Fact-checked by Grok 2 weeks ago

Contrast-to-noise ratio

Contrast-to-noise ratio (CNR) is a measure used to determine image quality by quantifying the difference in signal intensity between two regions of interest relative to the level of background noise, thereby assessing the visibility and detectability of structures. It is particularly prominent in medical imaging, such as for tissues or lesions, but also applied in non-medical fields like non-destructive testing.^[1] This measure is essential for evaluating image quality, as higher CNR values indicate superior contrast resolution, enabling clearer differentiation of features against noise, which directly impacts diagnostic accuracy in medical contexts.^[1] Unlike signal-to-noise ratio (SNR), which focuses solely on signal strength relative to noise, CNR incorporates the contrast between specific regions, making it particularly useful for tasks like lesion detection.^[2] The standard formula for CNR is typically expressed as CNR = |μ₁ - μ₂| / σ, where μ₁ and μ₂ represent the mean signal intensities in the two regions, and σ is the standard deviation of the noise.^[3] Variations of this formula exist across imaging modalities; for instance, in X-ray imaging, it may account for quantum noise, while in ultrasound, generalized forms like gCNR = 1 - overlap integral of probability density functions address limitations in dynamic range.^[4]^[5] CNR plays a critical role in optimizing imaging parameters—for example, adjusting kVp and mA in X-rays or TE and TR in MRI—to balance radiation dose, scan time, and diagnostic utility.^[1] Across modalities such as computed tomography (CT), magnetic resonance imaging (MRI), ultrasound, and digital radiography, CNR guides protocol development and quality assurance, with applications including dose reduction in CT via iterative reconstruction and enhanced vascular visualization in contrast-enhanced ultrasound.^[1] In clinical practice, CNR thresholds help determine the minimum requirements for reliable detection of abnormalities, such as tumors or fractures, ensuring that images maintain sufficient quality for effective radiological diagnosis.^[4] Research continues to refine CNR metrics, particularly for advanced beamforming in ultrasound^[5] and high-field MRI,^[6] to improve lesion detectability and overall imaging performance.

Definition and Fundamentals

Definition

The contrast-to-noise ratio (CNR) serves as a quantitative measure of an image's ability to distinguish signals from distinct regions, such as tissues or objects, relative to the level of background noise present.^[7] This metric evaluates image quality by balancing the perceptible differences between features against the confounding effects of noise, enabling more reliable identification of subtle variations in intensity.^[8] The concept of assessing image quality by combining contrast and noise emerged in the late 1970s with contrast-detail-dose analysis in computed tomography (CT), as developed by Cohen and DiBianca (1979) to address limitations of standalone contrast metrics. The contrast-to-noise ratio (CNR) was formally introduced in 1983 by Patterson and Foster in the context of ultrasound imaging.^[9]^[10] Intuitively, CNR gauges the detectability of structural differences in an image, with higher values signifying enhanced clarity in separating elements like pathological features from adjacent areas despite noise.^[11] For example, in a grayscale image, a high CNR would indicate that a low-intensity tumor region is prominently distinguishable from a noisy mid-intensity background, aiding in its visual isolation.^[4] CNR relates to but extends beyond signal-to-noise ratio (SNR) by incorporating inter-region contrast, providing a more targeted assessment of feature separability.^[12]

Relation to Contrast and Noise

Contrast in imaging refers to the absolute difference in signal intensity between two distinct regions, such as a target area and its background, often expressed as |μ_A - μ_B|, where μ_A and μ_B represent the mean intensities of those regions.^[13] This difference arises from variations in tissue properties or object characteristics that affect how signals are captured, enabling the visual separation of features like lesions from surrounding tissue.^[14] Noise, on the other hand, manifests as random fluctuations in pixel values across the image, primarily due to stochastic processes like photon detection variability in X-ray imaging or thermal effects in other modalities. It is typically quantified by the standard deviation (σ) of pixel intensities in uniform regions, such as the background or signal areas, capturing the inherent variability that degrades image fidelity.^[15]^[16] While contrast provides the raw distinction between regions, it alone is insufficient for assessing image quality because high noise levels can mask these differences, rendering subtle features indistinguishable even when contrast is present. For instance, increasing contrast without addressing noise may not improve visibility if fluctuations obscure the signal boundaries, highlighting the need for a metric that normalizes contrast against noise to evaluate detectability reliably.^[7] To visualize this, contrast resembles the brightness disparity between two objects in a scene, whereas noise acts like static interference on a television screen, which can overwhelm and hide those details regardless of their inherent difference.^[17]

