Dual Energy Radiography Acquisition and Processing

A major limitation of projection radiography is the projection of the three-dimensional patient volume and anatomy on a two-dimensional image plane. In chest imaging, for example, the bony structure of the ribs, clavicle, etc. will often hide subtle soft-tissue lesions in the lung because of anatomical overlap caused by the projection process. Removing the bony structure, therefore, might aid in the visualization of otherwise undetectable lesions; similarly, removing the soft-tissue components and emphasizing the bony structures might allow the discrimination of soft versus calcified lesions. The higher differential attenuation of bones as a function of energy compared to soft tissue allows the ability to decompose two images taken at different x-ray energies into tissue-selective representation of the anatomy, namely “soft-tissue only” and “bone-only” images. 

As the above graph shows, two images of the same object acquired with two distinct x-ray energy beams will exhibit different attenuation characteristics. Note the much greater attenuation of bone at low energies, and similar attenuation of bone and soft tissue at higher energies. The x-axis represents mono-energetic photon energy. Effective energy of a spectrum is equal to the energy of a mono-energetic x-ray photon with the same overall attenuation. Shown are typical x-ray “effective energies” for spectra generated at 60 kVp (green) and 120 kVp (red). If two x-ray images are acquired at these energies, one can “weight” one image relative to the other that when subtracted, will null the signal due to either bone or soft tissues, depending on the weighting factors.

Illustrated in the figure above are two images acquired at low energy and high energy using a Computed Radiography dual-energy system (more about the technology subsequently). Note the higher bone-tissue contrast on the left, which represents the “low energy” image. Below each image is a stylized rendition of the relative attenuation of soft tissue and bone in the low energy image (left) and the high energy image (right), which illustrates the larger overall bone signal (8 units in the low energy image, 4 units in the high energy image), and less energy-dependent soft tissue signals of lower signal in the low and high energy images, respectively.

Dual energy processing involves weighting each image according to the desire to null the signal due to bone for a “soft-tissue only” image, or to null the signal due to soft tissue for a “bone only” image. In the stylized illustration shown below, to remove bone requires that the signal due to bone be zeroed out. This can be achieved by multiplying the high energy image by 2 and the low energy image by 1, subtracting the weighted high from the weighted low image, and scaling the residual tissue signal over a range to produce a tissue-only image, as shown in the illustration below.

Similarly, to remove soft tissue requires that the signal due to soft tissue be zeroed out. This can be achieved by multiplying the low energy image by 2 and the low energy image by 3, subtracting the weighted low energy image from the weighted high energy image, and scaling the residual bone signal over a range to produce a bone-only image, as shown in the illustration below.

How is dual-energy radiography performed?

Currently there are two clinical systems available for dual-energy radiography

  • One specialized radiography system employs “passive” photostimulable storage phosphor imaging plates to acquire two images simultaneously. The imaging plates are stacked, geometrically aligned, and separated by a copper filter, which preferentially absorbs lower x-ray energies. A low energy image (front imaging plate) and high energy image (back imaging plate) are acquired with a single kVp x-ray beam. This is illustrated below.

With this technology, the low and high energy images are acquired simultaneously, essentially eliminating any artifacts due to patient motion, but the energy separation between the two image pairs is small, which results in a relatively low SNR for the tissue and bone images at typical patient exposures.

  • Another dual-energy capable radiography system uses an “active” flat-panel detector, where a low energy image (~60 kVp) is initially acquired, rapidly read out and detector reset, followed by a high energy image (~120 kVp) immediately afterward. This is illustrated below.

This dual-energy method uses a flat-panel detector with a fast readout capability, enabling the use of two separate x-ray beams producing large differences in the effective energy of the beams. The first acquisition occurs with the high (120 kVp) energy beam, then by image readout of the TFT flat-panel array, followed immediately by the low (60 kVp) energy beam acquisition, then by image readout. Energy separation is large, allowing for a relatively high SNR for a given patient exposure. However, because of the delay time required for acquiring two images and the readout time of the flat-panel array, the difference in time between the images often results in motion artifacts due to involuntary and voluntary patient motion over the ~230 ms acquisition time for both images and the readout time between images.

