Speckle is a grainy, salt-and-pepper texture that appears on virtually every ultrasound image, caused by the way sound waves bounce off tiny structures in tissue and interfere with one another before returning to the transducer. It is not a reflection of real anatomy but an artifact of the imaging physics itself, and it has been recognized as a fundamental challenge in diagnostic ultrasound since the technology’s early days. What makes speckle interesting, though, is that it is not purely a nuisance. The same phenomenon that degrades image quality can, in certain applications, be harnessed as a diagnostic tool.
How Speckle Forms
When an ultrasound transducer sends a pulse into the body, the sound scatters off structures that are smaller than the wavelength of the pulse. These scatterers are everywhere in soft tissue: cell walls, collagen fibers, tiny interfaces between slightly different materials. Each scatterer reflects a small echo back toward the transducer, and because there are many of them within any given region, their echoes overlap. Some arrive in sync and reinforce each other (constructive interference), while others arrive out of step and partly cancel out (destructive interference). The combined pattern of bright and dark spots that results is speckle.1Journal of Medical Imaging and Radiation Sciences. Speckles Suppression Techniques for Ultrasound Images
One important detail: speckle is deterministic for a given arrangement of scatterers. If you hold the transducer perfectly still and nothing in the tissue moves, the speckle pattern stays the same. Change the angle of the sound beam even slightly, or shift the transmitted frequency, and you get a different pattern. This reproducibility is both the reason speckle is so persistent in standard imaging and the key to most strategies for reducing it.
The theoretical groundwork for understanding acoustic speckle was laid in the late 1970s, drawing on parallels with laser speckle in optics. A foundational study presented both theory and experiments confirming how acoustic speckle arises and how it differs from its optical cousin, while also demonstrating that averaging multiple images could reduce the effect.2Ultrasonic Imaging. Acoustic speckle: Theory and experimental analysis Statistically, speckle in a simple imaging scenario follows a Rayleigh distribution, though in real tissue the statistics can be more complex. More general models, such as the Nakagami distribution, have been developed to better capture speckle behavior across different tissue types and imaging conditions.3PubMed. Nakagami-based total variation method for speckle reduction in thyroid ultrasound images
Why Speckle Makes Diagnosis Harder
Speckle degrades ultrasound images in a specific way: it reduces the contrast between a lesion or structure of interest and the surrounding tissue. Imagine trying to spot a slightly darker circle on a background that is itself randomly mottled with bright and dark speckles. Even if the circle has a genuinely different average brightness, the random variation from speckle can mask that difference. Researchers have formalized this idea using measures like lesion signal-to-noise ratio, which quantifies how detectable a lesion is in the presence of speckle.4PubMed. Improvement of lesion detectability by speckle reduction filtering: a quantitative study
Under idealized conditions with no noise other than speckle, alternative imaging approaches such as short-lag spatial coherence (SLSC) imaging perform roughly on par with conventional B-mode for lesion detection, suggesting that speckle alone does not make detection impossible. But the real world adds electronic noise, patient motion, and varying tissue backgrounds, and in those conditions speckle’s masking effect becomes clinically meaningful.5PubMed Central. Lesion detectability in diagnostic ultrasound with short-lag spatial coherence imaging Radiologists reading breast, thyroid, liver, and cardiac ultrasound all contend with speckle that can blur the edges of masses, obscure subtle texture differences, and make small lesions harder to distinguish from background noise.
When Speckle Becomes Useful
Here is where speckle’s story takes an unexpected turn. Because the speckle pattern is tied to the physical arrangement of scatterers, it moves with the tissue. If the heart wall contracts and shifts a fraction of a millimeter, the speckle pattern within that wall shifts by the same amount. Track those speckle patterns from frame to frame and you can measure how the tissue is deforming in real time.
This is the basis of speckle-tracking echocardiography, a technique now endorsed by the American Heart Association for assessing how the left ventricle contracts. Two-dimensional speckle-tracking echocardiography measures heart muscle deformation (myocardial strain) along three directions of motion: longitudinal, circumferential, and radial.6PubMed. Speckle-Tracking Strain Echocardiography for the Assessment of Left Ventricular Structure and Function: A Scientific Statement From the American Heart Association A custom tracking program follows the two-dimensional motion of speckle patterns across successive B-mode frames, and the resulting strain measurements have been validated against both implanted crystal measurements and tagged MRI.7PubMed. Noninvasive myocardial strain measurement by speckle tracking echocardiography: validation against sonomicrometry and tagged magnetic resonance imaging
The practical significance is hard to overstate. Ejection fraction, the traditional measure of how well the heart pumps, can look normal even when the muscle is starting to fail in certain patterns. Global longitudinal strain detected via speckle tracking picks up subtle dysfunction earlier, particularly in patients undergoing chemotherapy for cancer or those with infiltrative heart diseases. In this context the artifact is not something to be removed. It is the signal itself.
