Deep Tissue Imaging by Collective Accumulation of Single-Scatterers

Add One

Optical microscopy can only penetrate a few hundred microns into thick tissue, a limit imposed by scattering. High resolution imaging requires single-scattering events, so when we have multiple-scattering from particles above and below the focal plane, the resolution and signal to noise ratio quickly degrade. The thicker the tissue (i.e. the deeper the plane of interest) the more the multiple-scattering events dominate over single scattering. Techniques such as optical coherence tomography (OCT) enhance the penetration depth by rejecting multiple-scattered light using what is effectively a time-of-flight measurement. This works because light that has been scattered multiple times will tend to have travelled further than light that has been scattered only once. However, even with this technique, the penetration depth seldom exceeds 1-2 mm, as some multiple-scattered photons will (by chance) have a time of flight close to that of the single scattered photons. As we try to go deeper, these events will begin to dominate again. Now, a group mainly from Korea University in Seoul have suggested an additional method of discriminating between single and multiple-scattered photons, using a technique they call “collective accumulation of single-scattered waves”.

In their paper1, which was published in the April edition of Nature Photonics, the authors Kang et al. explain that their approach relies on the conservation of wavevector for single scattering. This can be illustrated by considering an object with a single spatial frequency (for example a sinusoidal grating). If we illuminate it at some incidence angle, then light will be scattered at some other angle. The relationship between these two angles depends on the spatial frequency of the grating. Thought of in a different way, these angles tell us the components of the incident and reflected wavevectors that are in the plane of the sample. The output component is simply the input component plus the spatial frequency of the sample. Crucially, this relationship is lost for multiple-scattering events, providing a method of identifying and rejecting them.

To make use of this principle, the authors illuminate a sample at a total of 2500 different angles. For each angle, they collect a holographic image using a technique called off-axis holography. This means that, instead of just collecting the intensity of light from the sample, they measure its amplitude and phase. They then take the 2D Fourier transform of each image, which tells them the spatial frequencies of the images, and hence the in-plane components of the output wavevectors. From this they can form a matrix, telling them the amplitude and phase of each output wavevector as a function of each input wavevector. Then, to determine the amplitude and phase of each spatial frequency of the object, they add together all the matrix elements that have the input/output wavevector relationship that corresponds to that spatial frequency. The final image is then obtained by taking the inverse Fourier transform.

The crucial part of all of this is that the single-scattered contributions add coherently because they are in-phase. The multiple-scattered contributions have essentially randomised phase and so add incoherently. These means that the single scattered events are amplified over the multiple-scattered events by a factor of the square root of the number of illumination angles.

The authors combine this new technique with low-coherence interferometry, which is the principle behind the optical sectioning seen in OCT. This means that only light from a thin plane around the focal depth generates the interference pattern detected in the off-axis holography. The combined technique allows the authors to image as far as 10 mean free paths into tissue. They demonstrate this by studying several targets covered by tissue phantoms and rat brain tissue.

As the authors acknowledge, there are several limitations of the approach. Most obviously, it isn’t compatible with fluorescence imaging, which rules out many potential applications. Secondly, they had to collect thousands of images to end up with just one reconstructed image – making the total acquisition extremely slow. And as a phase-based technique, it will be sensitive to even the smallest of movements during the acquisition, which makes it questionable as an approach for imaging live tissue. Nevertheless, it appears to hold promise as an interesting new technique for pushing beyond the current limits of deep tissue imaging.


  1. Kang, Sungsam, Seungwon Jeong, Wonjun Choi, Hakseok Ko, Taeseok D. Yang, Jang Ho Joo, Jae-Seung Lee, Yong-Sik Lim, Q-Han Park, and Wonshik Choi. “”Imaging deep within a scattering medium using collective accumulation of single-scattered waves.” Nature Photonics 9, no. 4 (2015): 253-258.

Leave a Comment

Your email address will not be published. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>