Advertisement

Clinical evaluation of a novel CT image reconstruction algorithm for direct dose calculations

Open AccessPublished:March 29, 2017DOI:https://doi.org/10.1016/j.phro.2017.03.001

      Abstract

      Background and purpose

      Computed tomography (CT) imaging is frequently used in radiation oncology to calculate radiation dose distributions. In order to calculate doses, the CT numbers must be converted into densities by an energy dependent conversion curve. A recently developed algorithm directly reconstructs CT projection data into relative electron densities which eliminates the use of separate conversion curves for different X-ray tube potentials. Our work evaluates this algorithm for various cancer sites and shows its applicability in a clinical workflow.

      Materials and methods

      The Gammex phantom with tissue mimicking inserts was scanned to characterize the CT number to density conversion curves. In total, 33 patients with various cancer sites were scanned using multiple tube potentials. All CT acquisitions were reconstructed with the standard filtered back-projection (FBP) and the new developed DirectDensity™ (DD) algorithm. 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.

      Results

      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.

      Conclusion

      The energy independent DD algorithm allows for accurate dose calculations over a variety of body sites. This novel algorithm eliminates the tube potential specific calibration procedure and thereby simplifies the clinical radiotherapy workflow.

      Keywords

      1. Introduction

      Computed tomography (CT) imaging is the most frequently used modality for radiotherapy treatment planning, i.e. organ delineation and dose calculations [
      • Medical imaging modalities in radiotherapy
      ,
      • Grau C.
      • Defourny N.
      • Malicki J.
      • Dunscombe P.
      • Borras J.M.
      • Coffey M.
      • et al.
      Radiotherapy equipment and departments in the European countries: final results from the ESTRO-HERO survey.
      ]. In most cases, a fixed X-ray tube potential (e.g. 120 kVp) is chosen during the CT acquisition, although different X-ray tube potentials may offer better contrast or signal-to-noise ratio for individual patients [
      • Scholtz J.E.
      • Wichmann J.L.
      • Hüsers K.
      • Albrecht M.H.
      • Beeres M.
      • Bauer R.W.
      • et al.
      Third-generation dual-source CT of the neck using automated tube voltage adaptation in combination with advanced modeled iterative reconstruction: evaluation of image quality and radiation dose.
      ,
      • Seyal A.R.
      • Arslanoglu A.
      • Abboud S.F.
      • Sahin A.
      • Horowitz J.M.
      • Yaghmai V.
      CT of the abdomen with reduced tube voltage in adults: a practical approach.
      ,
      • Lurz M.
      • Lell M.M.
      • Wuest W.
      • Eller A.
      • Scharf M.
      • Uder M.
      • et al.
      Automated tube voltage selection in thoracoabdominal computed tomography at high pitch using a third-generation dual-source scanner: image quality and radiation dose performance.
      ].
      The CT scan of the patient is an essential part of the radiation dose calculation procedure in the treatment planning system (TPS). After delineation of target volumes and organs at risk (OAR) the treatment plan is evaluated based on the dose constraints of the OAR. Before a dose distribution can be calculated in the TPS, the CT numbers must be converted into relative electron or mass densities. Depending on the dose calculation method, either relative electron density (RED) or mass density (MD) is used. The CT number to density conversion is performed using a calibration curve based on phantom measurements with inserts of known RED and MD [
      • Schneider U.
      • Pedroni E.
      • Lomax A.
      The calibration of CT Hounsfield units for radiotherapy treatment planning.
      ,
      • Constantinou C.
      • Harrington J.C.
      • DeWerd L.A.
      An electron density calibration phantom for CT-based treatment planning computers.
      ,
      • Das I.J.
      • Cheng C.-W.
      • Cao M.
      • Johnstone P.A.S.
      Computed tomography imaging parameters for inhomogeneity correction in radiation treatment planning.
      ]. The shape and characteristics of the conversion curve is dependent on the chosen X-ray tube potential (kVp) during the calibration process.
      A novel image reconstruction algorithm was recently developed that directly reconstructs raw CT projection data into RED images. In contrast to conventional CT images, where the CT numbers quantify the linear photon attenuation and therefore depends on the energy of the X-ray photons, the novel image reconstruction algorithm quantifies the RED of the scanned object which does not depend on the X-ray photon energy, but only on the composition of the scanned object itself.
      This image reconstruction algorithm eliminates the use of a CT number to density conversion curve for each selectable X-ray tube potential in the TPS and instead a single curve can be used. Therefore, this image reconstruction algorithm has the potential to make the treatment planning workflow more simple and robust.
      Our study evaluates the accuracy of this image reconstruction algorithm in radiotherapy treatment plans by quantifying differences between the dose distributions. The differences are calculated for different cancer sites between the dose distributions of the conventional 120 kVp CT images and the RED CT images obtained with different X-ray tube potentials.

