of-art filtering techniques in the aspect of higher PSNR value and preserves structural similarity in terms of SSIM value near to unity.
Abstract: Rheumatic heart disease has a substantial impact on morbidity and mortality for both men and women in developing countries. It is a complication of autoimmune phenomenon known as acute rheumatic fever in response to group A streptococcus bacteria. It often causes damage to valves and lacks its functionality. Initial manifestations of prolapsed valves are evident in echocardiography in the form of valve bulging, commissural fusion and restricted leaflet motion. However, these echocardiogram images are inevitably degraded by multiplicative noise known as speckle noise at the time of image acquisition and transmission. Hence despeckling is of vital importance for enhancing ultrasound image quality. In this paper, a comparative analysis of well established despeckling filters compiled for speckle suppression and feature enhancement has been performed. In addition to this, the performance of these filters is validated using different quantitative metrics. The experimental result shows that feature enhanced Fast Non-Local Means (FNLM) filter outperforms other state-of-art filtering techniques in the aspect of higher PSNR value and preserves structural similarity in terms of SSIM value near to unity. In conclusion, the preprocessing intends to suppress speckle noise and enhancing perceived visual quality that aids for further development of virtual cardiac model.
TL;DR: Support vector machine (SVM) has outperformed other technique on sensitivity and time complexity, hence chosen for abnormality classification in this work and an algorithm has been devised to use combination of RHOOF and SVM for the detection of atrial septal defect (ASD).
Abstract: In the medical field various motion tracking techniques like block matching, optical flow, and histogram of oriented optical flow (HOOF) are being experimented for the abnormality detection. The information furnished by the existing techniques is inadequate for medical diagnosis. This technique has an inherent drawback, as the entire image is considered for motion vector calculation, increasing the time complexity. Also, the motion vectors of unwanted objects are getting accounted during abnormality detection, leading to misidentification / misdiagnosis. In this research, our main objective is to focus more on the region of abnormality by avoiding the unwanted motion vectors from the rest of the portion of the heart, allowing better time complexity. Proposed a region-based HOOF (RHOOF) for blood motion tracking and estimation; after experimentation, it is observed that RHOOF is four times faster than HOOF. The performance of supervised machine learning techniques was evaluated based on accuracy, precision, sensitivity, specificity, and area under the curve. In the medical field more importance is given to the sensitivity than accuracy. Support vector machine (SVM) has outperformed other technique on sensitivity and time complexity, hence chosen for abnormality classification in this work. An algorithm has been devised to use combination of RHOOF and SVM for the detection of atrial septal defect (ASD).
References
More filters
01 Jul 2010
TL;DR: This paper presents several plaque-image analysis methods that have been developed over the past years, including clinical methods for visual classification that have led to standardized methods for image acquisition, method for image segmentation and denoizing, and a summary of emerging trends in 3-D imaging methods and plaque-motion analysis.
Abstract: Noninvasive ultrasound imaging of carotid plaques allows for the development of plaque-image analysis methods associated with the risk of stroke. This paper presents several plaque-image analysis methods that have been developed over the past years. The paper begins with a review of clinical methods for visual classification that have led to standardized methods for image acquisition, describes methods for image segmentation and denoizing, and provides an overview of the several texture-feature extraction and classification methods that have been applied. We provide a summary of emerging trends in 3-D imaging methods and plaque-motion analysis. Finally, we provide a discussion of the emerging trends and future directions in our concluding remarks.
82 citations
TL;DR: A modified version of the NL-means method is presented that incorporates an ultrasound dedicated noise model, as well as a GPU implementation of the algorithm that demonstrates that the proposed method is very efficient in terms of denoising quality and is real-time.
Abstract: Image denoising is the process of removing the noise that perturbs image analysis methods. In some applications like segmentation or registration, denoising is intended to smooth homogeneous areas while preserving the contours. In many applications like video analysis, visual servoing or image-guided surgical interventions, real-time denoising is required. This paper presents a method for real-time denoising of ultrasound images: a modified version of the NL-means method is presented that incorporates an ultrasound dedicated noise model, as well as a GPU implementation of the algorithm. Results demonstrate that the proposed method is very efficient in terms of denoising quality and is real-time.
78 citations
"Efficient Speckle Reduction Techniq..." refers background in this paper
...In paper [25] usage of modified version of NL (Non Local) is showcased for real despeckling of US images....
[...]
TL;DR: It is demonstrated that the proposed FESR method can improve the image quality of ultrasound B-mode imaging by enhancing the visualization of lesion features while effectively suppressing speckle noise.
