Evaluation of Direct Density Algorithm

White Paper Evaluation of the Direct Density Algorithm for Energy‐Independent Radiotherapy Treatment Planning Alisha Shutler1, Abhirup Sarkar1, Guillaume Grousset2, Jainil Shah2, Firas Mourtada1 1Radiation Oncology Dept., Helen F. Graham Cancer Center, Christiana Care, Newark, DE 2Siemens Healthineers USA, Malvern, PA 7 siemens-healthineers.us SIEMENS Healthineers White Paper · Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning Abstract Purpose: The commonly accepted practice in RT is to image all patients at 120kVp in order to avoid potential errors with the energy dependent electron/mass density calibration curve. While this is safe practice, the disadvantage is in missing potential superior soft tissue contrast when you vary the tube voltage. A novel Direct Density™ (DD) algorithm is available to allow use of energy- independent CT# to density, simplifying clinical workflow with one calibration curve. In order to commission DD for our clinic, we compared the dose calculated from DD reconstructed CT images at a variety of tube potentials to doses produced using the standard 120kVp images. Methods: Four different phantom studies were conducted. Two with tissue equivalent slabs (homogenous solid water and heterogeneous using ICRU slabs (solid water, bone, and lung equivalent slabs)); and two using thorax Rando phantoms. Scans were performed using a standard reconstruction at 120 kVp and a DD reconstruction for differing kVp (70 – 140kVp) on a SOMATOM Definition Edge (Siemens GMBH, Forchheim, Germany). Two distinct CT density curves were implemented in the treatment planning system (RaystationV9) to read both standard and DD images. Average CT numbers for each ROI were recorded. Point doses were calculated and measured for 200 MU AP plans at 6, 10, and 15 MV, and dose differences were compared. The Rando phantoms were scanned using both kernels at 120kVp, and a VMAT plan was simulated on each. DVH plots were created for assessment. Results: In all instances, computed DD doses were nearly identical to the standard kernel dose. Point dose measurements differed by ≤1%. The largest difference was for the 70kVp AP plan, producing dosimetric error of around 3cGy. VMAT plans showed negligible differences. Conclusions: With an appropriate CT density curve, DD reconstruction algorithm is as accurate as standard algorithms at dose prediction, but allows the flexibility of using variable kVp to improve image quality for certain tissues. Alisha Shutler, M.S. Firas Mourtada, Guillaume Grousset, Ph.D. Medical Physics Resident M.S.E.,Ph.D., DABR, FAAPM Advanced Therapy - Radiation Radiation Oncology, Chief of Clinical Physics Oncology KOL Engagement Helen F. Graham Cancer Center Radiation Oncology, Siemens Healthineers & Research Institute Helen F. Graham Cancer Center 40 Liberty Boulevard, 4701 Ogletown-Stanton Rd & Research Institute Malvern, PA 19355 Christiana Care Health System 4701 Ogletown-Stanton Rd Newark, DE 19713 Christiana Care Health System Newark, DE 19713 Abhirup Sarkar, M.S., DABR Tel: 302-623-4691 Jainil Shah, Ph.D. Radiation Oncology, Email: R&D Collaborations Manager– CT Helen F. Graham Cancer Center [email protected] for Radiation Oncology, & Research Institute Siemens Healthineers 4701 Ogletown-Stanton Rd 221 Gregson Drive Christiana Care Health System Cary, NC 27511 Newark, DE 19713 2 Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning · White Paper Contents Introduction Brief Background: What is Direct Density? Implementation: How can my TPS read RED values via CT#s?? Creating the CT-Mass Density Curve Verification: Does the Direct Density kernel work? > Do the CT #s remain constant? > Thorax phantoms evaluation Discussion Conclusions 3 White Paper · Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning Introduction Siemens Healthineers has recently introduced DirectDensity™ (DD), a novel technique that allows the reconstruction of images acquired at any kV to be directly interpreted