      2. Materials and methods

      2.1 Image reconstruction algorithm

      The developed image reconstruction algorithm (DirectDensity™, Siemens Healthcare GmbH, Germany) operates in both the projection and the image space whereby raw single energy CT projection data is reconstructed into CT images which are interpretable as RED images, independent from the selected X-ray tube potential. The DD image reconstruction algorithm achieves a material decomposition into water and bone, based on a threshold value for bone. Effective water and bone thicknesses are obtained by forward projecting the CT images after bone thresholding. The final RED line integrals are calculated by summing the multiplication values of each material effective thickness and its corresponding RED. A more detailed explanation of the DD image reconstruction algorithm is given in Supplemental Data 1.
      A compact notation was introduced here to denote the combinational use of the image reconstruction algorithm and the X-ray tube potential. In this notation, the first part indicates the image reconstruction algorithm: ‘FPB’ is used for filtered back-projection (Siemens B30 kernel) and ‘DD’ is used for DirectDensity™ (Siemens E30 kernel). The following number in the notation indicates the X-ray tube potential, e.g. DD-140 kVp defines a CT scan acquired at an X-ray tube potential of 140 kVp and reconstructed with the novel image reconstruction algorithm.

      2.2 Phantom calibration

      Both the FBP and the DD image reconstruction algorithms were validated with the Gammex RMI 467 phantom (Gammex, Middleton, WI). The Gammex phantom with tissue mimicking inserts (e.g. lung, adipose, soft tissues, calcium and iodine) was scanned at three different X-ray tube potentials (80, 120 and 140 kVp) with a SOMATOM Confidence® RT Pro (Siemens Healthcare GmbH, Germany) 64-slice CT scanner. The three CT scans of the phantom were obtained with a tube current of 250 mAs and reconstructed with a 3 mm slice thickness.
      To obtain CT number to RED or MD curves, all three CT scans of the Gammex phantom were reconstructed with the DD algorithm, and the 120 kVp CT scan was additionally reconstructed with the FBP algorithm.
      The data points of the CT number to density conversion curves were obtained by calculating mean CT numbers in a region of interest with respect to the certified RED and MD of the tissue mimicking Gammex inserts. The unit of the CT numbers depends on the image reconstruction algorithm: the FBP algorithm reconstructs the raw projection data in Hounsfield Units (HU) and the DD algorithm provides a simple CT number scaling proportional to the RED, described by Eq. (1).
      Further in this study, the term ‘CT number’ is only used as a general term to describe pixel values of a CT image reconstruction. In order to maintain the consistency between the unit of the CT numbers, CTHU is used to denote a FBP image reconstruction in HU, CTRED is used to denote a DD image reconstruction in the linear RED scaling (Eq. (1)).
      RED=CTRED1000+1
      (1)


