Light scattered in tissue usually ceases to be a useful source of information. But what if, instead of forcing it to travel along a single path, we allowed it to wander a little differently each time? Scientists from ICTER tested this seemingly risky idea and showed that a series of measurements that differ from one another can produce a clearer image than a single frame.
The new method was developed by scientists from the International Centre for Translational Eye Research (ICTER), part of the Institute of Physical Chemistry of the Polish Academy of Sciences. The study’s first author is Klaudia Nowacka-Pieszak, MSc Eng. She worked with Piotr Kasprzycki, PhD; Patricio Espinoza Guevara, MSc; Karol Karnowski, PhD; and Dawid Borycki, PhD, DSc.
The team combined a mathematical model, computer simulations, and experiments using optical coherence tomography (OCT). The researchers wanted to determine whether controlled changes in the way light is scattered could reduce speckle noise and make images less sensitive to selected optical aberrations. The findings are described in the article “Reducing spatial coherence via dynamic scattering media enables aberration and speckle suppression in optical imaging” , published in “Scientific Reports” .
The same object, different distortions
Light travelling through tissue encounters cell membranes, fibres, blood vessels and boundaries between structures with different optical properties. Some rays change direction, while the phase of the light wave is disrupted. As a result, the image may lose sharpness, become distorted or be covered by a grainy texture.
This texture is known as speckle noise. It arises when waves scattered by different parts of a sample overlap. In some places they reinforce one another, while in others they cancel one another out. The resulting image contains a random pattern of bright and dark spots that can obscure fine structures.
Speckle is not ordinary electronic noise; it arises from the properties of the light being used (in this case, coherent laser light). Speckle is intrinsically linked to the use of spatially coherent light. Without coherence, it disappears, but so does the ability to perform interferometric imaging such as OCT, in which coherence is used to select photons that carry structural information.
The method developed by the ICTER team does not try to provide identical conditions for every measurement. It does the opposite: it deliberately changes the way light reaches a stationary object. Each frame therefore contains a slightly different arrangement of speckle and distortions. If successive images are sufficiently independent of one another, they can be averaged. The fixed structure of the object is repeated and preserved. Random artefacts shift position, so they gradually weaken as the data are combined.
“Scattering is usually seen as the enemy of high-resolution imaging. We show that, when it is carefully controlled and made dynamic, it can actually become a useful tool: by changing the way light reaches the sample from one measurement to the next, we can suppress speckle and reduce image distortions. This opens a new way of improving the robustness of optical imaging without relying on increasingly complex optical hardware,” said Klaudia Nowacka-Pieszak, first author of the paper.
A Layer That Sets the Image in Motion
The diversity of measurements was provided by a dynamic scattering medium (DSM), which produced small, time-varying changes in the direction and phase of the light. Each successive measurement therefore shows the same object illuminated by a slightly different light wave. The differences cannot be too large. Excessive scattering would destroy the useful signal and severely blur the image. A balance is needed: the variation must be sufficient to generate independent speckle patterns, but it must not remove information about the object.
The dynamic scattering medium reduces the effective spatial coherence of the light. High spatial coherence is necessary for interferometric measurements, but it also promotes persistent speckle. The dynamic layer changes the phase relationships between successive measurements, allowing the random component of the image to be weakened through averaging.
The authors compared two methods of combining the images. The first was magnitude averaging, which primarily uses information about signal intensity. It is simpler and less sensitive to interferometer instability. The second method was complex averaging, also known as field averaging. It takes both the amplitude and phase of the light wave into account. It can preserve more information, but it requires successive measurements to be aligned precisely. Even slight drift in the system can prevent them from being combined correctly.
Both methods reduced the graininess of the speckle pattern. Magnitude averaging, however, left a bright background, described in the paper as a constant background, or pedestal. This raised the intensity of dark areas and reduced the difference between the bright and dark elements of the target. Complex averaging removed this component more effectively. The background remained darker and fine elements were separated more clearly. Achieving this benefit, however, required phase stabilisation and more advanced data processing.
What did the simulations show?
The first stage of the study involved simulations of 100 independent images. The scientists analysed, among other factors, defocus, astigmatism, and static layers that distorted the phase of the light. In individual images, the speckle contrast K ranged from 0.69 to 0.93. The higher this value, the grainier the image. After magnitude averaging, it fell to between 0.06 and 0.09. This shows that the random speckle pattern was strongly suppressed.
With complex averaging, the numerical value of K could be slightly higher under the strongest distortions. This did not mean that the resulting image was worse. The background became considerably darker, and because K also depends on the mean intensity of the analysed region, its value could rise even though the constant pedestal had been removed.
Michelson contrast, which describes the relationship between bright and dark elements in an image, revealed a clear difference between the methods. With strong defocus and a dynamic phase variation of 0.95 rad, magnitude averaging produced a contrast of approximately 0.51, while complex averaging gave approximately 0.74. With weaker phase variations, the values rose to approximately 0.79 and 0.97, respectively. For astigmatism, the contrast was 0.59 after magnitude averaging and 0.76 after complex averaging. The mechanism was therefore not limited to a single type of distortion.
The mathematical model made it possible to separate two effects of the dynamic scattering medium. Changes in the direction of the light were primarily responsible for blurring. The magnitude of the phase variations and the number of combined images, on the other hand, affected the amount of residual speckle noise and the resulting contrast. This means that the trade-off between image clarity and sharpness can be predicted while the device is still being designed.
Milk as part of the optical system
After the simulations, the team tested the method using OCT imaging. The object was a standard USAF resolution target containing sets of progressively finer lines. Ordinary milk served as the dynamic scattering medium. Its microscopic constituents are constantly moving, so each measurement contained a slightly different speckle pattern.
