Rays of hope from active sensors
DECEMBER 17, 2007--Huge improvements in medical x-ray and gamma ray imaging are expected to become possible in the near term, thanks to the development of large-area, wafer-sized solid-state sensor arrays.
DECEMBER 17, 2007--Huge improvements in medical x-ray and gamma ray imaging are expected to become possible in the near term, thanks to the development of large-area, wafer-sized solid-state sensor arrays. The breakthrough comes out of a UK research-council-sponsored initiative to develop new kinds of imaging systems--and the medical research fraternity seems to have been first to grasp the opportunities.
X-ray images of items can now be obtained by using phase-contrast techniques that show up features in delicate samples that are invisible using conventional methods. As a result, detection of the early stages of breast cancer is likely to become easier and more reliable, and many other branches of scientific research greatly assisted, since the fundamental technology works over most of the electromagnetic spectrum.
Known as multidimensional integrated intelligent imaging (MI-3), the technology has been developed by an 11-member consortium under the leadership of Nigel Allinson at the department of electronic and electrical engineering, University of Sheffield. The project has £4.4 million of funding under the Research Council Basic Technology program.
At a recent seminar, Allinson explained that the project involves having arrays of photodiodes produced in CMOS. This allows electronic circuitry to be placed alongside the sensing elements, so that they become active pixels. This is the alternative solid-state way of capturing solid-state images to CCDs--the technology used in most digital cameras--although active pixel arrays are now the most common devices used in mobile phone cameras.
With on-chip processing, it becomes possible to boost sensitivity, resolve images at finer scales than the sizes of individual pixels and undertake various kinds of image enhancement. The MI-3 sensors are intended to cover a spectral range that extends from ionizing particles to infrared.
"Using the on-chip intelligence down to pixel level, we can read out just the bits of interest--when they are of interest," states Allinson. He confirms that the sensors are ultimately intended to observe events occurring over time scales ranging from femtoseconds to hours.
The program has already produced a number of usable devices. The first, called StarTracker, is based on a pre-existing design developed by the Rutherford Appleton Laboratory with 525 x 525 pixels, each 25 μm square. This has led on to another version called the Vanilla Sensor, which has 512 x 512 pixels, each 25 μm square. Fill factor is 85%, and the device can be read to 12-bits resolution. Eureka was shown an image of a mouse brain derived from radiation from tritium. It took 28 days to obtain using conventional photographic emulsion, but Kevin Wells and Jorge Cabello at the University of Surrey took only a day and a half to produce using a back-thinned Vanilla sensor.
This has led on to a large-area sensor with 1400 x 1400, 40-μm pixels, giving an active area of 58 x 58 mm. Meanwhile Renato Turchetta from the Rutherford Appleton Laboratory describes future wafer-scale integrated sensors as a possibility.
One of the big attractions of having a large area image sensor is that it can be made to work with x-rays, which are extremely difficult to focus. Says Robert Speller, from the department of medical physics and bioengineering at University College, London: "When using radiation to treat cancers, one wants to deliver a uniform dose, while saving the surrounding tissues." The tumor may be moving as the patient breathes, he points out, so it would be desirable to be able to image the x-rays coming through the patient in real time, so that they could be targeted on the tumor and nowhere else.
He says it is also possible to image the distribution of single gamma ray photon emitters, such as radioactive technetium in tissue, by using a Vanilla sensor with cesium iodide needles deposited in holes in front of the sensing layer. The cesium iodide then scintillates when it is struck by the gamma rays.
Due to the extreme sensitivity of the sensors, it is possible to use them to study tissue, or anything else being x-rayed, in two novel ways. The first of these is tissue diffraction imaging: looking at diffracted x-rays instead of those that have passed straight through. "In many instances," says Speller, "the disease cannot be seen [using conventional technology]. What you see in a conventional mammographic image is rarely the tumor." But by using a diffraction method, it is very clear there is distinction, he adds. His team has been conducting experiments using a 'Pink beam' and Vanilla sensors.
Another technique the team has been working with is x-ray phase contrast imaging--or xPCi. By this method, a conventional x-ray source with 'Coded apertures' can reveal x-ray images of the internals of a common wasp that are quite impossible to see in a conventional x-ray image.
"This is proof of concept," says team member Alessandro Olivo. "We have demonstrated it is feasible. We have built the setup and compared simulation and experiment. What we need now is a good grant to build a full-sized prototype. If you use it to look at nylon fiber 300 μm in diameter, you don't even see it in a normal x-ray image--the contrast is less than 1% but phase effects give a contrast of 24%." The coded apertures, he says, are arrays of L-shaped slots, 20 μm in each direction.