Mathematical Formulation

Basic Formula

The contrast-to-noise ratio (CNR) is fundamentally expressed as

\text{CNR} = \frac{|\mu_{\text{object}} - \mu_{\text{background}}|}{\sigma_{\text{noise}}}

where \mu_{\text{object}} and \mu_{\text{background}} represent the mean signal intensities in the regions of interest for the object (e.g., tissue of interest) and background, respectively, and \sigma_{\text{noise}} denotes the standard deviation of the noise, typically estimated from the background region to reflect signal variability.^[18] This formulation originates from early work on visual perception and imaging detectability, where Albert Rose established that reliable detection requires the signal contrast to exceed noise fluctuations by a factor related to the standard deviation, laying the groundwork for CNR as a normalized metric.^[19] The derivation begins with the basic contrast, defined as the absolute difference |\mu_{\text{object}} - \mu_{\text{background}}|, which quantifies the inherent difference in signal levels between regions without accounting for variability. To incorporate the effects of noise, which degrades separability, this difference is divided by \sigma_{\text{noise}}, yielding a ratio that indicates how prominently the contrast stands relative to random fluctuations—essentially, the number of standard deviations separating the means. This normalization assumes an additive noise model, where the observed image intensity I at each pixel is I = S + N, with S as the true signal and N as zero-mean noise; under Gaussian noise conditions, the distributions of object and background signals are normal with variance \sigma_{\text{noise}}^2.^[18]^[20] Key assumptions underlying this basic formula include spatially uniform noise across the image (i.e., constant \sigma_{\text{noise}} variance) and an additive, Gaussian-distributed noise process, which holds approximately in many linear imaging systems like X-ray or MRI under low-noise regimes. The metric is dimensionless, allowing direct comparability across modalities or acquisition settings without unit dependencies.^[18] These assumptions simplify analysis but may require adjustments in scenarios with non-uniform or structured noise. For illustration, consider an image where \mu_{\text{object}} = 100, \mu_{\text{background}} = 80, and \sigma_{\text{noise}} = 5; the CNR is then (100 - 80)/5 = 4, suggesting moderate separability where the contrast spans four noise standard deviations, sufficient for detection in many contexts per the Rose criterion threshold of around 5.

