Single Pixel Camera Takes Images Through Breast Tissue

“Ghost imaging” technique could revolutionise medical imaging, say optical physicists

Back in the Victorian era, one hi-tech method for spotting testicular cancer was to hold a lamp behind the testes and look for shadows caused by the presence of tumours. Victorian doctors used a similar technique to spot breast cancer. Later, they tested different light sources to see how colour influenced the amount of light that was transmitted through human tissue. Visible light, they discovered, is strongly scattered by human tissue.

Since then, most medical imaging has focused on other wavelengths, such as infrared which travels much further through human tissue, and on other techniques, such as magnetic resonance imaging.

Now the tables have turned full circle and attention is turning once again to the possibility of visible light imaging through human tissue. That has become possible thanks to a new generation of techniques that eliminate problems caused by light scattering.

Today, Vicente Durán at Chalmers University of Technology in Sweden and a few pals, demonstrate how to take a clear image of an object hidden behind an otherwise opaque layer of chicken breast. And if that weren’t extraordinary enough, they’ve done it by using a single pixel to record the image.

The new technique, called ghost imaging, is based on a clever mathematical trick. When light from a scene is randomly scattered, it is easy to imagine that the resulting image recorded by a single pixel is entirely random as well. That’s not quite true.

Imagine that the experiment is repeated but this time the light is scattered randomly in a different way and then recorded. Next, the light is scattered randomly in another way and recorded and so on. The result is a set of seemingly random data points, each representing the light field at the point it was recorded by the single pixel.

But although these data points seem to be random, they are not. All these data points have one thing in common: the original scene before it was scattered.

In recent years, mathematicians have worked out how to mine the data from many images taken in this way, to find the correlation between them. When this data mining algorithm finds a correlation, it can then be used to reconstruct the original scene before it was scattered.

Physicists have used this single pixel technique to create 3-D images, 3-D movies and cameras that require no lenses at all.

And now Durán and co have used it for medical imaging. These guys sandwiched the number “25” between two slices of chicken breast, each about three millimetres thick. They then illuminated the chicken breast using a light source randomised by bouncing it off an array of digital micro-mirrors that had been randomly arranged. They collected the light that passed through both slices of chicken breast by focusing it on to a single pixel.

They then changed the array of digital micro-mirrors and repeated the measurement some 500 times.

Having number-crunched the data, the resulting image clearly shows the number 25.

That’s a fascinating result that has significant potential for medical imaging. It is by no means perfect, however. For a start, it works only for high contrast objects. In this case, the number consisted of a mask which allowed light through in a pattern shaped like 25. Medically relevant structures are unlikely to have this kind of high contrast property.

Nevertheless, the technique is still in its infancy and would no doubt delight the Victorian doctors who first experimented with light passing through tissues some 200 years ago.

Ref: arxiv.org/abs/1411.2731 : Imaging At Depth In Tissue With A Single-Pixel Camera

Follow the Physics arXiv Blog on Twitter at @arxivblog, on Facebook and by hitting the Follow button below