Adaptive Noise Reduction Filter Research Articles

Adaptive Fuzzy Switching Noise Reduction Filter for Iris Pattern Recognition

Noise reduction is a necessary procedure for the iris recognition systems. This paper proposes an adaptive fuzzy switching noise reduction (AFSNR) filter to reduce noise for iris pattern recognition. The proposed low complexity AFSNR filter removes noise pixels by fuzzy switching between an adaptive median filter and the filling method. The threshold values of AFSNR filter are calculated on the basis of the histogram statistics of eyelashes, pupils, eyelids, and light illumination. The experimental results on the CASIA V3.0 iris database, with genuine acceptance rate equals 99.72%, show the success of the proposed method.

TU-F-18C-03: X-Ray Scatter Correction in Breast CT: Advances and Patient Testing

Purpose: To further develop and perform patient testing of an x-ray scatter correction algorithm for dedicated breast computed tomography (BCT). Methods: A previously proposed algorithm for x-ray scatter signal reduction in BCT imaging was modified and tested with a phantom and on patients. A wireless electronic positioner system was designed and added to the BCT system that positions a tungsten plate in and out of the x-ray beam. The interpolation used by the algorithm was replaced with a radial basis function-based algorithm, with automated exclusion of non-valid sampled points due to patient motion or other factors. A 3D adaptive noise reduction filter was also introduced to reduce the impact of scatter quantum noise post-reconstruction. The impact on image quality of the improved algorithm was evaluated using a breast phantom and seven patient breasts, using quantitative metrics such signal difference (SD) and signal difference-to-noise ratios (SDNR) and qualitatively using image profiles. Results: The improvements in the algorithm resulted in a more robust interpolation step, with no introduction of image artifacts, especially at the imaged object boundaries, which was an issue in the previous implementation. Qualitative evaluation of the reconstructed slices and corresponding profiles show excellent homogeneity of both the background andmore » the higher density features throughout the whole imaged object, as well as increased accuracy in the Hounsfield Units (HU) values of the tissues. Profiles also demonstrate substantial increase in both SD and SDNR between glandular and adipose regions compared to both the uncorrected and system-corrected images. Conclusion: The improved scatter correction algorithm can be reliably used during patient BCT acquisitions with no introduction of artifacts, resulting in substantial improvement in image quality. Its impact on actual clinical performance needs to be evaluated in the future. Research Agreement, Koning Corp., Hologic Inc., Barco Corp Consultant Agreement, Fujifilm Medical Systems.« less

Echocardiographic image tracker with a speckle adaptive noise reduction filter for the automatic measurement of the left atrial volume curve

Since tracking accuracy in left atrial (LA) images decreases due to low image quality around the LA in the apical view, a practical tracking method for LA images has not yet been proposed. The aim of this study was to assess an accurate and high-speed LA volume tracking (LAVT) method for the automatic measurement of LA volume (LAV) curves. We used three approved protocols in this study: (i) LAV curves were measured by LAVT on computer-simulated images; (ii) in 20 healthy volunteers, we assessed the feasibility and accuracy of this method compared with expert's measurements; and (iii) echocardiography and multi-detector row computed tomography (MDCT) imaging were performed on the same day in 20 patients with suspected coronary artery disease. On computer-simulated images, mean absolute percentage LAVT error in one cardiac cycle was 3% in filtered images and 16% in original images. In 20 healthy volunteers, there are strong correlations between LAVT and the expert's LA measurements (LA maximum volume; R = 0.93, P < 0.001). In 400 LA images with 20 patients, an excellent correlation was obtained between LAVs using echocardiography and MDCT (R = 0.98, P < 0.001), with a small bias (-14% of the mean) and narrow limits of agreement (+15% of the mean). The mean time required for the LAVT analysis was 1.8 min, for the MDCT analysis was 35.8 min, and for the manual echocardiographic analysis was 14.0 min. This LAVT method is fast, valid, accurate, and reproducible for determining LAV in both simulated images and the clinical setting.

