Residency · Residency · Diagnostic Radiology

Digital Radiography and Image Quality

Digital Detector Systems

Computed Radiography (CR)

Computed radiography was the first widely adopted digital x-ray technology and uses photostimulable phosphor (PSP) plates, typically composed of barium fluorohalide doped with europium (BaFBr:Eu2+). When x-rays strike the plate, they create a latent image by trapping electrons in metastable energy states known as F-centers. To read the image, the plate is scanned with a focused red laser beam. This stimulates the trapped electrons to release their stored energy as blue-violet light (a process called stimulated luminescence), with the intensity of light proportional to the x-ray energy originally absorbed at each point. The light is collected by a light guide and photomultiplier tube, then digitized to create the image. CR has the advantage of being compatible with existing cassette-based radiography systems and offers portable cassettes, but its detective quantum efficiency (DQE) is lower than that of direct digital radiography, and the workflow is slower because the plates must be physically transported to a separate reader unit. Ghosting artifacts can also occur if plates are not completely erased between uses.

Direct Digital Radiography (DR)

Indirect Conversion DR

The most common DR technology in clinical use today is indirect conversion DR. These systems use a scintillator layer, made of either cesium iodide (CsI) or gadolinium oxysulfide (Gd2O2S), to convert incoming x-rays into visible light. CsI scintillators are particularly effective because they are grown in a needle-like columnar structure that channels light downward toward the detector elements, minimizing lateral spread and preserving spatial resolution. Beneath the scintillator sits a thin-film transistor (TFT) array with amorphous silicon (a-Si) photodiodes that convert the light into electrical charge. Each pixel in the detector has its own photodiode and TFT switch for readout.

Direct Conversion DR

Direct conversion DR systems take a different approach, using an amorphous selenium (a-Se) photoconductor to convert x-rays directly into electrical charge without an intermediate light step. An electric field applied across the selenium layer directs charge carriers to collection electrodes on a TFT array. Because there is no light-spread step, direct conversion systems offer superior spatial resolution, but their x-ray absorption efficiency is somewhat lower than that of CsI-based systems.

Comparison: CR vs. DR

DR systems offer substantially higher DQE than CR (typically 60-70% versus 25-35% at typical diagnostic energies), meaning they convert incident x-ray photons into useful signal more efficiently. DR also provides faster image availability, with images appearing in seconds rather than the minutes required for CR plate processing. Although DR systems have a higher upfront cost, they offer significant advantages in dose reduction potential and workflow efficiency. As a result, CR is being phased out in many departments in favor of DR.

Image Quality Parameters

Spatial Resolution

Spatial resolution is the ability to distinguish closely spaced small objects and is measured in line pairs per millimeter (lp/mm). In digital systems, it is determined by pixel size, detector element spacing, and focal spot size. Both DR and CR systems typically achieve 2.5 to 5 lp/mm, depending on pixel pitch (100 to 200 micrometers for DR) or laser beam diameter and phosphor particle size (for CR). The modulation transfer function (MTF) provides a more complete description of resolution performance across all spatial frequencies, rather than a single number.

Contrast Resolution

Contrast resolution refers to the ability to distinguish objects with small differences in x-ray attenuation. Digital systems have a dramatically wider dynamic range than the film-screen systems they replaced (roughly 10,000:1 versus 30:1). This wider dynamic range means that over- or underexposed images, which would have required repeat exposures on film, can often be rescued through post-processing. Window and level adjustments allow the display contrast to be optimized for different tissues after the image has already been acquired.

Detective Quantum Efficiency (DQE)

DQE is considered the single most important metric of detector performance. It measures how efficiently a detector converts incident x-ray photons into a useful signal relative to a theoretically perfect detector, expressed as DQE = (SNRout)^2 / (SNRin)^2. A higher DQE means the detector can produce better image quality at a lower dose, or equivalent image quality with less radiation. DQE depends on spatial frequency, exposure level, and beam quality. Typical values are 60-70% for CsI flat-panel DR, 40-55% for gadolinium oxysulfide flat-panel DR, and 25-35% for CR.

Signal-to-Noise Ratio (SNR)

The signal-to-noise ratio (SNR = signal / noise) quantifies how well the useful signal stands out from background noise. In properly exposed digital radiographs, quantum noise (the statistical fluctuation inherent in photon detection, which follows a Poisson distribution) is the dominant noise source. SNR improves with the square root of the number of detected photons, which means that doubling the dose improves SNR by a factor of approximately 1.4 (the square root of 2), not by a factor of 2.

Noise Sources

Three types of noise affect digital radiographs. Quantum noise, arising from the random statistical variation in how many photons are detected at each pixel, is the dominant source. Electronic noise from readout electronics and amplifiers becomes significant only at very low exposures. Structured noise, which is a fixed pattern caused by non-uniformities in the detector elements, is corrected through flat-field calibration performed during routine quality assurance.

Image Processing

Exposure Indicators

Exposure index (EI) systems quantify the detector dose for each acquisition and provide feedback on whether the exposure was adequate. The IEC standard, adopted by major manufacturers, defines an EI proportional to detector dose, with a target EI (EIt) representing the optimal exposure level. The deviation index (DI), calculated as DI = 10 x log10(EI / EIt), tells you how far the actual exposure was from the target. A DI of 0 is optimal; values above +1 suggest overexposure, and values below -1 suggest underexposure. Monitoring the DI is critical for detecting dose creep, a phenomenon in which technologists gradually increase technique factors over time because the wide dynamic range of digital detectors makes overexposed images look acceptable.