Variations and Generalized Forms

One variation of the basic contrast-to-noise ratio (CNR) accounts for noise measured separately in the two regions of interest, particularly when noise is independent but potentially varying between regions. In this form, the CNR is calculated as \text{CNR} = \frac{|\mu_A - \mu_B|}{\sqrt{\sigma_A^2 + \sigma_B^2}}, where \mu_A and \mu_B are the mean intensities in regions A and B, and \sigma_A and \sigma_B are the standard deviations of noise in each region, respectively; this approach captures the variance of the intensity difference more accurately than using a single noise estimate.^[21] A generalized form of CNR, known as the generalized CNR (gCNR), extends the metric to multiple regions or lesions within an image, making it robust to scaling and applicable across diverse image types like photoacoustic or ultrasound data. Defined as \text{gCNR} = \sqrt{ \sum_i \left( \frac{\mu_i - \mu_{\text{bg}}}{\sigma_{\text{bg}}} \right)^2 }, where \mu_i are means in target regions i, \mu_{\text{bg}} is the background mean, and \sigma_{\text{bg}} is the background noise standard deviation, this formulation aggregates individual CNR contributions into a single, scale-invariant value that better reflects overall lesion detectability; it was introduced in research on photoacoustic image quality assessment in 2020.^[22] Other adaptations integrate perceptual models for human detectability or extend CNR to multichannel images. The Rose model, a seminal framework for visual signal detection, posits that CNR thresholds of 2 to 5 are required for reliable human perception of low-contrast features, with values above 3-5 ensuring visibility in noisy backgrounds by relating CNR to the minimum detectable signal relative to noise fluctuations.^[23] For color images, CNR is often computed on the luminance channel (e.g., Y in YCbCr space) to prioritize achromatic contrast, as human vision is more sensitive to luminance variations than chrominance, yielding \text{CNR} = \frac{|\mu_{Y_A} - \mu_{Y_B}|}{\sigma_Y} where parameters are derived from the luminance component.^[15] The basic CNR assumes uniform (homoscedastic) noise across the image, which limits its applicability in modern digital systems where noise varies spatially due to factors like photon shot noise or processing artifacts. In contrast, generalized forms like gCNR accommodate heteroscedastic noise by incorporating local or background-specific statistics, providing more reliable assessments in non-uniform environments such as advanced beamforming in ultrasound.^[5]

Applications

In Medical Imaging

In computed tomography (CT) and X-ray imaging, contrast-to-noise ratio (CNR) plays a pivotal role in detecting low-contrast lesions, such as tumors, where subtle differences in tissue attenuation must be distinguished from background noise.^[4] A typical CNR value exceeding 3 is often required for reliable lesion detection, as lower values compromise visibility and increase the risk of oversight in diagnostic assessments.^[24] CNR in these modalities is influenced by acquisition parameters like kilovolt peak (kVp) and milliampere-seconds (mAs), with higher kVp enhancing CNR by reducing noise through increased beam penetration, while elevated mAs lowers noise by improving photon statistics.^[25] In magnetic resonance imaging (MRI), CNR quantifies tissue differentiation, particularly between gray and white matter, enabling precise anatomical delineation essential for neurological diagnostics. In T1-weighted images, CNR values typically range from 10 to 20, reflecting strong signal differences modulated by relaxation times and field strength (e.g., higher at 3T than 1.5T).^[18] Contrast agents, such as gadolinium-based compounds, further enhance CNR by shortening T1 relaxation times in targeted tissues, improving lesion conspicuity in applications like tumor evaluation.^[26] Ultrasound imaging relies on CNR to evaluate echo signal differences against speckle noise, a multiplicative artifact inherent to the modality that reduces overall image clarity. CNR values in ultrasound are generally lower than in CT or MRI due to this noise profile, which limits resolution for soft-tissue interfaces like cysts or vascular structures.^[5]^[27] Clinically, CNR thresholds guide diagnostic reliability across modalities; values below 2 frequently result in false negatives by obscuring subtle pathologies, as supported by radiology guidelines emphasizing minimum CNR for low-contrast detectability. For instance, the American College of Radiology (ACR) standards as of 2023 maintain a CNR threshold of at least 1 for CT accreditation in low-contrast phantom testing, but clinical protocols advocate higher values (e.g., >3) to minimize diagnostic errors in patient imaging.^[28]^[29]