The efficacy of 2D, non-linear noise reduction filtering in cardiac imaging: a pilot study

Computed tomography (CT) is becoming increasingly popular as a non-invasive method for visualizing the coronary arteries but patient radiation doses are still an issue. Postprocessing filters such as 2D adaptive non-linear filters might help to reduce the dose without loss of image quality. To investigate whether the use of a 2D, non-linear adaptive noise reduction filter can improve image quality in cardiac computed tomography angiography (CCTA). CCTA examinations were performed in 36 clinical patients on a dual source CT using two patient dose levels: maximum dose during diastole and reduced dose (20% of maximum dose) during systole. One full-dose and one reduced-dose image were selected from each of the examinations. The reduced-dose image was duplicated and one copy postprocessed using a 2D non-linear adaptive noise reduction filter, resulting in three images per patient. Image quality was assessed using visual grading with three criteria from the European guidelines for assessment of image quality and two additional criteria regarding the left main artery and the overall image quality. Also, the HU value and its standard deviation were measured in the ascending and descending aorta. Data were analyzed using Visual Grading Regression and paired t-test. For all five criteria, there was a significant (P < 0.01 or better) improvement in perceived image quality when comparing postprocessed low-dose images with low-dose images without noise reduction. Comparing full dose images with postprocessed low-dose images resulted in a considerably larger, significant (P < 0.001) difference. Also, there was a significant reduction of the standard deviation of the HU values in the ascending and descending aorta when comparing postprocessed low-dose images with low-dose images without postprocessing. Even with an 80% dose reduction, there was a significant improvement in the perceived image quality when using a 2D noise-reduction filter, though not approaching the quality of full-dose images. This indicates that cardiac CT examinations could benefit from noise-reducing postprocessing with 2D non-linear adaptive filters.

CT urography in the urinary bladder: To compare excretory phase images using a low noise index and a high noise index with adaptive noise reduction filter

Although CT urography (CTU) is widely used for the evaluation of the entire urinary tract, the most important drawback is the radiation exposure. To evaluate the effect of a noise reduction filter (NRF) using a phantom and to quantitatively and qualitatively compare excretory phase (EP) images using a low noise index (NI) with those using a high NI and postprocessing NRF (pNRF). Each NI value was defined for a slice thickness of 5 mm, and reconstructed images with a slice thickness of 1.25 mm were assessed. Sixty patients who were at high risk of developing bladder tumors (BT) were divided into two groups according to whether their EP images were obtained using an NI of 9.88 (29 patients; group A) or an NI of 20 and pNRF (31 patients; group B). The CT dose index volume (CTDI(vol)) and the contrast-to-noise ratio (CNR) of the bladder with respect to the anterior pelvic fat were compared in both groups. Qualitative assessment of the urinary bladder for image noise, sharpness, streak artifacts, homogeneity, and the conspicuity of polypoid or sessile-shaped BTs with a short-axis diameter greater than 10 mm was performed using a 3-point scale. The phantom study showed noise reduction of approximately 40% and 76% dose reduction between group A and group B. CTDI(vol) demonstrated a 73% reduction in group B (4.6 ± 1.1 mGy) compared with group A (16.9 ± 3.4 mGy). The CNR value was not significantly different (P = 0.60) between group A (16.1 ± 5.1) and group B (16.6 ± 7.6). Although group A was superior (P < 0.01) to group B with regard to image noise, other qualitative analyses did not show significant differences. EP images using a high NI and pNRF were quantitatively and qualitatively comparable to those using a low NI, except with regard to image noise.

Low-Dose MDCT Urography: Feasibility Study of Low-Tube-Voltage Technique and Adaptive Noise Reduction Filter

The purpose of this study was to investigate the feasibility of performance of MDCT urography with low tube voltage and an adaptive noise reduction filter. Thirty-one patients underwent excretory phase (300 seconds after administration of 100 mL of iopamidol) 40-MDCT of the urinary tract at 120 and 80 kVp. The 80-kVp images were postprocessed with an adaptive noise reduction filter. Using a 3-point scale for homogeneity of the urinary tract and sharpness of contour, streak artifacts, and overall image quality, two radiologists evaluated coronal multiplanar reconstruction images generated from 120-kVp, unfiltered 80-kVp, and filtered 80-kVp images. Attenuation values of the abdominal aorta, renal pelvis, renal cortex, psoas muscle, vertebral body, and retroperitoneal fat and image noise of the psoas muscle were measured. The effective radiation dose was estimated for each patient. At visual evaluation of images of the upper urinary tract, the quality of filtered 80-kVp images was comparable with that of 120-kVp images. At evaluation of images of the lower urinary tract, however, filtered 80-kVp images were of inferior quality. Except for those of fat tissue, attenuation values were significantly higher on 80-kVp than on 120-kVp images (paired Student's t test, p < 0.01). Noise values did not differ significantly between 120- and filtered 80-kVp images (Dunnett test, p = 0.37). The mean effective doses for 120- and 80-kVp scans were 7.0 and 2.9 mSv. MDCT urography is feasible with a low-tube-voltage technique and an adaptive noise reduction filter. The technique allows reduction in radiation dose without marked degradation of image quality and can be used in clinical assessment of the renal collecting system and upper ureter. For evaluation of the pelvic ureter and urinary bladder, however, image quality is not sufficient, and a compensatory increase in tube current may be necessary.