Lookup Tables (LUTs) and Histogram Analysis

After an image is acquired, the system analyzes the histogram of pixel values to identify the anatomically relevant region. These values of interest (VOI) are then mapped to the display grayscale using a lookup table (LUT). When the histogram analysis goes wrong, such as when collimation is not correctly detected or a prosthesis is in the field, processing failures can produce images that are inappropriately dark or light. Proper collimation improves histogram analysis accuracy by providing the algorithm with a cleaner dataset.

Image Enhancement

Several post-processing techniques are used to improve image quality. Edge enhancement, also known as unsharp masking, increases the conspicuity of edges and fine detail, though over-enhancement can create halo artifacts at boundaries. Noise reduction algorithms smooth quantum mottle while attempting to preserve structural detail. Multi-frequency processing decomposes the image into different frequency bands that can be optimized independently, allowing simultaneous enhancement of both fine detail and large-area contrast.

Common Digital Radiography Artifacts

Exposure-Related Artifacts

Underexposure produces quantum mottle, giving the image a grainy, noisy appearance with reduced contrast-to-noise ratio. Overexposure may cause saturation or blooming and is usually masked by post-processing, meaning the image looks fine even though the patient received more radiation than necessary. Dose creep is the systematic tendency toward overexposure over time, driven by the fact that overexposed digital images are not obviously degraded the way overexposed films were. Monitoring exposure indices is the primary defense against this problem.

Detector Artifacts

Dead pixels or dead lines result from defective detector elements and require calibration maps to compensate. Ghosting, seen primarily in CR, occurs when a residual image from a prior exposure persists due to incomplete plate erasure. Lag, the DR equivalent, happens when residual electrical charge from a prior exposure carries into the next image. Grid lines can appear if the anti-scatter grid is not properly aligned with the detector array.

Processing Artifacts

Histogram analysis failure, where the system selects the wrong VOI, leads to images that are too dark or too light. Stitching artifacts appear as misregistration in long-length imaging studies such as scoliosis series or lower extremity alignment studies. Moire patterns, appearing as wavy interference lines, occur when the spatial frequency of an anti-scatter grid interferes with the detector's pixel array.

Patient and Positioning Artifacts

Motion blur, external objects overlapping anatomy (clothing, jewelry, tubing), and improper collimation that confuses the histogram analysis algorithm are the most common artifacts in this category.

<image>A side-by-side comparison diagram of indirect DR and direct DR detector architectures. Left panel: indirect conversion showing incoming x-ray photons striking a columnar CsI scintillator layer, light photons channeling down the needle-like columns, reaching an amorphous silicon photodiode array and TFT readout. Right panel: direct conversion showing x-rays striking an amorphous selenium layer with an applied electric field, charge carriers migrating directly to collection electrodes on a TFT array. Labels highlight the key difference: the intermediate light-conversion step in indirect DR versus direct charge collection in direct DR.</image>

<image>A set of four radiographic images demonstrating common digital radiography artifacts: (1) quantum mottle from underexposure showing a grainy, noisy chest radiograph; (2) histogram analysis failure showing an inappropriately dark image with loss of parenchymal detail; (3) ghost artifact on CR showing a faint superimposed image from a prior examination; (4) Moire pattern artifact showing wavy interference lines across a chest image due to grid-detector frequency mismatch. Each image is labeled with the artifact name and cause.</image>

<image>A graph showing Detective Quantum Efficiency (DQE) curves for three detector types: CsI flat-panel DR (highest, reaching approximately 0.65 at low spatial frequencies), gadolinium oxysulfide flat-panel DR (middle, approximately 0.45), and CR with barium fluorohalide (lowest, approximately 0.30). The x-axis shows spatial frequency in lp/mm from 0 to 5, and the y-axis shows DQE from 0 to 0.8. All curves decline with increasing spatial frequency. Annotations explain that higher DQE enables lower patient dose for equivalent image quality.</image>

Clinical Pearls

Exposure indices should be monitored on every study to prevent dose creep. A deviation index persistently above +1 indicates systematic overexposure that needs to be addressed. The wide dynamic range of digital detectors is a double-edged sword: it virtually eliminates repeat exposures due to incorrect technique, but it can silently mask overexposure, leading to unnecessary patient dose over time. CsI-based indirect DR detectors offer the best DQE among current technologies and are the preferred choice for dose-sensitive applications such as pediatric and screening examinations. Proper collimation is not merely good practice for dose reduction; it directly improves image processing by providing the histogram algorithm with a cleaner dataset and more accurate VOI selection. When image quality appears unexpectedly poor, processing failures (histogram errors, wrong body part selected in the system) should be considered before assuming the problem is a technique error. Finally, CR is being phased out in favor of DR in most departments due to DR's superior DQE, faster workflow, and dose reduction potential.

References

• Bushberg JT, et al. The Essential Physics of Medical Imaging, 4th edition
• Seibert JA. "Flat-panel detectors: how good are they?" Pediatric Radiology, 2006
• IEC 62494-1: Standard for Exposure Index in digital radiography
• ACR-AAPM-SIIM Practice Parameter for Digital Radiography
• Defined AAPM Report No. 116: "An Exposure Indicator for Digital Radiography"

Read this lecture as Markdown