Speckle analysis has also been explored for characterizing tissue at a level below what the image can resolve directly. Since the late 1970s, researchers have used first-order speckle statistics to try to distinguish different tissue types or detect pathological changes based on how the speckle pattern’s statistics deviate from expected models. A recent review of this quantitative ultrasound literature noted that while the approach holds promise, relatively little attention has been devoted to validating basic assumptions like how linear the imaging system’s response actually is in practice.8PubMed Central. A Systematized Review of Quantitative Ultrasound Based on First-Order Speckle Statistics
Hardware-Based Speckle Reduction
The oldest and still most widely used approach to reducing speckle relies on a simple principle: if you acquire multiple images of the same region with slightly different speckle patterns, averaging them together smooths out the random variation while preserving the real anatomy underneath. The challenge is getting those different patterns in the first place.
Compounding techniques accomplish this by varying the imaging conditions between frames. In spatial or angular compounding, the transducer steers beams at different angles and then averages the resulting images. Since each angle produces a partially independent speckle pattern, the combined image has less granularity. A motion compounding variant achieves a similar effect by combining partially correlated images acquired as the transducer moves slightly across the skin surface.9PubMed Central. A motion compounding technique for speckle reduction in ultrasound images
Frequency compounding takes a different route. Instead of changing the beam angle, it changes the transmitted frequency. Even modest shifts in center frequency alter the interference pattern enough to produce partially independent speckle. This technique was proposed and quantitatively measured decades ago, with researchers documenting how the speckle pattern decorrelates as the center frequency changes.10PubMed. A quantitative approach to speckle reduction via frequency compounding
Modern systems often combine both approaches. Phantom experiments have shown that when frequency compounding and angular compounding are applied together, the speckle reduction multiplies rather than merely adding up. One experiment measured roughly a ninefold improvement in the ratio of mean signal to speckle variation when both methods were combined, consistent with each method contributing independently.11bioRxiv. Multiplicative frequency and angular speckle reduction in ultrasound imaging Most mid-range and high-end clinical ultrasound machines shipped today have some form of compounding built in, often activated by default.
Software Filters for Speckle Reduction
When hardware compounding is not enough, or when images have already been acquired without it, software-based filters can reduce speckle after the fact. The essential difficulty is telling the difference between speckle (which you want to smooth away) and real edges or textures (which you want to preserve). A filter that is too aggressive blurs everything, destroying the diagnostic detail you are trying to see more clearly.
Anisotropic diffusion filters address this by smoothing the image more aggressively in uniform regions while slowing down or halting the smoothing near edges. The result is that speckle within a region of similar tissue is reduced, while the boundaries between different structures stay relatively sharp.12Science China Technological Sciences. Anisotropic Diffusion Filtering for Ultrasound Speckle Reduction More advanced versions use fractional-order mathematics and combine anisotropic diffusion with total variation approaches, which helps preserve both sharp edges and the subtle, low-contrast boundaries that standard filters tend to erase.13PubMed Central. Phase asymmetry ultrasound despeckling with fractional anisotropic diffusion and total variation
Another family of filters draws on the nonlocal means approach, which was originally designed for removing ordinary camera noise from photographs. The idea is to look across the entire image for patches that resemble the one you are trying to clean up, then average those similar patches together. Adapted to the specific noise characteristics of ultrasound (which are multiplicative rather than additive), nonlocal means filtering has been shown to preserve edges and structural details more accurately than many older methods.14PubMed Central. Nonlocal means-based speckle filtering for ultrasound images
A practical tension runs through all of these filters. Removing speckle can make an image look subjectively cleaner and more pleasing, but if a filter alters the texture or brightness of tissue in a way that mimics or masks pathology, the “improved” image could actually mislead a clinician. This is part of why speckle reduction has not simply replaced raw B-mode imaging in all contexts. Many sonographers and radiologists prefer to keep the original speckle-bearing image available alongside any filtered version.
Deep Learning and Advanced Beamforming
The latest wave of speckle reduction leans on neural networks trained to distinguish real anatomy from speckle noise. A U-Net-based deep learning model, for instance, was trained on images corrupted with synthetic speckle noise at multiple intensity levels and then tested on standard benchmark datasets. The approach consistently outperformed other U-Net variants in preserving image quality as measured by structural similarity and peak signal-to-noise ratio.15PubMed Central. An effective image despeckling and reconstruction approach using U-Net based model and comparative analysis These models learn from examples what speckle looks like and what real tissue features look like, and then suppress the former while keeping the latter. The promise is significant, though questions remain about how well models trained on synthetic noise or phantom images transfer to the full range of clinical scenarios.