In Non-Medical Fields

In non-destructive testing (NDT), particularly radiographic inspection, the contrast-to-noise ratio (CNR) serves as a key metric for evaluating the visibility of defects such as cracks in welds or material inclusions. By quantifying the difference in signal intensity between the defect and surrounding material relative to image noise, CNR helps determine if flaws are detectable against background variations. Standards like ASTM E2698 specify measurement of CNR using image quality indicators, requiring a minimum value of 2.5 in the 4T hole for acceptable digital radiographic image quality in industrial applications.^[30] This threshold ensures reliable defect detection in safety-critical components, such as pipelines or aircraft structures, where low CNR could lead to overlooked flaws.^[31] In astronomy, CNR is adapted to assess the detectability of faint celestial objects, like exoplanets, against stellar glare and cosmic noise sources, including photon noise from limited light collection. High-contrast imaging techniques optimize CNR by suppressing starlight while preserving planet signals, enabling differentiation in photon-noise-limited regimes. For exoplanet detection, algorithms directly maximize the nonlinear signal-to-noise equation, which aligns with CNR principles, to achieve contrasts as low as 10^{-7} at small angular separations.^[32] This application is crucial for instruments like coronagraphs on telescopes, where photon noise dominates and CNR guides observation strategies for habitable zone planets.^[33] Digital photography and machine vision leverage CNR to quantify image sharpness and reliability for automated tasks, such as defect inspection in manufacturing processes. In camera systems, CNR evaluates how well features stand out from noise, influencing overall image quality metrics integrated into specifications since the early 2010s, alongside dynamic range assessments. For machine vision, low CNR values hinder automated detection of surface defects like scratches on automotive parts, prompting the use of CNR thresholds to validate imaging setups.^[34] In radiographic machine vision for aerospace composites, CNR quantifies small defect contrast against noise, with values below 3 often indicating insufficient discriminability for automated grading.^[35] Emerging applications in remote sensing utilize CNR to enhance the interpretation of satellite imagery for environmental monitoring, including land cover changes from deforestation or urbanization. By measuring contrast between vegetation, water, or urban features relative to atmospheric and sensor noise, CNR improves classification accuracy in multispectral data. Restoration techniques in remote sensing boost CNR to distinguish subtle land cover transitions, supporting quantitative assessments of environmental shifts over decadal scales.^[36]

Comparison with Other Metrics

Versus Signal-to-Noise Ratio

The signal-to-noise ratio (SNR) is defined as the ratio of the mean signal intensity in a region of interest to the standard deviation of the noise, expressed as SNR = \frac{\mu_{\text{signal}}}{\sigma_{\text{noise}}}. This metric primarily assesses the absolute strength of a signal relative to the background noise, providing a measure of overall image fidelity and clarity in uniform regions.^[20]^[18] In contrast, the contrast-to-noise ratio (CNR) focuses on the differential signal between two regions, given by CNR = \frac{|\mu_A - \mu_B|}{\sigma_{\text{noise}}}, where \mu_A and \mu_B represent the mean intensities of the regions of interest. This emphasis on relative contrast makes CNR particularly valuable for detecting subtle changes, such as lesions against surrounding tissue in medical imaging, whereas SNR is less sensitive to such differences and better suits evaluations of global signal robustness.^[20]^[18] SNR is commonly applied in scenarios involving uniform signals, like phantom testing to benchmark scanner performance, while CNR excels in heterogeneous environments, such as delineating tissue boundaries where contrast visibility is paramount. For regions experiencing similar noise levels, CNR can be approximated as the difference between their individual SNRs, CNR \approx SNR_A - SNR_B, highlighting the interconnected yet distinct roles of these metrics.^[20]

Versus Peak Signal-to-Noise Ratio

The Peak Signal-to-Noise Ratio (PSNR) is defined as

\text{PSNR} = 10 \log_{10} \left( \frac{\MAX^2}{\MSE} \right),

where \MSE denotes the mean squared error between the reference and processed images, and \MAX is the maximum possible pixel value in the image.^[37] This metric is primarily employed in image compression and fidelity assessment to quantify overall distortion levels.^[37] Unlike CNR, which provides a direct ratio of contrast differences to noise for task-specific evaluation, PSNR functions as a logarithmic measure of pixel-wise errors, rendering it less attuned to localized contrast variations that affect perceptual interpretation.^[38] CNR's emphasis on relative signal differences makes it more aligned with human visual tasks, whereas PSNR aggregates errors globally without prioritizing feature separability.^[5] PSNR exhibits limitations in capturing human visual perception, as it correlates poorly with subjective quality assessments and detectability, often overlooking how contrast influences visibility.^[39] For example, high PSNR values may not reflect noticeable artifacts in low-contrast regions. In contrast, CNR better supports tasks like lesion detection in medical imaging by quantifying separability, with 2010s studies validating its predictive power for low-contrast object visibility.^[25]^[40] Certain imaging workflows integrate both metrics for a multifaceted quality analysis, yet CNR is favored in contrast-critical domains over PSNR's focus on peak error minimization.