as electron density images, thus eliminating the need to perform a Hounsfield Unit (HU) to electron density calibration. While the principles behind DD have been summarized in an earlier white paper1, incorporating a new technology in the clinic can often be challenging and time consuming. While as a community we continue to explore how best to integrate the new technology in the clinic, the goal of this work is to establish a step-by-step methodology to assist the successful implementation of DD in clinical routine. In this paper, we describe the commissioning process at our clinic and present on the dose calculated from DD reconstructed CT images at a variety of tube potentials to doses produced using the 120kVp reconstructed images using the standard filtered back projection (FBP) algorithm. Brief Background What is Direct Density? Direct Density is a reconstruction algorithm/kernel (not to be confused with Siemen’s Dual Energy scanning protocol) that provides HUs scaled to the Relative Electron Density. (These scaled HU values will be referred to as CT #s.) The CT #s are energy independent, meaning that any scan, at any energy, will produce values that give an accurate representation of the true relative electron density or relative mass density (RD or relative density) of the material. This can be especially useful for differentiation of soft tissues that would benefit from lower energy scans, or scans requiring the use of higher energies for harder density materials; no matter the energy used, by selecting the direct density reconstruction, the RD value doesn’t vary with energy. Since it represents a physical property of the materials, the CT #s, which directly then represent the RD, will remain the same. 4 Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning · White Paper Implementation How can my TPS read RD values via CT#s? CT #s are simply RD values that have been scaled so that they resemble traditional HUs: RD = CT # + 1000 1000 In modern treatment planning systems (TPS), the HU to mass density function is usually required to perform dose calculations. Similarly, for DD implementation, CT#s to mass density function will be required and then inputted into the TPS. This can be accomplished by scanning the CT Density phantom, as usually done now in the clinic, with known mass density plugs (it is recommended that scans performed using these phantoms use plugs that range in density from very low to very high, or near zero to roughly 3-4 g/cm3). After the CT acquisition and applying the DD recon kernel Sd-40, each plug is contoured as a Volume-of-Interest (VOI) and then a table that relates the average CT #s for each VOI and its mass density can be created. This table can then be entered into the TPS. Relative electron density DirectDensity image value 0.000 -1000 1.000 0 4.072 3072 Table 1 RED to CT#s values provided by Siemens Healthineers 5 White Paper · Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning Creating the CT‐ Mass Density Curve A high quality CT Density Phantom, such as CIRS (Norfolk, VA, U.S.A.) and Gammex (now Sun Nuclear, Middleton, WI, U.S.A.) should be used. These phantoms are composed of tissue-equivalent materials with a variety of “plugs,” or different density materials, ranging in composition from lung to dense bone or even titanium, duplicated around both an inner and an outer ring. To verify that the Direct Density kernel mapped all tissue - --- densities to the appropriate mass density, the density phantom was scanned at 70, 80, 100, 120, and 140 kVp, --- respectively, using a standard, clinical acquisition protocol of Br38 and then reconstructed using the DD Sd40 kernel. For comparison, the density phantom was also scanned with conventional 120kVp protocol and reconstructed using a standard FBP kernel (Br38). VOIs were created for each plug type, using a single ROI for both the inner plug Figure 1 CIRS model 62 density phantom shown with water plug in and outer plug of a given tissue equivalent