      To prevent negative RED values in Eq. (1), CTRED numbers between −1024 and −1000 were forced to be −1000 by applying thresholding. The mean CTRED numbers obtained from the DD-80 kVp, DD-120 kVp and DD-140 kVp image reconstructions were combined into a mean data point of the energy independent CTRED number to density conversion curve for the novel image reconstruction algorithm. In this fit we assumed the energy independence of the CTRED numbers. The CTHU number to density conversion curve for the FBP-120 kVp image reconstruction was obtained by calculating mean HU in a region of interest as a function of the phantom insert densities.
      The CT number to density conversion curve in the TPS was interpreted as a linear interpolation between the mean data points and was not used as a linear fit between all data points.
      The CTRED number to density conversion curve, shown in the right panel of Fig. 1, was used in the TPS to perform dose calculation on the DD image reconstructions obtained at different X-ray tube potentials.
      Figure thumbnail gr1
      Fig. 1The CTHU number to mass density (MD) conversion curves (left) if the CT scans of the Gammex phantom and its tissue mimicking inserts were reconstructed with the filtered back-projection (FBP) image reconstruction algorithm and the energy independent CTRED number to MD conversion curve (right) if the same CT scans were reconstructed with the DirectDensity™ (DD) image reconstruction algorithm.

      2.3 Patient population

      The DD image reconstruction algorithm was evaluated in a clinical imaging and radiation treatment workflow. A group of 26 patients with various cancer sites were scanned with two different imaging protocols (Table 1). First the patient was scanned with the conventional X-ray tube potential of 120 kVp, whereafter an additional dual-spiral dual-energy (DE) CT acquisition was made at 80 kVp and 140 kVp. The CT scans were performed according to a clinical imaging protocol whereby an equivalent imaging dose was opted between the first 120 kVp CT acquisition and the second dual-spiral DECT acquisition.
      Table 1Relative mean dose differences (ΔD¯), differences in the volume percentage that receives more than 95% of the prescribed dose (ΔV95%) and standard deviations are calculated in the planning target volume (PTV) or organs at risk between two image reconstructions per cancer patient subgroup. The total number of cancer patients in the subgroup (N) is indicated in brackets. (FBP, filtered back-projection; DD, DirectDensity™).
      Relative differenceFBP-120 kVp

      DD-80 kVp (%)
      FPB-120 kVp

      DD-120 kVp (%)
      FBP-120 kVp

      DD-140 kVp (%)
      Prostate (N = 3)
      ΔD¯PTV0.0 ± 0.60.4 ± 0.60.4 ± 0.5
      ΔV95%,PTV−0.9 ± 0.10.7 ± 0.90.2 ± 0.5
      ΔD¯Rectum−0.2 ± 0.90.7 ± 0.60.1 ± 0.9
      ΔD¯Anal canal3.5 ± 2.20.7 ± 0.53.7 ± 2.3
      ΔD¯Bladder0.4 ± 0.80.4 ± 0.40.7 ± 0.7
      Rectum (N=4)
      ΔD¯PTV−0.2 ± 0.40.1 ± 0.10.0 ± 0.3
      ΔV95%,PTV−0.2 ± 0.20.0 ± 0.10.0 ± 0.2
      ΔD¯Bowels0.2 ± 0.70.2 ± 0.20.2 ± 0.5
      ΔD¯Bladder−0.1 ± 0.30.2 ± 0.10.0 ± 0.3
      Bone metastasis (N=7)
      ΔD¯PTV−0.1 ± 0.30.1 ± 0.10.1 ± 0.3
      ΔV95%,PTV−0.2 ± 0.50.1 ± 0.20.1 ± 0.4
      Head (N=5)
      ΔD¯PTV−0.4 ± 0.4−0.3 ± 0.5−0.4 ± 0.4
      ΔV95%,PTV0.5 ± 1.00.4 ± 0.70.5 ± 1.2
      Miscellaneous (N=7)
      ΔD¯PTV−0.2 ± 0.4−0.0 ± 0.3−0.2 ± 0.6
      ΔV95%,PTV−0.4 ± 0.7−0.3 ± 0.8−0.1 ± 0.9
      The CT scans were reconstructed to obtain four image datasets for each patient: DD-80 kVp, DD-120 kVp, DD-140 kVp and FBP-120 kVp. Patients were delineated and planned (Eclipse™ version 11, Varian Medical Systems, Palo Alto, CA) based on the FBP-120 kVp image reconstruction. Both the clinical contours and the clinical treatment plan were directly copied to the DD-80 kVp, DD-120 kVp and DD-140 kVp images. Due to the sequential acquisitions at different X-ray tube potentials, patient motion could take place between the three CT acquisitions. No correction was applied to compensate for these deformations.
      Additionally, a cohort of seven breast cancer patients was used to investigate the automatic adjustment of the X-ray tube potential and the tube current based on the topogram [] (CARE kV, Siemens Healthcare GmbH, Germany). The standard CT scan was performed with a conventional X-ray tube potential of 120 kVp and if a different tube potential was suggested (mainly 100 kVp) also this image was acquired. An Institutional Review Board approval was given for all patients included in our study.