Factors Influencing CNR

Acquisition Parameters

In X-ray and computed tomography (CT) imaging, radiation dose parameters such as milliampere-seconds (mAs) directly influence CNR by modulating the photon flux, which affects both signal intensity and noise levels. Increasing mAs enhances the signal strength while reducing quantum noise, thereby improving CNR, though this comes at the cost of elevated patient radiation exposure risk. Specifically, CNR scales with the square root of the dose due to the inverse relationship between image noise and the square root of the number of detected photons.^[41] In magnetic resonance imaging (MRI), the main field strength significantly impacts CNR, with higher Tesla values generally yielding superior performance. For instance, 3T systems provide substantially higher CNR compared to 1.5T systems, often achieving approximately double the signal-to-noise ratio (SNR) under similar conditions, which translates to enhanced contrast differentiation. Additionally, sequence parameters like repetition time (TR) and echo time (TE) trade off acquisition duration for CNR optimization; longer TR allows fuller longitudinal relaxation for stronger signals and better CNR, while adjusted TE can maximize T2 contrast at the expense of signal decay and increased scan time.^[42] For ultrasound imaging, transducer frequency and receiver gain settings play key roles in balancing resolution and CNR during acquisition. Higher frequencies improve axial resolution and near-field CNR by enhancing tissue differentiation, but they increase attenuation, reducing penetration depth and degrading CNR in deeper tissues due to amplified noise from scattering and absorption. Receiver gain amplifies both echo signals and inherent noise proportionally, maintaining CNR but potentially introducing artifacts if over-applied, thus requiring careful adjustment to preserve dynamic range without saturation.^[43]^[44] Across modalities, acquisition parameters often involve inherent trade-offs, such as between spatial resolution and CNR, where smaller voxel sizes enhance detail but amplify noise by reducing signal per voxel, thereby lowering CNR. In radiation-based modalities like X-ray and CT, recent guidelines emphasize the ALARA (As Low As Reasonably Achievable) principle to balance these factors, advocating minimal effective dose while ensuring adequate CNR for diagnostic utility.^[45]