materials (e.g. center (syringe). Titanium plug not pictured. two “exhale lung” plugs both were contoured under the same VOI name). The ROIs were expanded volumetrically to include as much of the plug as possible without taking the VOI to any edge, resulting in volumes of around 20 cm3 for all but the smallest plugs. Average CT #s were noted for each ROI, per energy, and plotted in a table with their physical mass density provided by the vendor of the CT density phantom. To ensure that the TPS is reading the materials densities properly, plots were constructed from the CIRS CT density phantom and compared to that provided by Siemens Healthineers’ literature (Fig. 3). In the treatment planning system (TPS), in order to have a single mass density curve for all energies, the new CT‐ Mass Density curve was created by entering the average CT# over the ROI produced by all Direct Density scans (70 – 140 kVp) for each material plug. Figure 2 Contoured ROIs for each plug on the CT density phantom 6 Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning · White Paper Sd40 Sd40 Sd40 Sd40 Sd40 Physical 70 kVp 80 kVp 100 kVp 120 kVp 140 kVp Sd40 AVG Br38 AVG CT# CT# CT# CT# Density CT# CT# CT# (g/cm3) Air -955 -955 -955 -955 -955 -955.0 -955.0 0.00121 Lung inhale -763.7 -781.4 -771.3 -782.1 -778.8 -775.5 -780.0 0.195 Lung exhale -456.2 -474.6 -470.6 -477.4 -476.3 -471.0 -478.0 0.495 Adipose -92.0 -80.0 -67.3 -62.2 -57.3 -71.8 -61.0 0.967 Breast -62.6 -46.4 -44.5 -36.3 -34.4 -44.8 -36.0 0.991 Water -21.6 -14.7 -5.6 -6.6 -3.05 -10.3 -13.0 1 Muscle 9.7 23.49 34.9 42.1 41.9 30.4 40.0 1.062 Liver 25.3 38.17 47.7 51.3 52.5 43.0 53.0 1.071 Trabecular Bone 116.4 123 114.52 118.65 116.2 117.7 208.0 1.161 Dense Bone 413.1 425.6 439.7 464.6 466.2 441.8 836.0 1.609 Titanium 2907 2912 2918 2941 2953 2926.2 3072.0 4.51 Table 2 Average CT# per VOI per energy for the CIRS CT Density Phantom reconstructed using both Direct Density (Sd40) and a standard, Br38 kernel used on a 120 kVp image 2.5 2.0 Standard 120 kVp Kernel Average DD Kernel Mass 1.5 70 kVp DD Density A 80 kVp DD (g/cm3) * 1.0 100 kVp DD * 120 kVp DD 140 kVp DD 0.5 The plotted lines are not visible because of the small difference 0 -1500.0 -1000.0 -500.0 0.0 500.0 1000.0 CT # Figure 3 CT # to mass density curve taken using Sd40 (Direct Density) scans at various energies and comparing this curve to the traditional density curve done at 120 kVp for a standard, Br38, reconstruction kernel 7 White Paper · Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning Verification Is the Direct Density kernel accurate for dose calculation? Several material slabs of different densities were 1. Homogenous Phantom assembled, scanned using the Siemens Healthineers A TPS plan was created in the TPS (Raystation V9A, EDGE CT scanner in our radiation oncology department, RaySearch, Stockholm, Sweden) to deliver 200 MUs to and reconstructed using the DD kernel. The first setup the homogenous slab configuration. Absorbed dose was for a homogenous phantom of solid water only. The measurements were made and compared to the second setup was for a heterogeneous phantom using a predicted dose from the TPS. Table 3 depicts the CT#s mixture of solid water, lung, and bone tissue-equivalents. for the homogenous phantom using both version of syngo.via VB10 and VB20, indicating the importance of recommission when the software version has changed. Table 4 depicts the dosimetric point dose results for the standard 120 kVp Br38 kernel and the Sd40 DD kernels. Point dose Solid 12 cm Water 30 cm Figure 4 Homogenous slab setup 8 Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning · White Paper syngo.via VB20 CT# Br38_120kVp 23.73 Sd40_120kVp 21.97 Sd40_70kVp 21.01 Sd40_80kVp 22.96 Sd40_100kVp 22.75 Sd40_140kVp 21.44 Sd40 Ave. CT# 22.03 %CV 3.8% *Data could not be acquired retrospectively Table 3 Average CT #s reported by Raystation TPS for the homogenous solid water phantom for syngo.via VB20. %CV is percent coefficient of variation 6 MV Dose (cGy) 10 MV Dose (cGy) 15 MV Dose (cGy) Br38_120kVp 165 175 179 Sd40_120kVp 165 175 179 Sd40_70kVp 165 175 179 Sd40_80kVp 165 175 179 Sd40_100kVp 165 175 179 Sd40_140kVp 165 175 179 Average Dose 165 175 179 %CV 0% 0% 0% Table 4 Predicted Dose from 200 MUs at 100 SSD for the Homogenous Phantom for photon energy of 6, 10, and 15 MV, respectively. All TPS results are calculated for the syngo.via VB20. Ave Dose is for Br38 and Sd40 kernels. %CV is percent coefficient of variation 9 White Paper · Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning 2. Heterogenous Phantom A TPS plan was created in the Raystation Solid 5 cm (V9A, RaySearch, Stockholm, Sweden) Water to deliver 200 MUs to the heterogenous Bone 1 cm slab configuration, consisting of ICRU slabs of solid water, bone, and lung Lung 5 cm (Fig 5). Absorbed dose measurements were made and compared to the Bone Point dose 1.5 predicted dose from the TPS. Table 4 cm depicts the CT#s for the heterogenous Solid 7 cm phantom using syngo.via VB20. Table 5 Water depicts the dosimetric point dose results for the standard 120 kVp Br38 kernel and the Sd40 DD kernels. 30 cm Figure 5 Heterogenous slab configuration Table 6 shows the ion chamber measurements to those calculated by Raystation for both phantoms, for the three photon energies used in our clinic. Solid Water Solid Water–Top Solid Water–Bottom Lung Bone Bone–Top Bone–Bottom CT# CT# CT# CT# CT# CT# CT# Br38_120kVp 51.4 44.7 54.9 -653.5 438.9 262.8 447.7 Sd40_120kVp 46.6 40.9 49.8 -657.9 279.9 150.2 288.3 Sd40_70kVp 46.8 35.6 53.0 -625.1 199.7 108.3 206.1 Sd40_80kVp 49.6 39.9 55.1 -637.9 218.5 122.0 224.6 Sd40_100kVp 45.6 35.4 51.8 -649.2 249.0 136.9 256.4 Sd40_140kVp 44.1 35.97 49.1 -660.5 296.2 161.2 306.1 Sd40 Ave. CT# 46.5 37.5 58.1 -646.1 248.6 135.7 256.3 %CV 1.5% 1.2% 1.3% 1.3% 1.3% 1.3% 1.3% Table 5 Average CT #s as reported by Raystation TPS for the Heterogenous Phantom, see Figure 6 for the slab arrangement. The “Solid Water” and “Bone” columns are the averages of the Top and Bottom solid water and bone columns, respectively. %CV is percent coefficient of variation. 10 Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning · White Paper Predicted Dose from 200 MUs at 88 SSD for Heterogenous Phantom 6 MV Dose (cGy) 10 MV Dose (cGy) 15 MV Dose (cGy) Br38_120kVp 142 155 161 Sd40_120kVp 142 155 161 Sd40_70kVp 141 154 160 Sd40_80kVp 142 154 161 Sd40_100kVp 142 155 161 Sd40_140kVp 142 155 161 Average Dose 141.8 154.6 160.8 %CV 1.4% 1.1% 1.4% Table 6 Depicts the dosimetric point dose results for the standard 120 kVp Br38 kernel and the Sd40 DD kernels. Ave Dose is for Br38 and Sd40 kernels. %CV is percent coefficient of variation. SSD = Source-to-Surface Distance, MU = Monitor Units TPS Dose (DD) Measured Dose (cGy) (cGy) % Difference 6 MV Homogenous 165 166.8 1.1% Heterogeneous 141.8 143.2 1.0% 10 MV Homogenous 175 174.8 0.1% Heterogeneous 154.6 154.6 0.0% 15 MV Homogenous 179 180.2 0.7% Heterogeneous 160.8 162.5 1.1% Table 7 Ion chamber point dose measurements for the 1) homogenous phantom setup and 2) the heterogenous phantom setup. TPS calculations performed with the average DD curve shown in Figure 5. 11 White Paper · Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning Do the CT #s remain constant? A Sensitivity Analysis There were some data during our investigation Ratio of 120 kVp Sd40 to 120 kVp Br38 Dose that suggested that CT #s for lower density materials 1.018 changed for scans that contained both low and high density materials. This is likely due to the method in 1.013 which the DD kernel searches for and establishes bone densities first, in conjunction with the non- 1.008 typical case of large heterogeneous slabs of bone on top of soft-tissues (see Discussion section). 1.003 Variations of the CT #s were the most pronounced 0.998 Relative Dose when comparing the difference between the DD 0.993 (Sd40) kernel at 70 kVp and at 120 kVp. However, clinically, these differences appeared to result in 0.988 dose differences of less than 1% as shown in 1.7 5.1 6.0 6.8 2.5 3.4 7.7 8.6 9.4 0.8 4.3 Figures 6 and 7 (right). 