Abstract: Goal: Effective speckle reduction in ultrasound B-mode imaging is important for enhancing the image quality and improving the accuracy in image analysis and interpretation. In this paper, a new feature-enhanced speckle reduction (FESR) method based on multiscale analysis and feature enhancement filtering is proposed for ultrasound B-mode imaging. In FESR, clinical features (e.g., boundaries and borders of lesions) are selectively emphasized by edge, coherence, and contrast enhancement filtering from fine to coarse scales while simultaneously suppressing speckle development via robust diffusion filtering. In the simulation study, the proposed FESR method showed statistically significant improvements in edge preservation, mean structure similarity, speckle signal-to-noise ratio, and contrast-to-noise ratio (CNR) compared with other speckle reduction methods, e.g., oriented speckle reducing anisotropic diffusion (OSRAD), nonlinear multiscale wavelet diffusion (NMWD), the Laplacian pyramid-based nonlinear diffusion and shock filter (LPNDSF), and the Bayesian nonlocal means filter (OBNLM). Similarly, the FESR method outperformed the OSRAD, NMWD, LPNDSF, and OBNLM methods in terms of CNR, i.e., 10.70 ± 0.06 versus 9.00 ± 0.06, 9.78 ± 0.06, 8.67 ± 0.04, and 9.22 ± 0.06 in the phantom study, respectively. Reconstructed B-mode images that were developed using the five speckle reduction methods were reviewed by three radiologists for evaluation based on each radiologist's diagnostic preferences. All three radiologists showed a significant preference for the abdominal liver images obtained using the FESR methods in terms of conspicuity, margin sharpness, artificiality, and contrast, p <0.0001. For the kidney and thyroid images, the FESR method showed similar improvement over other methods. However, the FESR method did not show statistically significant improvement compared with the OBNLM method in margin sharpness for the kidney and thyroid images. These results demonstrate that the proposed FESR method can improve the image quality of ultrasound B-mode imaging by enhancing the visualization of lesion features while effectively suppressing speckle noise.
53 citations
"Efficient Speckle Reduction Techniq..." refers background in this paper
...Ultrasound echoes gets interfere creating artificially large or small signals because of tissues propagation, attenuation and scattering [1]....
[...]
...Ultrasound incident wavelength is much larger than these echoed signals, creates phase summation or cancellation and such patters is defined as multiplicative rather than additive [1]....
[...]
...Effective noise reduction of Ultrasound image is important for image quality enhancement, accuracy improvement, analysis and interpretation [1]....
[...]
TL;DR: It is suggested that the measurement of the CCA IMT on one side only is enough (and this is in agreement with other studies), as well as automated measurements can be used.
Abstract: The intima-media thickness (IMT) of the common carotid artery (CCA) is an established indicator of cardiovascular disease (CVD). There have been reports about the difference between the left and the right sides of the CCA IMT and their relation with CVD. In this paper, we propose an automated system based on image normalization, speckle reduction filtering, and snakes segmentation, for segmenting the CCA, perform IMT measurements, and provide the differences between the left and the right sides. The study was performed on 1104 longitudinal-section ultrasound images acquired from 568 men and 536 women out of which 125 had cardiovascular symptoms (CVD). A cardiovascular expert manually delineated the IMT for the normal and the CVD groups. The corresponding (normal versus CVD) IMT mean ± standard deviation values for the left and the right sides were 0.74 ± 0.24 versus 0.87 ± 0.24 mm and 0.70 ± 0.17 versus 0.80 ± 0.18 mm, respectively. The main findings of this paper can be summarized as follows: 1) there was no significant difference between the CCA left side IMT and the right side IMT. These findings suggest that the measurement of the CCA IMT on one side only is needed for the normal group (and this is in agreement with other studies); 2) there were statistical significant differences for the IMT measurements between the normal group and the CVD group for both the left and the right sides; 3) there was an increasing linear relationship of the left and the right IMT measurements with age for the normal group; and to a lesser extend for the CVD group; 4) no statistical significant differences were found between the manual and the automated IMT measurements for both sides; and 5) the best result for classification disease modeling, using support vector machines, to discriminate between the normal and the CVD groups was a 64%±3.5% correct classifications score when using both the left and the right IMT automated measurements. Further research is required for estimating differences and similarities between left and right intima media complex structure and morphology and their variability with texture features for differentiating between the normal and the CVD group.
53 citations
"Efficient Speckle Reduction Techniq..." refers background in this paper
...DsFlsmv filter suppresses the content from the original image which are higher in frequency [4]....
[...]
TL;DR: To objectively and systematically compare the performance of eleven despeckle filters for the breast ultrasound images, several comparison methods are used, such as the full-reference image quality metrics, the nonreference/blind imagequality metrics, observing the removed noise images, as well as the visual evaluation of experts.
Abstract: It is well known that the quality of ultrasound image is significantly degraded by the speckle noise, which has restricted the development of automatic diagnostic techniques for ultrasound images, especially for the breast ultrasound images. This necessitates the need to choose an optimal speckle filtering algorithm for the specific clinical application with different required criteria. In this paper, the study focuses on the comparison of despeckle filters for the breast ultrasound images. Firstly, the models of speckle noise for medical ultrasound images are discussed. After that, eleven despeckle filters which are classified into five categories (local adaptive filter, anisotropic diffusion filter, multi-scale filter, nonlocal means filter, and hybrid filter) are described. Then, the comparative experiments of eleven despeckle filters for the two types of simulated images and clinical ultrasound breast images are presented. Finally, to objectively and systematically compare the performance of eleven despeckle filters, several comparison methods are used, such as the full-reference image quality metrics, the nonreference/blind image quality metrics, observing the removed noise images, as well as the visual evaluation of experts.
43 citations