The system used light with a central wavelength of 795 nm. Its experimentally measured lateral resolution was 4.6 µm. The researchers acquired 100 three-dimensional images, or volumes. The model indicated that the milk particles would need approximately 0.4 s to move far enough to generate a new speckle pattern. In the experiment, however, the researchers used much longer intervals of 60 s. This was a conservative choice intended to ensure that successive images were fully independent.
Such long intervals increased the risk of interferometer drift. A two-stage phase correction was therefore required before complex averaging. After stabilisation, the remaining fluctuations corresponded to changes in the optical path length of approximately 25-50 nm.
For an unobstructed USAF target, the OCT system alone achieved a Michelson contrast of 0.87. After the dynamic scattering medium was added, the contrast was 0.81 with magnitude averaging and 0.84 with complex averaging. Under straightforward imaging conditions, the dynamic layer therefore did not improve an image that was already of high quality. Its purpose is primarily to help when the light encounters additional elements that introduce distortions.
Reducing these disturbances comes at the cost of some image sharpness. The measured edge width increased from 4.6 µm for OCT alone to approximately 7.7 µm after magnitude averaging and 7.5 µm after complex averaging. This is not an unforeseen flaw, but the central trade-off of the method. Dynamic scattering reduces speckle and sensitivity to certain distortions, but it also introduces a predictable degree of blur. The model developed by the authors makes it possible to estimate this cost and select parameters for a specific application.
The toughest test: imaging beneath tissue
The most important part of the experiment concerned situations in which the light encountered an obstacle.
First, the USAF target was covered with transparent adhesive tape, which acted as an irregular layer that distorted the light wave. Without the dynamic scattering medium, the image contained strong, grainy artefacts. Milk was then introduced into the system to serve as the dynamic scattering medium. The researchers recorded 100 images in its presence and then averaged them. This made the lines of the target much clearer, without the need to measure and correct each distortion separately.
The researchers then placed a 250-µm-thick section of mouse brain above the target. The tissue strongly scattered and attenuated the light. In an image acquired using OCT alone, the target was largely obscured by speckle and distortions.
After the sample was immersed in the dynamic medium and 100 volumes were combined, elements of the target became visible again. The resulting image had lower resolution and a weaker signal than a direct image of the target, but the experiment confirmed that the method worked beneath real tissue, rather than only in simulations.
“For biological imaging, this is especially interesting because real tissue is never an ideal optical system. If we can make imaging less sensitive to the distortions introduced by tissue itself, we may be able to see structures that would otherwise be much harder to resolve,” said Klaudia Nowacka-Pieszak.
What does this mean?
At this stage, the method is a laboratory proof of concept. The study involved stationary objects and ex vivo tissue. It did not involve patients or imaging of a living eye. It is therefore too early to say that the method will improve the detection of any particular disease.
Its potential is nevertheless considerable. In the future, the method could help in the development of more robust OCT systems. This may be particularly important in ophthalmology, where image quality is affected by the properties of the cornea, lens and vitreous body. The mechanism could also be applied in biological microscopy, digital holography and other techniques that use coherent light.
The dynamic scattering medium does not correct each distortion separately. Instead, it causes the effect of the distortions to vary between measurements so that it can be reduced through averaging. In a future imaging system, the dynamic medium could be enclosed in a small chamber with controlled flow or integrated into a microscope objective.
Such a solution would not require a spatial light modulator, a deformable mirror or a separate wavefront sensor. This does not yet mean that a finished device would be less expensive. The results do, however, indicate a possible route towards more compact designs containing fewer active components.
“In the longer term, we hope this approach can inspire new optical designs in which scattering is not simply something to remove, but something that can be deliberately controlled and used to improve the robustness of imaging,” explained Klaudia Nowacka-Pieszak.
Imaging Living Tissue Is Still a Long Way Off
The method has clear limitations. It cannot recover information that has been completely lost. According to the authors’ model, averaging ceases to help when the distortions are so strong that no discernible signal from the object remains in the individual images.
Complex averaging also requires phase stability. Overly aggressive correction can remove not only disturbances but also part of the useful signal. When this happens, its advantage over the simpler magnitude-averaging approach diminishes.
Speed is the greatest obstacle to applying the method in vivo. In the experiment, successive volumes were recorded at intervals of 60 s. Such a protocol is suitable for laboratory work, but not for imaging a moving eye or other living tissue. A future device would require a much faster dynamic medium and a substantially shorter acquisition time.
“The mathematical model was just as important to us as the experiment itself. It showed that reducing speckle noise comes at a specific cost in the form of partial image blur, but it also allowed us to predict that cost. This means that the system parameters can be selected deliberately, depending on whether sharpness, contrast or robustness to imaging disturbances is more important in a given application,” said Piotr Kasprzycki, PhD, co-author of the paper.
The study’s central conclusion is simple: a single perfect image is not always necessary. Multiple measurements can be collected, each affected by disturbances in a different way, and their differences can then be used to reduce noise. Scattering will still destroy the image if it is too strong and uncontrolled. When it can be controlled, however, it ceases to be only a problem and becomes part of the solution.
Klaudia Nowacka-Pieszak, Piotr Kasprzycki, Patricio Espinoza Guevara, Karol Karnowski, Dawid Borycki (2026). Reducing spatial coherence via dynamic scattering media enables aberration and speckle suppression in optical imaging . Scientific Reports .
DOI: https://doi.org/10.1038/s41598-026-60563-1
Author: Scientific Editor Marcin Powęska
Scientific Reports
Experimental study
Cells
Reducing spatial coherence via dynamic scattering media enables aberration and speckle suppression in optical imaging
7-Jul-2026