10.3 11.1 12.0 12.9 13.7 14.6 15.4 16.3 17.2 Depth in Phantom (cm) Solid water–bottom Bone–bottom Lung Bone–top Solid water–top Figure 6 Ratio of calculated line doses through the heterogenous phantom for two scans at 120 kVp, one using a standard Br38 reconstruction kernel and the other using DD Ratio of 70 kVp Sd40 to 120 kVp Br38 Dose 1.018 1.013 1.008 1.003 0.998 Relative Dose 0.993 0.988 1.7 5.1 6.0 6.8 2.5 3.4 7.7 8.6 9.4 0.8 4.3 10.3 11.1 12.0 12.9 13.7 14.6 15.4 16.3 17.2 Depth in Phantom (cm) Solid water–bottom Bone–bottom Lung Bone–top Solid water–top Figure 7 Ratio of calculated line doses through the heterogenous phantom for two scans, one using standard reconstruction at 120 kVp Br 38 and the other using DD at 70 kVp Sd40 12 Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning · White Paper Thorax Phantoms evaluation The performance of the Sd40 kernel was also evaluated on heterogenous phantoms that are more clinically realistic. Two different humanoid phantoms were used for this investigation - the IROC-Houston lung phantom2 and an in-house RANDO Thorax phantom. A. The IROC lung phantom model, as shown in Figure 8, is well described in the literature2. It was scanned using the Siemens Healthineers EDGE scanner with the standard 120 kVp Br38 FBP and then Photon Lung Phantom reconstructed with 120 kVp Sd40 direct density kernel. No dosimetric differences were seen Surrounding Soft Tissue between any of the VMAT plans (Fig 9) as shown in the DVH results below were essentially Heart psilateral Lung identical (Fig 10). (Compressed Cork) Contralateral Lung (Compressed Colly Figure 8 Images from: Steinmann, A., Alvarez, P., Lee, H., Court, L., Stafford, R., Sawakuchi, G., Wen, Z., Fuller, C. and Followill, D. (2019), MRIgRT dynamic lung motion thorax anthropomorphic QA phantom: Design, development, reproducibility, and feasibility study. Med. Phys., 46: 5124-5133. doi:10.1002/mp.13757 and rpc. Spinal Cord Ipsilateral Lung In-house petroleum jedy styrofoam boll maj mdanderson.org/RPC Figure 9 VMAT plan on PTV on IROC Figure 10 DVH comparison for VMAT plan on IROC phantom using Br38 and Sd40 images phantom taken at 120 kVp. DVHs are overlapped and no difference detected 13 White Paper · Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning B. Rando thorax phantom: This method of clinical verification was repeated for a more complex, more true-to-life Alderson RANDO thorax phantom (http://rsdphantoms.com/radiation-therapy/ the-alderson-radiation-therapy-phantom/). CT #s were noted for relevant ROIs, then 3D plans and VMAT plans were made on a standard kernel image at 120 kVp and on Direct Density images taken at all available kVp (140-70 kVp). Figure 11 Alderson RANDO phantom used to verify DD-based dose calculations in Raystation CT Recon Kernel PTV Left Lung Right Lung Heart Cord Sternum Br38 120 kVp -44.5 -507.3 -507.6 25.6 42.5 146.4 Sd40 120 kVp -47.4 -504.0 -504.4 23.6 49.0 63.7 Sd40 70 kVp -62.2 -499.3 -501.5 14.0 48.6 62.0 Sd40 80 kVp -51.1 -500.0 -503.3 16.7 47.4 66.0 Sd40 100 kVp -47.6 -500.2 -503.3 20.7 47.6 67.7 Sd40 140 kVp -57.1 -504.7 -508.5 30.7 47.2 64.0 Sd40 Average CT# -51.6 -502.6 -504.8 21.8 47.1 64.7 %CV 12.1% -0.5% -0.5% 30.6% 1.7% 3.4% Table 7 CT#s for Rando Thorax phantom 14 Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning · White Paper CT Recon Kernel Calc Point cGy Max Dose cGy PTV Average Dose cGy Br38 120 4,600 5,327 4,676 Sd40 120 4,600 5,330 4,676 Sd40 70 4,600 5,328 4,689 Sd40 80 4,600 5,340 4,692 Sd40 100 4,600 5,345 4,694 Sd40 140 4,600 5,328 4,692 Average Dose (cGy) 4,600 5,333 4,686.5 %CV 0% 0.1% 0.2% Table 8 Predicted Dose for 3D Rando Thorax Plans CT Scan Dose at Volume, Max Dose, % PTV at 5000, Heart Dose, cGy cGy cGy cGy Br38 120 5,014 5,658 96.36 56 Sd40 120 5,033 5,666 97.91 67 Sd40 70 5,034 5,552 98.45 67 Sd40 80 5,066 5,721 99.6 71 Sd40 100 5,028 5,482 98.82 72 Sd40 140 5,058 5,537 99.25 78 