Noise Reduction Techniques

Spatial filtering techniques, such as Gaussian and median filters, are commonly applied post-acquisition to reduce noise in medical images, thereby enhancing the contrast-to-noise ratio (CNR) by smoothing out high-frequency noise components. Gaussian filters apply a weighted average based on a bell-shaped kernel, effectively lowering the standard deviation of noise (σ_noise) while preserving overall image structure, though they can introduce blurring that slightly attenuates contrast in fine details. Median filters, being non-linear, replace each pixel with the median value of its neighborhood, which is particularly effective for impulse noise and can effectively reduce noise in uniform regions without the Gaussian's smoothing artifacts. Adaptive variants of these filters, which adjust kernel size or weights based on local image statistics, further mitigate blurring by preserving edges, leading to CNR improvements in low-contrast areas like soft tissues in MRI or CT scans.^[46] Advanced denoising algorithms offer more sophisticated noise suppression while better maintaining spatial resolution and contrast. Wavelet-based denoising decomposes the image into frequency subbands using wavelet transforms, applying thresholding to remove noise-dominated coefficients before reconstruction; this approach has been shown to significantly enhance CNR in magnetic resonance (MR) images by sharpening edges and improving low-intensity feature contrast, with studies reporting CNR gains of approximately 1.7 times in hepatobiliary-phase liver imaging compared to non-denoised counterparts.^[47] Since 2018, deep learning methods, particularly convolutional neural network (CNN)-based models, have emerged as powerful tools for CNR enhancement, learning noise patterns from paired noisy-clean datasets to suppress artifacts without excessive detail loss; for instance, CNN denoisers applied to non-contrast head CT images achieve approximately 2-3 times higher SNR (and thus CNR) between gray and white matter regions relative to traditional methods like block-matching 3D filtering.^[48] These techniques excel in handling complex noise distributions, such as Rician noise in magnitude MR images, yielding sharper reconstructions and substantial CNR improvements in clinical datasets. In computed tomography (CT) and magnetic resonance imaging (MRI), iterative reconstruction algorithms model noise statistically during the reconstruction process, iteratively refining the image to minimize discrepancies between measured and simulated projections, which boosts CNR at fixed radiation doses. Adaptive statistical iterative reconstruction (ASIR), a widely adopted method in CT, reduces image noise by 20-40% compared to filtered back projection while increasing signal-to-noise ratio (SNR)—and thus CNR—by 30-40% in coronary and abdominal scans, enabling dose reductions of up to 50% without compromising diagnostic quality.^[49] Model-based iterative reconstruction (MBIR) extends this by incorporating prior knowledge of image physics, achieving greater CNR enhancements, such as approximately 1.3 times higher values in liver tissues at quarter-dose levels versus standard full-dose reconstruction, as validated in upper abdominal CT studies.^[50] Despite these benefits, noise reduction techniques involve inherent trade-offs, where excessive denoising can diminish contrast by oversmoothing edges and textures, potentially degrading CNR in regions requiring high spatial fidelity. Over-denoising may introduce biases or artifacts that mask subtle anatomical features, as observed in evaluations of deep learning and wavelet methods, where aggressive noise suppression reduced variance but compromised low-contrast detectability in MRI and CT. Validation through metrics like CNR pre- and post-processing, alongside subjective assessments, is essential to balance noise reduction with contrast preservation, ensuring clinical utility without loss of diagnostic information.

