Technical Note: The use of DirectDensityTM and dual‐energy CT in the radiation oncology clinic

Abstract Purpose Two new tools available in Radiation Oncology clinics are Dual‐energy CT (DECT) and Siemens’ DirectDensity™ (DD) reconstruction algorithm, which allows scans of any kV setting to use the same calibration. This study demonstrates why DD scans should not be used in combination with DECT and quantifies the magnitude of potential errors in image quality and dose. Methods A CatPhan 504 phantom was scanned with a dual‐pass DECT and reconstructed with many different kernels, including several DD kernels. The HU values of various inserts were measured. The RANDO ® man phantom was also scanned. Bone was contoured and then histograms of the bone HU values were analyzed for Filtered‐Backprojection (FBP) and DD reconstructions of the 80 and 140 kV scans, as well as several virtual, monoenergetic reconstructions generated from FBP and DD reconstructions. “Standard” dose distributions were calculated on several reconstructions of both phantoms for comparison. Results The DD kernel overcorrected the high‐Z material inserts relative to bone, giving an excessively low relative electron density (RED). A unique artifact was observed in the high density inserts of the CatPhan in the monoenergetic scans when utilizing a DD kernel, due to the overcorrection in the DD scan of the material, especially at lower kV. Conclusions While DD and DECT perform as expected when used independently, errors from their combined use were demonstrated. Dose errors from misuse of the DD kernel with DECT post‐processing were as large as 2.5%. The DECT post‐processing was without value because the HU differences between low and high energy were removed by the DD kernel. When using DD and DECT, we recommend the use of a DD reconstruction of the high energy scan for the dose calculation, and use of a FBP filter for the low and high energy scans for the DECT post‐processing.


| INTRODUCTION
Multiple new technologies and imaging processing approaches have recently become available for CT Simulation. Two significant examples of this are DirectDensity™ (DD-Siemens Medical) and Dual Energy CT (DECT). The vendor supplied white paper on DD states the following regarding the use of DD in combination with DECT: "It is technically possible to select DirectDensity™ kernels in reconstructions of Dual Energy scans, but the resulting DirectDensity™ images are not suitable for Dual Energy post-processing." 1 It further states: "Non-natural materials, for example metals and contrast agents like iodine, will decrease accuracy andas with conventional CT imagescan potentially lead to image artifacts." However, the console allows DD kernels to be used on DECT scans. In the list of DECT algorithms on the Siemens Confidence CT 45 Simulator, there is a DD kernel, which could easily be selected without the user being aware of the potential problems involved. To our knowledge there has been no peer reviewed published data characterizing the various types of potential artifacts that may manifest when utilizing DD. We are also unaware of peer reviewed published data demonstrating the potential dosimetric consequences of performing dose calculations on scans using DD in combination with DECT. Here, we endeavor to explore and characterize both of these questions.

1.A | Dual energy CT
In Hounsfield's first paper on CT, he discussed DECT and how various materials can be better differentiated by utilizing DECT postprocessing. 2 The uses of DECT within the realm of Radiation Oncology are still a very active field of research. 3 There are multiple hardware approaches to achieving DECT, each of which has its own strengths and weaknesses. One approach is a dual-source system which has two x-ray tubes 90°apart on the gantry. This approach is able to acquire the low and high energy scans simultaneously.
Because it uses multiple tubes, it is able to use ideal mAs settings for the low energy tube and the high energy tube, and is capable of providing the best spectral differences of all the dual-energy approaches. A dual-source system is much more expensive due to a near doubling of the hardware involved. 4,5 Another approach is fast kV switching, which acquires low and high energy using the same tube in rapid succession, such that every other projection is high or low energy. 6 This approach allows both low and high energy to be acquired in the same geometry and at approximately the same time. Because it uses one tube, the mAs is not ideal for both low and high energy.
Split filter is another approach which uses one tube and filtering of one half of the beam to create the high energy spectra. This approach allows both low and high energy to be acquired simultaneously. However, because the high energy is created solely by filtration, the spectral separation is not ideal. Another approach is to separate the low and high energy photons at the level of the detector. 5,7 Dual-pass DECT uses one tube and two scans separated temporally. As such, it is the most susceptible to motion artifacts. However, it does allow for excellent spectral differences between low and high energy. This approach is also easiest to implement in existing clinics, as it requires no additional hardware.
There are many DECT post-processing techniques which aid diagnostic clinical applications, such as metal artifact reduction, virtual noncontrast scans, material decomposition for higher Z materials relative to specific soft tissues 8,9 and virtual monoenergetic images. 10 Virtual monoenergetic reconstructions are helpful in several ways. A low energy monoenergetic reconstruction provides much greater contrast.
A high energy monoenergetic scan provides low noise and can be used to minimize metal artifacts. Often, the two are blended together to get the best contrast to noise ratio (CNR) for the best tumor visualization.