Average Dose (cGy) 5,038.8 5,602.7 98.4 68.5 %CV 0.4% 1.6% 1.2% 11% Table 9 Predicted Dose for VMAT Rando Thorax Plans 15 White Paper · Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning Discussion Our results indicate that the DD algorithm has produced CT#s with a variance from the original 120kVP Br38 kernel (filtered back projection (FBP)) used in our clinic. It is important to realize this and evaluate the magnitude of the dosimetric impact. This is expected since DD relies on single energy two- material decomposition, there are of course slight variations in CT values depending on the setting/patient. Siemens Healthineers indicates that variations in the order of up to 20-25 CT #s are normal and to be expected. As shown in Table 4, the largest deviation in CT#s is found in the bone area (CT# of 33) for the heterogenous slab phantom. As always in CT (and especially for corrections and reconstruction), assumptions are made on the objects that are typically to be imaged in clinical situations (i.e. patients). Some of those assumptions are: an overall oval shape, the high-density objects are bones (or implants) and that they are always “far” from the edges of the patient (in other words, we assume that the patient is always surrounded by some fat/soft tissue). The design of presented heterogeneous phantom is typical of an RT phantom (successive layers of material with a hole for an ion chamber, etc.…). This is typically used to verify predicted delivered dose for external beam, but those phantoms are suboptimal for imaging studies, due to their abrupt heterogenous nature, and users should exert caution when analyzing results obtained with those type of phantoms. The square design causes some serious challenges from a “patient outline” continuity standpoint, which can significantly affect the homogeneity of the CT values within the object. Secondly, one should consider the fact that when the CT tube is either at 90 degrees or 270 degrees (assuming the 0 degree position is above the phantom), the x-ray beam “sees” 2 slabs of bone that have the same thickness as the object (which of course goes against the assumption we make as to what a patient usually looks like). This creates some significant disturbance in the sinogram and has non-negligible beam hardening effects that would also translate to variations in CT values. The dosimetric impact was largest for the 70 kVp Sd40, and on the order of 1% compared to the standard Br 38 120kVp kernel dose predictions. 16 Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning · White Paper Several reports using clinical patient data have reached similar findings. Flatten et al conducted a phantom study (simple and anthropomorphic) which also included metallic implants. Differences were found mainly in pure air and high-density materials such as bones3. The difference of the mean dose was below 0.7%, in most cases below 0.4%. No indication was found that the algorithm is corrupted by metal inserts, enabling the application for all clinical cases. van der Heyden et al performed a retrospective study on the accuracy of DD dose calculation using 33 patients with various cancer types4. All CT acquisitions were reconstructed with the standard FBP and DD. The mean tumor doses and the volume percentage that receives more than 95% of the prescribed dose were calculated for the planning target volume. Relevant parameters for the organs at risk for each tumor site were also calculated. The relative mean dose differences between the standard 120 kVp FBP CT scan workflow and the DD CT scans (80, 100, 120 and 140 kVp) were in general less than 1% for the planned target volume and organs at risk. Changes to Clinical Workflow Implementing DirectDensity in the clinic is fairly straightforward. As stated, a new CT Density curve will need to be created in the TPS. Then, appending the DirectDensity reconstruction as a secondary reconstruction for all existing protocols, will also need to be done. In this manner, physicians will be able to use the standard protocol, done at a different kVp from the usual 120 kVp, to draw the GTV/CTV and OARs contours with the benefits of enhanced tissue contrast. Dosimetry/Physics will map the contours using rigid registration tools in the TPS to the DD-reconstructed image and continue with treatment planning as normal. 