References

[1]
Enhancing Radiological Diagnosis: A Comprehensive Review ... - NIH
Jun 24, 2024 · The CNR quantifies the difference in signal intensity between two regions of interest relative to the noise level in the image. Higher CNR ...
[2]
CONTRAST-TO-NOISE RATIO VERSUS SIGNAL-TO ... - Vidisco
Nov 7, 2021 · Contrast-to-noise ratio (CNR) is a measure used to determine image quality. CNR is similar to the metric signal-to-noise ratio (SNR), but subtracts a term ...
[3]
Contrast-to-Noise Ratio (CNR) MRI | CNR Formula MRI - MRI Master
CNR stands for Contrast-to-Noise Ratio in MRI. It is a metric used to assess the quality of an MRI image by measuring the contrast between different tissues ...
[4]
X-ray Contrast To Noise (CNR) Illustrated Examples Of Image Noise ...
The Contrast to Noise Ratio (CNR) in a medical image is a measure of the contrast between the tissue of interest and the background (ie the neighboring tissue).Signal to Noise Ratio (SNR... · Quantum Noise (a.k.a... · Perceived Image Quality...
[5]
The generalized contrast-to-noise ratio: a formal definition for lesion ...
In the last 30 years the contrast-to-noise ratio (CNR) has been used to estimate contrast and lesion detectability in ultrasound images.
[6]
Radiographic Techniques, Contrast, and Noise in X-Ray Imaging
Jan 23, 2015 · The ratio of lesion contrast to image mottle is known as the contrast-to-noise ratio (CNR). This ratio is an indicator of the relative image ...
[7]
Assessing contrast on MR images. | Radiology - RSNA Journals
The ability to visually differentiate two objects on the basis of signal intensity depends on the contrast-to-noise ratio (CNR). This parameter, however ...Missing: definition | Show results with:definition
[8]
Contrast-Detail-Dose Analysis of Six Different Computed... - LWW
Contrast-Detail-Dose Analysis of Six Different Computed Tomographic Scanners ... Journal of Computer Assisted Tomography 3(2):p 197-203, April 1979.
[9]
detail--dose analysis of six different computed tomographic scanners
Contrast--detail--dose analysis of six different computed tomographic scanners. J Comput Assist Tomogr. 1979 Apr;3(2):197-203. ... Author. G Cohen. PMID: 429628 ...
[10]
Contrast to Noise Ratio - an overview | ScienceDirect Topics
Contrast-to-noise ratio (CNR) measures image quality, quantifying contrast relative to background noise, calculated as (I o − I b) / σ b.
[11]
CNR vs. SNR – What Radiological Technologists Need to Know
Contrast-to-noise ratio (CNR) analysis is a method of evaluating contrast resolution in an X-ray image. The CNR gives evaluation results similar to those of a ...Missing: definition radiology
[12]
On the Definition of Signal-To-Noise Ratio and Contrast-To ... - NIH
Nov 6, 2013 · Also the relationship between SNR and CNR (contrast-to-noise ratio, a measure of image quality based on a contrast rather than the raw signal) ...
[13]
Image quality - Radiology Cafe
1. Contrast. Contrast is the difference in the displayed or image signal intensity between two areas of interest e.g. a lesion and background tissue.
[14]
Noise | Imatest
Imatest subtracts gradual pixel level variations from the image before calculating noise (the standard deviation of pixel levels in the region under test).
[15]
[PDF] Image Noise in Radiography and Tomography - Juniper Publishers
Oct 11, 2019 · Image noise can be calculated statistically from the standard deviation of CT number or pixel intensity values in a physical uniform region. As ...<|control11|><|separator|>
[16]
Contrast to Noise Ratio (CNR) - Telerad Physics Teaching
Jul 14, 2018 · If the contrast between things is large, then you will still be able to see a difference if there is a lot of noise. If contrast is small, then ...
[17]
Measurement of Signal-to-Noise and Contrast-to-Noise in the fBIRN ...
This study looks at the SNR and CNR ratios in structural fast spin-echo T 2 -weighted scans acquired in five individuals across ten sites.
[18]
On the Definition of Signal-To-Noise Ratio and Contrast-To-Noise ...
In science and engineering, the signal-to-noise ratio (SNR) is a measure that compares the level of a desired signal to the level of background noise.
[19]
Contrast to Noise Ratio - an overview | ScienceDirect Topics
(11) fSNR = aSNR / 1 + λ 2 aSNR 2 . Figure 11 compares measured anatomic and functional signal-to-noise ratios, varied by using two field strengths ...
[20]
Application of the generalized contrast-to-noise ratio to assess ... - NIH
Jun 10, 2020 · Because gCNR is fundamentally a measurement of the separability between two regions in an image, it is applicable to other medical images, such ...
[21]
[PDF] Contrast
The more appropriate measure is the ratio of the contrast to the noise standard deviation, known as the contrast-to-noise ratio (CNR):. The utility of this ...
[22]
Detectability of Hepatic Lesions in DOTATATE PET