1.B | CT sim
CT Simulation for radiation oncology imposes multiple demands, including an increased need for high spatial accuracy and HU accuracy, relative to diagnostic CT scanners. 11,12 The HU accuracy is important in Radiation Therapy because the CT scans' HU values are converted to electron density which is then used to calculate the treatment dose. 13 The CT-to-density conversion curve is energy dependent. 14,15 Radiation Oncology traditionally uses one standard energy (typically 120 kVp) at CT SIM because using multiple energies would require multiple CT-density curves which is cumbersome in addition to the potential for misapplication of CT-density curves. 16

1.C | DirectDensity™
Using one energy is not ideal for all patients or treatment sites.
Lower energy scans provide much better soft-tissue contrast. Patient size will also affect which energy is ideal. The larger the patient is, the higher the ideal scan energy to ensure sufficient signal. DirectDensity™ (DD) was developed by Siemens to address this issue of kV energy and CT-density curves. 1,17 DD reconstructs and thresholds for bone; the bone is then forward projected into the sinogram space. The sinogram space representation is corrected for the length of bone along each ray path. Final reconstruction results in a CT that uses relative electron density (RED) units, not HU. This allows any beam energy to use the same CT-density curve. Thus, the CT SIM software solution is capable of scanning with an ideal scan energy rather than always using the same kV settings.
However, one must be careful when acquiring images using DECT protocols and reconstructing the images using DD and performing DECT post-processing. This study aims to quantify the potential image quality and dose implications of ignoring the vendor recommendations by performing DECT processing on DD images.

| MATERIALS AND METHODS
To evaluate the effect of kernel selection on the resulting CT-density curves, a dual energy scan of a CatPhan Model 504 was acquired on a SOMATOM Confidence ® RT Pro (Siemens Healthcare GmbH, Erlangen, Germany). The DECT was acquired using the DE_Abdo- Multislice contours were created on each of the sensitometry inserts in the CTP404 module. The mean HU values within the contours were measured for each reconstruction. These were then plotted on a CT-density curve and reviewed. Particular attention was paid to how the DD reconstructions handled the Teflon and Delrin inserts given the manufacturer warning that such materials could cause artifacts. 1 To evaluate how well the DD algorithm handled bone, a RANDO ® man phantom (Rando) (The Phantom Laboratory, Greenwich, NY, USA) was used to perform additional scans. The same protocol was used for this as was used for the CatPhan. Because the focus was on the effect that DD had on DECT processing, only two kernels were used, one FBP kernel (B30f) and the DD kernel (E30f).
These two data sets then had DECT post-processing to create monoenergetic reconstructions at 40, 50, 70, 100, 120, 140, and 190 keV. A bone contour was created on the B30f 140 kVp scan (not the monoenergetic scan) using a threshold of 85 HU and above.
That generated contour was evaluated to ensure that it accurately represented bone, and was then used to measure the average HU   high density objects have a higher HU on lower energy scans, and low density objects have a lower HU on lower energy scans. The DD scans closely follow the FBP scans for low density objects. This makes sense because the DD algorithm thresholds for bone and then in the conversion to RED from HU it will primarily adjust those voxels, not the low HU voxels. At higher densities with non-naturalbody materials, we see the DD algorithm overcorrect the values. Siemens alerts users that non-natural-body materials, like the high density inserts in the CatPhan, will have decreased accuracy and could lead to artifacts. This overcorrection inverts the now RED values for high density objects relative to their FBP HU values. The DECT processing then accentuates this overcorrection for low energy monoenergetic reconstructions. This is particularly apparent when looking at the Teflon HU/RED values compared to Bone HU/RED values (Fig. 2). This is also what gave rise to these unique image artifacts seen in the high density inserts in Fig. 3. It should also be noted that while this study only saw an overcorrection by DD of non-natural-body dense objects, there is the potential for the DD algorithm to under-correct for other materials.
The Rando phantom was also scanned and reconstructed with DD kernels and then had DECT post-processing. The unique artifacts visible on the CatPhan (Fig. 3) were not visible on any of these reconstructions. As shown in Fig. 2(a), the bone was not overcorrected by DD which is what had led to the artifact on the CatPhan. Given the impacts on dose and image quality, the biggest concern for a clinic that utilizes DD with DECT post-processing appears to be the potential for inaccurate contouring due to artifacts from non-natural-body materials such as iodine contrast or metal implants.  vendor recommendations about their use. If ignored, unique DD/ DECT artifacts can occur. Knowledge of this is particularly important while there is a DD kernel listed in a group of DECT kernels. While the dosimetric impact observed for cases explored here was relatively small, subsequent artifacts from non-naturalbody materials could lead to inaccurate target and OAR delineation, which could subsequently lead to more significant dose distribution errors. If clinics wish to utilize both technologies simultaneously, the DECT processing should only be performed with FBP kernels. The DD reconstruction should be created using DE Rho post-processing to minimize artifacts from non-naturalbody materials.

ACKNOWLEDGMENTS
The authors acknowledge the feedback and information provided by Nilesh Mistry and Guillaume Grousset from Siemens.

CONF LICT OF I NTEREST
There are no conflicts of interest to report.