Over the past several years, radiography has started transitioning from screen-film detectors to digital detectors.  Digital detectors offer several logistical advantages over screen-film.  As the system stores its images digitally, a radiologist may access the images at any workstation or several radiologists may access the images simultaneously.  Digital imaging systems also allow for the use of image processing algorithms, which enhance various anatomical features.  Digital systems also allow the clinician to alter the image contrast post-acquisition, which frees digital systems from the contrast limitations imposed by screen-film systems.

With the flexibility brought about by the transition of medical imaging from analog to digital modalities, there is an opportunity and a challenge to acquire images with the highest quality for rendering diagnosis.  As a first step for measuring image quality, we have developed several methods for evaluating the physical performance of a detector.  Physical characterization typically includes measuring the resolution, noise, and signal to noise performance through the Modulation Transfer Function (MTF), Noise Power Spectrum (NPS), and Detective Quantum Efficiency (DQE), respectively.  We have improved physical characterization methods by developing different software methods for measuring the MTF, exploring the use of different filter materials in MTF and NPS measurements, and comparing several edge devices for MTF measurement.

Another consideration for radiography systems is how it manages x-ray scatter.  Scattered radiation plays an important role in imaging.  For example, up to 95% of photons in chest radiography are scattered photons.  This radiation does not contain any anatomical information and degrades image quality.  Therefore, we have developed methods to measure the magnitude of scatter in various imaging situations, such as chest radiography, and how well a system rejects that scattered radiation.  Recently, we have developed Monte Carlo models of mammography systems to understand how scatter affects the resolution and noise of images.  These models also help us understand how different anti-scatter grids reject scatter.  We have also measured the scatter performance of different grids and slot-scanning technologies to learn how these different technologies reject scattered radiation.

Theoretical MTF of primary radiation, scattered radiation and primary+scatter beam.  The MTFs were calculated using Monte Carlo software developed at RAILabs.  The MTF was calculated using the edge technique by acquiring an image of a tungsten edge with a selenium detector. Scatter is generated through a 6 cm heterogeneous breast (50% adipose/50% glandular) imaged with a 28 kVp Mo/Mo beam.  The scatter leads to a substantial low-frequency drop of the MTF.

Influence of scatter on image quality.  The image on the left represents an image acquired on a full-field digital chest radiography system (120 kVp, 0.02 mSv) while the image on the right was acquired using a slot-scan detector (140 kVp, 0.02 mSv).  The enhanced contrast and reduced noise are brought about by superior scatter rejection properties of the slot-scan system, even though the system has an inherently lower DQE.

Identical regions of a lung image with resolution and noise corresponding to a direct radiography system (left) and an indirect system (right).  We conducted an observer study with several chest radiologists to examine the impact the differing resolution and noise properties of these two detectors have on the detection of lung nodules.