      2.4 Analysis of the dose distributions

      Dose calculations, calculated with Acuros® XB (v.11.0.31, Eclipse, Varian Medical Systems, Palo Alto, CA) and reported in dose to medium, were performed on all CT image reconstructions of each patient included in our study [
      • Onizuka R.
      • Araki F.
      • Ohno T.
      • Nakaguchi Y.
      • Kai Y.
      • Tomiyama Y.
      • et al.
      Accuracy of dose calculation algorithms for virtual heterogeneous phantoms and intensity-modulated radiation therapy in the head and neck.
      ,
      • Zifodya J.M.
      • Challens C.H.
      • Hsieh W.L.
      From AAA to Acuros XB-clinical implications of selecting either Acuros XB dose-to-water or dose-to-medium.
      ,
      • Tsuruta Y.
      • Nakata M.
      • Nakamura M.
      • Matsuo Y.
      • Higashimura K.
      • Monzen H.
      • et al.
      Dosimetric comparison of Acuros XB, AAA, and XVMC in stereotactic body radiotherapy for lung cancer.
      ,
      • Kroon P.S.
      • Hol S.
      • Essers M.
      Dosimetric accuracy and clinical quality of Acuros XB and AAA dose calculation algorithm for stereotactic and conventional lung volumetric modulated arc therapy plans.
      ,
      • Fogliata A.
      • Nicolini G.
      • Clivio A.
      • Vanetti E.
      • Cozzi L.
      Dosimetric evaluation of Acuros XB Advanced Dose Calculation algorithm in heterogeneous media.
      ]. The Acuros® dose calculation engine uses MD to assign materials inside the voxels, hence a CT number to MD conversion curve was required. Note that a CT number to RED conversion must be used if the dose calculation engine scales the pre-calculated dose kernels according to the RED matrix as is done e.g. in the analytical anisotropic algorithm (AAA, Eclipse™, Varian, Palo Alto, CA) [
      • Van Esch A.
      • Tillikainen L.
      • Pyykkonen J.
      • Tenhunen M.
      • Helminen H.
      • Siljamãki
      • et al.
      Testing of the analytical anisotropic algorithm for photon dose calculation.
      ].

      2.5 Analysis and statistics

      The first group of 26 patients was subdivided into five smaller patient subgroups, the group of seven breast cancer patients was considered as the sixth subgroup.
      The subgroups and the number of patients inside each subgroup, indicated between brackets, were subsequently listed: prostate cancer (N = 3), rectal cancer (N = 4), bone metastasis (N = 7), brain metastases (N = 5), breast cancer (N = 7) and miscellaneous tumor sites (N = 7). The miscellaneous tumor site group consisted of cancers in the bladder, spleen, stomach, groin metastasis and seminoma testis metastasis patients.
      For each of the six subgroups, mean doses and the volume percentage that receives more than 95% of the prescribed dose (V95%) were calculated in the planning target volume (PTV). Because the OAR delineation depended on the tumor position, the treatment plan and intent, the same OAR were not always delineated throughout the entire cancer patient subgroup. The mean doses delivered in the OAR were only calculated for the prostate, rectum and lung cancer patient subgroup. For each of the OAR, the dose volume histogram parameters that were used as planning constraints were compared between the different dose distributions.