Radiation dose reduction in hepatic multidetector computed tomography with a novel adaptive noise reduction filter

The aim of this study was to optimize a novel adaptive noise reduction filter based on patient body weight and to investigate its utility for improving the image quality of low-dose hepatic computed tomography (CT) scans. The tube current-time product was changed from 140 to 180 and from 60 to 100 mAs at standard- and low-dose CT, respectively, based on the body weights of 45 patients. Unenhanced and two-phase contrast-enhanced helical scans were obtained at the standard dose during the hepatic arterial and equilibrium phases. During the equilibrium phase, we obtained low-dose scans of the liver immediately after standard-dose scans. The low-dose CT images were postprocessed with the filter. Two radiologists visually evaluated artifacts in the liver parenchyma and its graininess, the sharpness of the liver contour, tumor conspicuity, homogeneity of the enhancement of the portal vein, and overall image quality. There was no statistically significant difference between standard and filtered low-dose images with respect to artifacts in the liver, the graininess of the liver parenchyma, tumor conspicuity, homogeneity of enhancement of the portal vein, or overall image quality. The adaptive noise reduction filter effectively reduced image noise. We confirmed the effectiveness of the filter by examining clinical hepatic images obtained at low-dose CT.

Improvement of Low-Contrast Detectability in Low-Dose Hepatic Multidetector Computed Tomography Using a Novel Adaptive Filter

The purpose of this study was to investigate how much radiation dose can be reduced without loss of low-contrast detectability with a newly developed adaptive noise reduction filter in hepatic multidetector computed tomography (MDCT) scans by using a computer-simulated liver phantom. Simulated CT images, including liver and intrahepatic tumors, were mathematically constructed using a computer workstation to evaluate low-contrast detectability by the observer performance test. Milliampere second for construction of simulated images were 60, 80, 100, and 120 mAs (low dose) and 160 mAs (standard dose) at 120 kVp. Images with 60, 80, 100, and 120 mAs were postprocessed with the adaptive noise reduction filter. A total of 432 images were prepared and receiver operating characteristic (ROC) analysis was performed by 5 radiologists. The detectability of simulated tumor by radiologists was estimated with the area under the ROC curves (Az values). In addition, we visually evaluated CT images of 15 patients with chronic liver damage for graininess of the liver parenchyma, sharpness of the liver contour, conspicuity and marginal sharpness of the liver tumors, and overall image quality. The mean Az value at 0.777 (60 mAs), 0.828 (80 mAs), and 0.844 (100 mAs) without filter was significantly lower than that of 160 mAs without filter (P < 0.001, 60 mAs; P = 0.010, 80 mAs; P = 0.040, 100 mAs). There was no statistical difference between the mean Az value at 80 mAs with and 160 mAs without the adaptive noise reduction filter (P = 0.220) and 100 mAs with and 160 mAs without the adaptive noise reduction filter (P = 0.979). In the visual evaluation of patient livers, there was no statistical difference in the graininess and sharpness of the liver, the conspicuity and marginal sharpness of the tumor, and the overall image quality between standard-dose and filtered low-dose images (Wilcoxon signed rank test, P > 0.05). The radiation dose can be reduced by 50% without loss of nodule detectability by applying the adaptive noise reduction filter to simulated and patient liver images obtained at MDCT.

Noise Reduction Filter of Cascaded Sandglass-type Neural Network Adaptively Controlled by Noise Intensity for Speech Signal.