Jun 1, 2024 · This high level of lesion detectability (99%) was maintained to a CNR threshold of > 3.5. Below this level, ranging from 2.0 to 3.5, a ...
[23]
The Feasibility of Contrast-to-Noise Ratio on Measurements to ...
Jul 6, 2020 · Contrast-to-noise ratio (CNR) is an important tool used to determine image quality. It is a measure of image quality established on a contrast.
[24]
Clinical feasibility of CS-VIBE accelerates MRI techniques ... - Nature
Jun 20, 2023 · The CNRWM/GM was calculated based on a previously published formula: (SI of white matter–SI of gray matter)/noise of white matter. Due to the ...
[25]
Assessment of Low-Contrast Resolution for the ACR CT ... - NIH
The CNR threshold used by the ACR CT accreditation program yields higher PPRs than visual assessment of LCR, potentially resulting in lower quality images.Missing: 2023 false negatives
[26]
Effect of miscentering and low-dose protocols on contrast resolution ...
Hence, a miscentered position should be avoidable as it can degrade the diagnostic value of the images, thereby leading to false-negative findings. To conclude, ...
[27]
[PDF] 8.3 Minimum Requirements for Digital Radiography ... - NDT.net
Measurement of Contrast to Noise Ratio in DDA Practice. • CNR shall be measured in the. 4T hole for proof of image quality. • A minimum CNR of 2.5 is required ...
[28]
[PDF] Importance of contrast-to-noise ratio sensitivity function in ...
Jan 7, 2025 · difficult or impractical to estimate product quality by destructive testing, one can estimate it using nondestructive testing. Fraction of ...Missing: ASME | Show results with:ASME
[29]
Improved Contrast in Images of Exoplanets Using Direct Signal-to ...
In this paper, we present an algorithm that directly optimizes the nonlinear equation for planet signal-to-noise ratio (S/N).
[30]
Exoplanet detection limits using spectral cross-correlation with ...
A contrast of a few 10−7 at 50 mas should be within reach for bright targets in a photon noise regime with molecular mapping. The spectral information that is ...
[31]
[PDF] Information capacity: a measure of potential image quality of a digital ...
The aim of the paper is to define an objective measurement for evaluating the performance of a digital camera. The challenge is to mix different flaws ...Missing: machine | Show results with:machine
[32]
Automated defect detection in digital radiography of aerospace ...
Feb 24, 2022 · Small defects can be difficult to distinguish from this noise, exhibiting a small contrast to noise ratio (CNR). To accurately capture small ...<|separator|>
[33]
Assessment Of Image Restoration Techniques To Remote Sensing ...
The resulting image after restoration is compared with original image and their performance has been evaluated in terms of CNR (Contrast-to-noise -ratio) ...
[34]
Assessment of the Impacts of Image Signal-to-Noise Ratios ... - MDPI
Estimation of noise contained within a remote sensing image demonstrated that the noise varied within individual land cover types [40,41,46].Missing: thresholds | Show results with:thresholds
[35]
Image texture, low contrast liver lesion detectability and impact on ...
These model observer demonstrate a stronger correlation with human observer performance compared to conventional CNR or standard Fourier metrics [46,47].
[36]
https://www.mdpi.com/2072-4292/11/22/2603
[37]
Contrast signal-to-noise ratio for image quality assessment
Peak signal-to-noise ratio (PSNR) is commonly used as an objective quality metric in signal processing. However, PSNR correlates poorly with the subjective ...Missing: CNR | Show results with:CNR
[38]
Free Energy Adjusted Peak Signal to Noise Ratio (FEA-PSNR) for ...
Peak signal to noise ratio (PSNR), the de facto universal image quality metric has been widely criticized as having poor correlation with human subjective ...
[39]
(PDF) The Feasibility of Contrast-to-Noise Ratio on Measurements ...
Jul 6, 2020 · Thicker slice images resulted in higher CNR values at 80 kVp with different mAs levels for the 80-MDCT scanner.<|separator|>
[40]
Reducing contrast media and radiation dose in CT angiography at ...
Mar 24, 2025 · CNR B = D CB / D CA CNR A, 1. The CNR is expected to be proportional to the square root of the radiation dose D R , because the image noise σ ...
[41]
Signal-to-Noise Ratio (SNR) in MRI | Factors affecting SNR
Increasing the TR will enhance the SNR, as a higher TR allows the longitudinal magnetization to approach its maximum and generate high signal intensities.
[42]
Ultrasound system for precise neuromodulation of human deep ...
Sep 5, 2025 · This trade-off is also linked to ultrasound frequency selection, as lower frequencies offer better skull penetration but poorer spatial ...
[43]
Assessing the performance of 360∘ ultrasound probes designed for ...
At high receive gains, the CNR remains broadly consistent for the − 5.7 dB target. However, for the − 9.5 dB target there is ...