Low-dose imaging
George Zentai from Varian Medical Systems (Palo Alto, CA) described the development of an amorphous selenium (a-Se) detector for use in mammography. The goal, he explained, is to generate a high-performance a-Se array with increased sensitivity and temperature resistance.
The imager comprises a 200 µm-thick a-Se film on a 5 x 5 cm thin-film-transistor (TFT) array with 50 µm pixels. Zentai and colleagues tested the device performance after temperature cycling between 0 and 70 °C, with 30 hours at each point. After 30 hours at 0 °C, the device exhibited almost no change in dark current and no significant change in sensitivity. After the 70 °C cycle, there was a slight increase in dark current, and practically no change in sensitivity.
Increasing the bias electrical field from 10 to 20 V/µm (thus reducing the energy required to produce an electron-hole pair) increased the sensitivity of the imager by more than 60%. A special blocking structure ensures that increasing the bias voltage did not greatly increase the dark current, which remained below 0.3 pA/mm2 up to 25 V/µm. Zentai also highlighted the low pixel-to-pixel inhomogeneity of less than 5%.
The device exhibited high detective quantum efficiency (DQE) under mammography conditions (28 kVp). The DQE did not decrease significantly at low exposures, down to 23 µGy at a bias field of 20 V/µm. The researchers used the device to image a breast phantom, and could clearly see all microcalcifications in a 5 x 5 cm image.
The team now plans to scale the a-Se device to a larger size and to perform detailed temperature, environmental and bias tests on the larger imager. "This material is an excellent candidate for low-dose mammography imaging, including tomography," Zentai concluded.
Large-area detection
Also looking at applications of amorphous-selenium, Kai Wang from the University of Waterloo in Ontario, Canada, presented details of a new detector architecture for large-area, high-speed digital X-ray imaging.
Wang explained that X-ray detectors can work in two ways: via direct conversion, using a photoconductor such as a-Se; or indirect conversion, using a scintillator to convert the X-rays into visible light that is then detected by photodiodes. Direct conversion offers a simple pixel architecture but requires a high bias voltage, while the latter approach has a more complex pixel architecture.
A promising alternative, Wang suggested, could be to use a lateral a-Se metal-semiconductor-metal (MSM) photodetector overlaid with a CsI(Na) scintillator. This design provides indirect conversion with a simplified pixel architecture. "Our approach is simply to replace the photodiode with a lateral a-Se MSM photodetector for indirect-conversion X-ray imaging," he explained.
Initial tests of device performance revealed a high responsivity and a high-speed photoresponse. The first evaluation of the device's X-ray response showed a signal visible above the noise. "Lateral a-Se MSM detectors can potentially be easily fabricated in a large-area format and integrated with either CMOS or TFT arrays," said Wang. "This extends a-Se to higher-energy applications, including real-time fluoroscopy and flat-panel CT."
Energy resolution
Reporting from Johns Hopkins University (Baltimore, MD), Xiaolan Wang told delegates about her research group's work on a microCT system based on an energy-resolved photon-counting X-ray detector. The detector comprises a pixellated cadmium telluride (CdTe) radiation sensor, with fast ASIC readout.
She explained that the first generation of this device provided useful spectral information but was fraught by severe ring artefacts caused by inter-pixel variation. So last year, the team began development of a second-generation device with more energy thresholds, smoother pixels and improved calibration.
Wang presented results from initial tests using a microCT set-up with a cylindrical PMMA phantom containing various materials (soft-tissue equivalent, bone equivalent and contrasts). "There was a lot of improvement in the second-generation device compared with the first-generation, but it was still not good enough," she noted.
To compensate for remaining inter-pixel threshold variations, the researchers incorporated user-accessible threshold adjustment circuits. This allowed fine tuning of the thresholds for individual pixels. Phantom tests before and after tuning showed that this process improved uniformity and substantially reduced ring artefacts in the reconstructed images. "We hope to combine this with pulse acquisition correction methods to further reduce these artefacts," Wang added.
The team examined the detector's ability to decompose images containing two or more constituents, using a two-cylinder phantom containing a range of materials. Wang noted that the use of more than two energy windows enabled separation of several types of materials, for example: soft-tissue, bone and various contrast agents.