10.2 Digital Image Characteristics & Exposure Index

The Digital Image as a Numeric Grid

A digital radiograph is a rectangular grid of picture elements, or pixels. The full grid is the matrix, described as rows by columns (for example, 2000 x 2500). Each pixel holds one numeric value representing the signal at that location. The physical area the matrix covers is the field of view (FOV). These three quantities are linked by one relationship you must be able to manipulate:

Pixel size = Field of view / Matrix size

So for a fixed FOV, a larger matrix means smaller pixels and finer detail; for a fixed matrix, a smaller FOV means smaller pixels. The center-to-center distance between adjacent pixels is the pixel pitch; a smaller pitch packs more pixels per millimeter and yields higher spatial resolution.

Spatial Resolution and the Nyquist Limit

Spatial resolution is the ability to distinguish small, closely spaced structures, measured in line pairs per millimeter (lp/mm). In a digital system, spatial resolution is capped by pixel pitch. The highest spatial frequency a detector can record is the Nyquist frequency, equal to 1 / (2 x pixel pitch). Detail finer than the Nyquist limit cannot be resolved and may alias. This is why direct-conversion DR and small-pixel systems resolve finer detail: their sampling is finer. Critically, no amount of post-processing can add spatial resolution the detector never sampled.

Bit Depth, Dynamic Range, and Contrast Resolution

Bit depth is the number of bits used to encode each pixel value, and it sets the number of possible shades of gray: a system stores 2 to the bit-depth power values. A 12-bit system encodes 4,096 shades; a 14-bit system encodes 16,384. Greater bit depth supports finer contrast resolution - the ability to distinguish small differences in signal - and a wider dynamic range, the range of exposures over which the detector produces useful signal. Digital detectors have a very wide, essentially linear dynamic range, far broader than film. That wide latitude is a double-edged sword: it visually forgives exposure errors, which is the root cause of dose creep.

Separate two kinds of resolution. Spatial resolution (detail) is governed mainly by pixel pitch and focal spot; contrast resolution (gray-shade discrimination) is governed by bit depth and noise. Digital systems trade slightly lower spatial resolution than the best film-screen for dramatically better contrast resolution and post-processing flexibility.

DQE, MTF, and SNR

Three performance metrics describe how good a detector is:

• Detective quantum efficiency (DQE) measures how efficiently a receptor converts incident x-ray photons into useful image signal. An ideal detector has a DQE of 1.0; real detectors are lower. Higher DQE means a diagnostic image at lower dose, so DQE is the headline dose-efficiency figure - CsI indirect and a-Se direct panels generally out-perform CR here.
• Modulation transfer function (MTF) describes how faithfully the system reproduces contrast at increasing spatial frequencies. MTF is 1.0 at zero frequency and falls toward zero near the resolution limit; a system with higher MTF at a given lp/mm renders fine detail with more contrast.
• Signal-to-noise ratio (SNR) compares useful signal to random noise. SNR improves when receptor exposure (mAs) is increased appropriately, because more photons reach the detector. Low receptor exposure produces a low-SNR image dominated by quantum mottle (grainy noise).

The Exposure Index (EI)

Because a digital image looks acceptable across a wide exposure range, the technologist cannot judge dose from brightness alone. The exposure index (EI) is a numeric value that reflects the amount of radiation the image receptor actually received in the anatomy of interest. In the standardized IEC/AAPM convention, EI is proportional to receptor exposure - double the exposure, double the EI. (Legacy vendor indicators differ: Fuji's S number is inversely proportional to exposure, while Kodak/Carestream EI and Agfa lgM rise with exposure. ARRT expects you to know that an indicator exists and that it reflects receptor exposure.)

Each exam and body part has a target exposure index (EIt) - the value expected when exposure is correct. Comparing actual EI to target tells you whether you over- or under-exposed. But raw EI numbers are hard to interpret across systems, which is why the deviation index was created.

The Deviation Index (DI)

The deviation index (DI) standardizes exposure feedback across all vendors with one formula:

DI = 10 x log10 (EI / EIt)

DI expresses how far actual receptor exposure sits from target on a logarithmic scale:

A DI of 0 means receptor exposure matches the target. Because the scale is logarithmic, each plus or minus 1 DI is roughly a 25 to 26 percent change, and plus or minus 3 DI corresponds to a factor of two. A DI near 0 (many departments accept about plus or minus 1) is ideal.

Dose Creep - the Central Digital Pitfall

Dose creep is the gradual, unintended rise in patient exposure that happens because over-exposed digital images still look excellent - the system simply rescales them. Without EI/DI monitoring, technologists drift toward higher techniques to guarantee low noise, silently increasing patient dose and violating ALARA. A persistently positive DI flags this drift, while a persistently negative DI (for example, -3) warns of underexposure producing quantum mottle. Watching the DI on every image converts an invisible dose problem into a number you can act on.