      3. Results

      3.1 Calibration and phantom evaluation

      The left panel of Fig. 1 shows the energy dependency of the HU on the material inserts. The energy dependence was reduced using the DD algorithm as shown in the conversion curve for the DD-80 kVp, DD-120 kVp and DD-140 kVp image reconstructions (Fig. 1 right panel). The CTRED numbers which are proportional to the RED (Eq. (1)) had a −0.1%±2.2% (1SD) residual from the linear identity line for all tissue mimicking Gammex inserts and all X-ray tube potentials. Individually, relative differences up to −0.1% ± 1.8% were observed for the DD-120 kVp image reconstruction, −0.1% ± 2.8% for the DD-80 kVp image reconstruction and 0.0% ± 1.9% for the DD-140 kVp image reconstruction.

      3.2 Patient dose distributions

      Relative differences in mean dose and in volumetric dose coverage were calculated between the clinically used FBP-120 kVp image reconstruction and the other DD-80 kVp, DD-100 kVp (breast cancer patient subgroup only), DD-120 kVp and DD-140 kVp image reconstructions. The relative mean dose differences and the mean differences in the V95% were averaged over all cancer patients in a specific subgroup, and are listed in Table 1, Table 2. All relative differences and standard deviations in mean dose and V95% for each treatment site were smaller than 1%, except for the dose differences in the anal canal. Here, the largest relative mean dose differences of almost 4% and large standard deviations of 2.3% were found in the DD-80 kVp and 140 kVp image reconstruction and are most likely caused by the small organ displacement in the images. The relative mean dose differences and standard deviation of the anal canal between the DD-120 kVp and DD-140 kVp image reconstructions were both less than 1%.
      Table 2Relative mean dose differences (ΔD¯), differences in the volume percentage that receives more than 95% of the prescribed dose (V95%) and standard deviations are calculated in the planning target volume (PTV) or organs at risk between the FBP-120 kVp and DD-100 kVp image reconstructions of the 7 breast cancer patient analyzed. (FBP, filtered back-projection; DD, DirectDensity™).
      Relative differenceFBP-120 

      kVp DD-100 kVp (%)
      Breast (N = 7)
      ΔD¯PTV−0.3 ± 0.2
      ΔV95%,PTV−0.6 ± 0.8
      ΔD¯Heart0.2 ± 0.4
      ΔD¯Left lung0.3 ± 0.4
      ΔD¯Right lung0.0 ± 0.2
      Fig. 2, Fig. 3 show the isodose lines of a prostate and a head-and-neck cancer patient, respectively. These isodose lines of both patients showed similar patterns between the dose distributions calculated based on different image reconstructions. Both patients had the largest relative mean dose differences in their patient group, due to their large body volume more voxels could be assigned incorrectly in the density to material conversion which is done before the dose calculation is started, although the relative difference for both tumor and OAR were smaller than 1%. The isodose lines for other cancer sites showed similar results.
      Figure thumbnail gr2
      Fig. 2Dose distributions of a clinical prostate cancer treatment plan calculated on the four different image reconstructions (window: width = 300, level = 40) (CTDIvol,80kVp = 7.9 mGy; CTDIvol,140kVp = 8.7 mGy; CTDIvol,120kVp = 17.7 mGy).
      Figure thumbnail gr3
      Fig. 3Dose distributions of a clinical treatment plan of a head and neck cancer patient recalculated on the four different image reconstructions (window: width = 300, level = 40) (CTDIvol,80kVp = 1.3 mGy; CTDIvol,140kVp = 2.0 mGy; CTDIvol,120kVp = 3.3 mGy).
      The higher image noise in the dual-spiral DECT images compared to the 120 kVp CT images was due to the lower volume CT dose index (CTDIvol) of the separate 80 kVp and 140 kVp CT acquisitions compared to the 120 kVp CT acquisition. A lower imaging dose results in a CT image with more noise. The total imaging dose of the 120 kVp CT scan is divided over the 80 kVp and the 140 kVp CT acquisitions.
      The CT images obtained through the DD image reconstruction algorithm gave a better image contrast inside dense regions (e.g. bone) compared to the standard FBP image reconstruction.