An adaptive noise reduction filter composed of Cascaded Sandglass-type Neural Network (CSNNRF) is proposed to develop a hearing aid appliance. The number of unit sandglass-type neural networks (SNNs) is controlled adaptively by noise intensity. Usually the hearing aid works outside where noise intensity is altering all the times. An adaptive noise reduction filter controlled by noise intensity is essential to cope with these circumstances. Each SNN has a three-layer structure and consists of the same number of neural units in the input and output layers and a single neural unit in the hidden layer.SNN are connected in cascaded to be CSNNRF. The number of unit SNNs is adaptively determined by our algorithms so that hearing intelligibity for speech signal may be processed more preferablely. To determine the number of SNNs, we regard significance on the improvement of intelligibility more than numerical value itself such as S/N ratio of speech signal.

Adaptive noise reduction by using cascaded sandglass‐type neural networks

AbstractAn adaptive noise reduction filter composed of cascaded sandglass‐type neural networks (CSNNRF) is proposed. A given number of unit sandglass‐type neural networks (SNN), each of which has a three‐layer structure and consists of the same number of neural units in the input and the output layers and a single neural unit the hidden layer, are connected in cascade. The number of unit SNNs is adaptively determined so as to be equal to a rank of covariance matrix of an original noise‐free signal (signal component). Outputs of hidden layer units in individual unit SNNs, whose variances are equivalent to eigenvalues of the covariance matrix of the observed signal, are statistically compared by use of ANOVA (analysis of variance) to estimate the rank. When a CSNNRF is composed of the number of unit SNNs equal to the rank, the observed signal is filtered without any loss of signal component while noise component is maximally reduced. It was shown by computer experiments that the rank was almost always estimated accurately in an adaptive manner, and that noise reduction from the signal was carried out optimally. © 2001 Scripta Technica, Electr Eng Jpn, 136(1): 37–46, 2001

An adaptive noise reduction filter for discrete signal by use of sandglass‐type neural network

An adaptive noise reduction filter for discrete signal by use of sandglass‐type neural network

An adaptive noise reduction filter for discrete signal by use of sandglass‐type neural network

An adaptive noise reduction filter for discrete signal by use of sandglass‐type neural network

An adaptive noise reduction filter for discrete signal by use of sandglass-type neural network

An adaptive noise reduction filter composed of a sandglass-type neural network (SNN) noise reduction filter (RF) is proposed in this paper. SNN was originally devised to work effectively for information compression. It is a hierarchial network and is symmetrically structured. SNN consists of the same number of units in the input and output layers and a smaller number of units in the hidden layer. It is known that SNN has signal processing performance which is equivalent to Karhunen–Loeve expansion after learning. We proved the theoretical suitability of SNN for an adaptive noise reduction filter for discrete signals. The SNNRF behaves optimally when the number of units in the hidden layer is equal to the rank of the covariance matrix of the signal components included in the input signal. Further we show by applying the recursive least squares method to learning of the SNNRF that the filter can process signals for on-line adaptive noise reduction. This is an extremely desirable feature for practical application. In order to verify the validity of SNNRF, we performed computer experiments examining how the noise reduction ability of SNNRF is affected by altering the properties of the input pattern, learning algorithm, and SNN. The results confirm that the SNNRF acquired appropriate characteristics for noise reduction from the input signals, and markedly improved the SNR of the signals. © 1999 Scripta Technica, Electr Eng Jpn, 127(4): 39–51, 1999

Evaluation of a digital adaptive filter, noise-reduction scheme for normal and hearing-impaired listeners

A common complaint of hearing aid wearers is that they have difficulty when listening in a background of noise. The ability of a digital adaptive noise reduction filter to reduce background noise without degrading speech perception has been evaluated. Each of four normal-hearing and six hearing-impaired subjects completed two behavioral tests of speech perception—the speech reception threshold (SRT) and the modified distinctive features differences test [DFD(m)]—and one subjective measure of listener preference (paired comparisons). Each test was performed in three types of background noise presented at a level of 65 dB: white noise, low-pass filtered at 8.0 kHz; white noise, low-pass filtered at 1 kHz; and multitalker speech babble. For hearing-impaired subjects, the speech-plus-noise signal was filtered and amplified to provide the NAL real-ear insertion gain. Results indicated that all subjects preferred the noise-reduction-processed speech-plus-noise signal over the unprocessed signal at positive signal to noise ratios. The effect of the adaptive digital filter on speech perception depended on hearing loss, noise type, and SNR. Feature analysis of DFD(m) responses indicated that processing had different effects on the error patterns for normal and hearing-impaired listeners.