On the beamforming side, coherent plane-wave compounding (CPWC) has attracted considerable research interest. Instead of focusing the ultrasound beam line by line, the transducer transmits unfocused plane waves at multiple angles and reconstructs the image computationally. This can dramatically increase frame rates while still compounding across angles to reduce speckle. More sophisticated beamformers improve on basic reconstruction by selectively canceling noise and interference, leading to better resolution and contrast.16PubMed Central. Improving Ultrasound B-Mode Image Quality with Coherent Plane-Wave Compounding Using Adaptive Beamformers Based on Minimum Variance Adaptive weighting strategies that assign different importance to different plane-wave angles have been shown to further improve the final image quality of CPWC.17PubMed Central. Plane wave compounding with adaptive joint coherence factor weighting
Synthetic aperture imaging takes a related approach. A preliminary clinical study in breast ultrasound found that speckle artifacts were observed less frequently in synthetic aperture images compared to conventional ones, with the improvement reaching statistical significance across multiple radiologists reading the same scans.18Academic Radiology. Synthetic aperture imaging in breast ultrasound: a preliminary clinical study The tradeoff was that reflection artifacts appeared more often in the synthetic aperture images, illustrating the familiar theme that eliminating one artifact can sometimes introduce or amplify another.
Speckle Beyond Ultrasound
Speckle is not unique to acoustic imaging. Any wave-based imaging system that uses coherent radiation produces it, including optical coherence tomography (OCT) and laser imaging. In OCT, which uses infrared light to image structures like the retina and skin, speckle arises from the same interference mechanism and creates the same grainy texture. Speckle reduction techniques developed for ultrasound have been adapted for OCT, including approaches based on repeated scanning at slightly different probe-to-surface distances. In imaging of basal cell carcinomas, for example, an OCT speckle reduction method improved image clarity by acquiring multiple scans with the probe repositioned slightly between each.19PubMed. Improved quality of optical coherence tomography imaging of basal cell carcinomas using speckle reduction
Laser speckle contrast imaging (LSCI) flips the script even more dramatically. Rather than trying to eliminate speckle, LSCI deliberately analyzes how fast the speckle pattern fluctuates in order to measure blood flow. When a laser illuminates tissue, moving red blood cells cause the speckle pattern to shift rapidly. If a camera’s exposure time is longer than these fluctuations, the speckle blurs, and the amount of blurring directly reflects the speed of blood flow. Regions with faster flow show more blurring and lower spatial contrast in the speckle pattern.20PubMed Central. Laser Speckle Contrast Imaging of Cerebral Blood Flow
LSCI has become a valuable tool in neuroscience for mapping cerebral blood flow in real time during animal experiments and neurosurgery. Recent work has extended the technique to thicker tissues using a transmissive detection approach, where light passes through the tissue rather than bouncing back from the surface. This variant showed higher sensitivity to blood flow changes at low velocities compared to conventional LSCI, opening the door to deeper microcirculation monitoring.21iScience. In vivo laser speckle contrast imaging of microvascular blood perfusion using a chip-on-tip camera Miniaturized chip-on-tip cameras have been developed to bring LSCI inside the body through endoscopes, with temporal contrast measurements correlating closely with benchtop reference cameras across a range of flow speeds.
What Sonographers and Patients Should Know
If you have ever looked at your own ultrasound image and noticed that the tissue seems grainy or speckled rather than smooth, that texture is almost certainly speckle artifact rather than anything anatomically real. It does not mean the image is bad or the machine is malfunctioning. It is an inherent feature of how ultrasound works, present on every scan from every machine. Higher-frequency transducers, which give better resolution for superficial structures, tend to produce finer speckle, while lower-frequency transducers used for deeper imaging produce coarser speckle. Neither is a sign of a problem.
For sonographers and clinicians, the practical decision is how much speckle reduction to apply. Most modern machines offer compounding and post-processing filters that can be adjusted in real time. Turning them up makes the image look smoother and more aesthetically pleasing, but aggressive filtering can blur subtle tissue boundaries and change the texture patterns that experienced sonographers use to distinguish normal from abnormal tissue. Many practitioners develop a personal preference, sometimes keeping compounding moderate and relying on their trained eye to mentally “look through” the residual speckle. When evaluating a specific lesion or structure, toggling between the filtered and unfiltered views can be informative, because a finding that appears only in one version warrants extra scrutiny.
The field continues to evolve rapidly. Deep learning models, advanced beamforming, and adaptive compounding strategies are gradually moving from research prototypes into commercial systems. The long-term direction points toward smarter processing that preserves clinically relevant texture while removing only the noise component of speckle, but that boundary remains fuzzy and context-dependent. Phantoms made from tissue-mimicking materials remain essential for validating new approaches, since they provide a known ground truth against which to measure whether a filter is truly improving the image or just making it look different.22PubMed Central. Characterization of soft tissue-mimicking materials for ultrasound training phantoms