      4. Discussion

      We evaluated a novel method to simplify the clinical radiation therapy workflow. The CTRED number to MD conversion curve needed for the DD image reconstruction algorithm was obtained through a calibration with the Gammex phantom and its certified tissue mimicking inserts and was independent of tube potential.
      The calibrated CTRED number to MD conversion curve (Fig. 1, right panel) was implemented in the TPS and could be used to perform dose calculations on DD image reconstructions independent from the X-ray tube potential of the CT scanner. The DD image reconstruction algorithm used in our study is based on the FBP image reconstruction algorithm, as also the conventional CT images in this study. However, there is also the possibility to reconstruct the DD images with an iterative reconstruction (SAFIRE or ADMIRE, Siemens terminology) variant of this algorithm (Siemens F30 kernel). This could potentially reduce the imaging dose and noise further [
      • Scholtz J.E.
      • Wichmann J.L.
      • Hüsers K.
      • Albrecht M.H.
      • Beeres M.
      • Bauer R.W.
      • et al.
      Third-generation dual-source CT of the neck using automated tube voltage adaptation in combination with advanced modeled iterative reconstruction: evaluation of image quality and radiation dose.
      ,
      • Rivers-Bowerman M.D.
      • Shankar J.J.
      Iterative reconstruction for head CT: effects on radiation dose and image quality.
      ,
      • Baumueller S.
      • Winklehner A.
      • Karlo C.
      • Goetti R.
      • Flohr T.
      • Russi E.W.
      • et al.
      Low-dose CT of the lung: potential value of iterative reconstructions.
      ,
      • Han B.K.
      • Grant K.L.
      • Garberich R.
      • Sedlmair M.
      • Lindberg J.
      • Lesser J.R.
      Assessment of an iterative reconstruction algorithm (SAFIRE) on image quality in pediatric cardiac CT datasets.
      ,
      • Wortman J.R.
      • Adduci A.J.
      • Sodickson A.D.
      Synergistic radiation dose reduction by combining automatic tube voltage selection and iterative reconstruction.
      ].
      Alternatively, a different method that still needs to be investigated could be found to derive the conversion curve: Beaulieu et al. [
      • Beaulieu L.
      • Carlsson Tedgren A.
      • Carrier J.F.
      • Davis S.D.
      • Mourtada F.
      • Rivard M.J.
      • et al.
      Report of the Task Group 186 on model-based dose calculation methods in brachytherapy beyond the TG-43 formalism: current status and recommendations for clinical implementation.
      ] used real human tissue compositions, taken from ICRP23 [

      International Commission on Radiological Protection, Report on the Task Group on Reference Man, ICRP Publication No. 23.