X-ray beam spectra for the dual-energy approaches
Depicted below are the typical x-ray beam spectra used for the two dual energy approaches described above.  
On the left is the single x-ray beam acquisition at 120 kVp using the CR dual detector / filter sandwich. Advantages include simpler x-ray operation and no patient motion. Disadvantages include relatively poor energy separation and lower detection efficiency (compared to the flat-panel detector).

On the right is the dual x-ray beam acquisition at 60 kVp and 120 kVp using the single, fast readout thin-film-transistor array detector. Advantages include better energy separation and better image quality at the same dose compared to the CR detector sandwich. Disadvantages include potential for patient motion artifacts, and the need for a more complicated system interface and more costly system.

Example dual-energy images

Dual energy radiography can assist in the differential diagnosis of
soft versus calcified lesions. In the dual-energy image acquisition using a CR sandwich detector pair shown in the figure below, it is clear that the lesions on the composite image (left) are calcified as seen in the bone-only image. For the flat-panel dual energy image acquisition, note the clearly visible soft tissue lesion in the tissue-only image. In retrospect, the lesion is reasonably easy to detect in the conventional composite image, but clearly the ribs project anatomical “noise” that interferes with the conspicuity of the relatively large lesion in the pulmonary tissues. Also of note is the cardiac motion visible in the bone-only image, where soft-tissue .

Dual energy image gallery

Dual energy images can be manipulated with different grayscale presentations like any other digital image. The next several sets of images demonstrate a variety of composite, tissue-only and bone-only images of the postero-anterior chest projection. Many of the images contain soft-tissue and calcified pulmonary lesions, and there are examples of energy-sensitive elements that project specific signals in either the tissue-only or bone-only images.

Example image sets illustrating the value of tissue-only and bone-only image presentations:

On the left is the conventional single-energy image; in the middle is the bone-subtracted “soft-tissue only” image; on the right is the soft-tissue subtracted “bone only” image. In the upper image set, the soft tissue pulmonary lesions are clearly visible in both the composite and soft-tissue only images, although there are other lesions in the heart region in the bone-only image that indicates the presence of calcium-containing lesions. These might be due to calcium deposits in the vasculature.   In the lower image set, there are no readily apparent lesions, but surgical clips are readily visualized on the bone only image.

Image examples demonstrating grayscale manipulation. Of interest in these images is the presence of silicone in the breasts of this patient, and a soft-tissue lesion in the left upper quadrant of the lung.

For flat-panel dual energy detector systems, motion can be a problem when the x-ray system is energized during the rapid contraction period of the heart (end systole). Most motion artifacts appear in and around the cardiac anatomy, and often in the pulmonary architecture and diaphragm area.

Dual energy radiography can often improve the diagnostic information content and sensitivity of projection radiography in many situations by removing the anatomic shadows that can mask soft tissue lesions. In the example below, the composite image (left) does not show evidence of a pulmonary lesion, which is hidden by the overlying rib signal. By selectively removing the bone signal, a soft tissue lesion is clearly visible in the “tissue only” image (middle). A subsequent CT scan for needle biopsy illustrates the cross-sectional volume of the lesion and the correlation to the dual-energy image. The value of cross-sectional imaging is nicely demonstrated, although with higher radiation dose and much higher costs.

Summary, Dual Energy Imaging

Dual energy imaging provides the capability of selectively imaging two clinically relevant materials, namely soft tissue and bone tissue. Energy dependent differences of bone versus soft tissue are used to eliminate one tissue or the other, determined by energy spectra differences used for acquiring independent images. Elimination of structured anatomy (noise) is the major benefit of the technique. 

Two major methods include a CR sandwich (passive detector) with inter-detector filter (copper) to achieve low (front) and high (back) image pairs. Attributes of the CR method is single-shot, no motion, but poor energy separation, resulting in noisy images for low dose typical of a chest x-ray examination. DR (using a fast flat-panel readout detector) acquires images with different kVp (usually ~60 kVp and ~120 kVp) to produce two distinct images with good energy separation but poor temporal response time, allowing motion artifacts to sometimes be a significant problem with image quality. Characteristics of the DR dual energy images are the possibility of involuntary patient motion (particularly the heart), but good energy separation and superior image noise properties for a given patient dose, resulting in images with excellent signal to noise ratio.