17 White Paper · Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning Conclusions With an appropriate CT density curve, DD reconstruction algorithm is as accurate as standard algorithms at dose prediction but allows the flexibility of using variable kVp to improve image quality for certain tissues. Our next step is to implement DirectDensityTM for routine clinical CT simulation in our clinic. Future report of the clinical outcomes will be documented in Part II of this white paper. 18 Evaluation of the Direct Density Algorithm for Energy-Independent Radiotherapy Treatment Planning · White Paper Notes 19 At Siemens Healthineers, our purpose is to enable The information in this document contains general healthcare providers to increase value by empowering technical descriptions of specifications and options them on their journey toward expanding precision as well as standard and optional features, which do medicine, transforming care delivery, and improving not always have to be present in individual cases. patient experience, all enabled by digitalizing healthcare. Siemens Healthineers reserves the right to modify An estimated 5 million patients globally benefit every the design, packaging, specifications, and options day from our innovative technologies and services described herein without prior notice. For the most in the areas of diagnostic and therapeutic imaging, current information, please contact your local sales laboratory diagnostics, and molecular medicine, as representative from Siemens Healthineers. well as digital health and enterprise services. Note: Any technical data contained in this document We’re a leading medical technology company with may vary within defined tolerances. Original images over 120 years of experience and 18,500 patents always lose a certain amount of detail when reproduced. globally. With about 50,000 dedicated colleagues in over 70 countries, we’ll continue to innovate and References shape the future of healthcare. 1Ritter, Direct Density: Principles and Implications for Radiotherapy Siemens Healthineers White Paper. 2Steinmann, A., et al., MRIgRT dynamic lung motion thorax On account of certain regional limitations of sales anthropomorphic QA phantom: Design, development, reproducibility, rights and service availability, we cannot guarantee and feasibility study. Med Phys, 2019. 46(11): p. 5124-5133. 3Flatten, V., et al., A phantom based evaluation of the dose prediction that all products included in this brochure are and effects in treatment plans, when calculating on a direct density available through the Siemens Healthineers sales CT reconstruction. J Appl Clin Med Phys, 2020. 21(3): p. 52-61. organization worldwide. Availability and packaging 4van der Heyden, B., et al., Clinical evaluation of a novel CT image reconstruction algorithm for direct dose calculations. Physics and may vary by country and is subject to change Imaging in Radiation Oncology, 2017. 2: p. 11-16. without prior notice. Some/All of the features and products described herein may not be available in the United States. The scientific overlay is not that of the individual pictured and is not from a device of Siemens Healthineers. It was modified for better visualization. Siemens Healthineers Headquarters USA Siemens Healthcare GmbH Siemens Medical Solutions USA, Inc. Henkestr. 127 Healthcare 91052 Erlangen, Germany 40 Liberty Boulevard Phone: +49 9131 84-0 Malvern, PA 19355-9998, USA siemens-healthineers.com Phone: +1-888-826-9702 siemens-healthineers.us Published by Siemens Medical Solutions USA, Inc. · Order No. RO-20-NAM-1086 · Printed in USA · 07.2020 · ©Siemens Medical Solutions USA, Inc., 2020