      ] and listed in Schneider et al. [
      • Schneider U.
      • Pedroni E.
      • Lomax A.
      The calibration of CT Hounsfield units for radiotherapy treatment planning.
      ], to fit a linear relationship between MD and RED. This linear fit is then derived from real human tissue compositions instead of tissue mimicking plastics and could possibly be used as a conversion curve of the DD image reconstruction algorithm, although scanner specific changes will not be taken into account.
      The relative mean dose differences in the PTV and the OAR were evaluated for all 33 patients with various cancer sites. For each patient, the relative mean dose difference between both image reconstruction algorithms at different X-ray tube potentials is smaller than 1%, excluding the anal canal because of the displacement of this relatively small structure.
      A dual-spiral DECT scan was used which could cause image reconstructions that were slightly different due to the patient’s respiratory motion. When small delineated structures, such as the anal canal (OAR of the prostate cancer patient subgroup), are copied from the FPB-120 kVp image reconstruction to other DD image reconstructions obtained from the DECT scan, the anal canal can be slightly displaced. This is supported by the fact that the mean dose difference of the anal canal (prostate cancer patient subgroup) between the FBP-120 kVp and the DD-120 kVp image reconstructions is much smaller because both DD image reconstructions were performed on the same raw projection data.
      By changing the clinical imaging workflow towards the DD algorithm, the CTRED number to density conversion curve in the TPS is now independent from the X-ray tube potential of the CT scanner. Only one CTRED number to density conversion curve should be implemented in the TPS to perform dose calculations on CT scans acquired at different kVp settings.
      The dose constraint differences between the clinically accepted FBP-120 kVp and the DD treatment plans were small, hence the treatment plan based on the novel DD image reconstruction can be seen as clinical acceptable for various X-ray tube potentials.
      The use of DD reduces both the clinical work load and the risk of selecting incorrect conversion curves in the TPS, but it also offers the opportunity to optimize each CT protocol in terms of image quality without concern for the accuracy of the dose calculations.

      5. Conclusion

      An image reconstruction that directly reconstructs relative electron density images independent from the X-ray tube potential was evaluated in a clinical workflow. We found small differences in dose distribution for the PTV and the OAR of various cancer sites. Therefore this reconstruction algorithm could be advantageous for implementation in the clinic as it has the potential to improve the workflow, selection of different kVp settings and make the treatment planning procedure more robust.

      Conflict of interest

      This study was partly supported by (Siemens Healthcare GmbH, Germany).

      Acknowledgements

      Vittorio Colombo from Siemens Healthineers is acknowledged for his valuable comments during this work.

      Appendix A. Supplementary data

      References

        • Medical imaging modalities in radiotherapy
        Lu Jiade J. Brady Luther W. Radiation oncology an evidence-based approach. Springer, Berlin Heidelberg2008: 625-639
        • Grau C.
        • Defourny N.
        • Malicki J.
        • Dunscombe P.
        • Borras J.M.
        • Coffey M.
        • et al.
        Radiotherapy equipment and departments in the European countries: final results from the ESTRO-HERO survey.
        Radiother Oncol. 2014; 112: 155-164
        • Scholtz J.E.
        • Wichmann J.L.
        • Hüsers K.
        • Albrecht M.H.
        • Beeres M.
        • Bauer R.W.
        • et al.
        Third-generation dual-source CT of the neck using automated tube voltage adaptation in combination with advanced modeled iterative reconstruction: evaluation of image quality and radiation dose.
        Eur Radiol. 2016; 26: 2623-2631
        • Seyal A.R.
        • Arslanoglu A.
        • Abboud S.F.
        • Sahin A.
        • Horowitz J.M.
        • Yaghmai V.
        CT of the abdomen with reduced tube voltage in adults: a practical approach.
        Radiographics. 2015; 35: 1922-1939
        • Lurz M.
        • Lell M.M.
        • Wuest W.
        • Eller A.
        • Scharf M.
        • Uder M.
        • et al.
        Automated tube voltage selection in thoracoabdominal computed tomography at high pitch using a third-generation dual-source scanner: image quality and radiation dose performance.
        Invest Radiol. 2015; 50: 352-360
        • Schneider U.
        • Pedroni E.
        • Lomax A.
        The calibration of CT Hounsfield units for radiotherapy treatment planning.
        Phys Med Biol. 1996; 41: 111-124
        • Constantinou C.
        • Harrington J.C.
        • DeWerd L.A.
        An electron density calibration phantom for CT-based treatment planning computers.
        Med Phys. 1992; 19: 325-327
        • Das I.J.
        • Cheng C.-W.
        • Cao M.
        • Johnstone P.A.S.
        Computed tomography imaging parameters for inhomogeneity correction in radiation treatment planning.
        J Med Phys. 2016; 41: 3-11
      1. Grant K, Schmidt B. CARE kV. White paper 2011. https://health.siemens.com/ct_applications/somatomsessions/index.php/care-kv-how-to-optimize-individualized-dose/ (Accessed: 29 December 2016).

        • Onizuka R.
        • Araki F.
        • Ohno T.
        • Nakaguchi Y.
        • Kai Y.
        • Tomiyama Y.
        • et al.
        Accuracy of dose calculation algorithms for virtual heterogeneous phantoms and intensity-modulated radiation therapy in the head and neck.
        Radiol Phys Technol. 2016; 9: 77-87
        • Zifodya J.M.
        • Challens C.H.
        • Hsieh W.L.
        From AAA to Acuros XB-clinical implications of selecting either Acuros XB dose-to-water or dose-to-medium.
        Australas Phys Eng Sci Med. 2016; 39: 431-439
        • Tsuruta Y.
        • Nakata M.
        • Nakamura M.
        • Matsuo Y.
        • Higashimura K.
        • Monzen H.
        • et al.
        Dosimetric comparison of Acuros XB, AAA, and XVMC in stereotactic body radiotherapy for lung cancer.
        Med Phys. 2014; 41: 081715
        • Kroon P.S.
        • Hol S.
        • Essers M.
        Dosimetric accuracy and clinical quality of Acuros XB and AAA dose calculation algorithm for stereotactic and conventional lung volumetric modulated arc therapy plans.
        Radiat Oncol. 2013; 8: 149
        • Fogliata A.
        • Nicolini G.
        • Clivio A.
        • Vanetti E.
        • Cozzi L.
        Dosimetric evaluation of Acuros XB Advanced Dose Calculation algorithm in heterogeneous media.
        Radiat Oncol. 2011; 6: 82
        • Van Esch A.
        • Tillikainen L.
        • Pyykkonen J.
        • Tenhunen M.
        • Helminen H.
        • Siljamãki
        • et al.
        Testing of the analytical anisotropic algorithm for photon dose calculation.
        Med Phys. 2006; 33: 4130-4148
        • Rivers-Bowerman M.D.
        • Shankar J.J.
        Iterative reconstruction for head CT: effects on radiation dose and image quality.
        Can J Neurol Sci. 2014; 41: 620-625
        • Baumueller S.
        • Winklehner A.
        • Karlo C.
        • Goetti R.
        • Flohr T.
        • Russi E.W.
        • et al.
        Low-dose CT of the lung: potential value of iterative reconstructions.
        Eur Radiol. 2012; 22: 2597-2606
        • Han B.K.
        • Grant K.L.
        • Garberich R.
        • Sedlmair M.
        • Lindberg J.
        • Lesser J.R.
        Assessment of an iterative reconstruction algorithm (SAFIRE) on image quality in pediatric cardiac CT datasets.
        J Cardiovasc Comput Tomogr. 2012; 6: 200-204
        • Wortman J.R.
        • Adduci A.J.
        • Sodickson A.D.
        Synergistic radiation dose reduction by combining automatic tube voltage selection and iterative reconstruction.
        J Thorac Imaging. 2016; 31: 111-118
        • Beaulieu L.
        • Carlsson Tedgren A.
        • Carrier J.F.
        • Davis S.D.
        • Mourtada F.
        • Rivard M.J.
        • et al.
        Report of the Task Group 186 on model-based dose calculation methods in brachytherapy beyond the TG-43 formalism: current status and recommendations for clinical implementation.
        Med Phys. 2012; 39: 9208-9236
      2. International Commission on Radiological Protection, Report on the Task Group on Reference Man, ICRP Publication No. 23.