Advertisement

In vivo assessment of tissue-specific radiological parameters with intra- and inter-patient variation using dual-energy computed tomography

  • Nils Peters
    Correspondence
    Corresponding author at: OncoRay – National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Helmholtz-Zentrum Dresden-Rossendorf, Händelallee 26, 01309 Dresden, Germany.
    Affiliations
    OncoRay – National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Helmholtz-Zentrum Dresden – Rossendorf, Dresden, Germany

    Helmholtz-Zentrum Dresden – Rossendorf, Institute of Radiooncology – OncoRay, Dresden, Germany
    Search for articles by this author
  • Aaron Kieslich
    Affiliations
    OncoRay – National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Helmholtz-Zentrum Dresden – Rossendorf, Dresden, Germany

    Helmholtz-Zentrum Dresden – Rossendorf, Institute of Radiooncology – OncoRay, Dresden, Germany
    Search for articles by this author
  • Author Footnotes
    1 Now with Siemens Healthineers, Forchheim, Germany.
    Patrick Wohlfahrt
    Footnotes
    1 Now with Siemens Healthineers, Forchheim, Germany.
    Affiliations
    OncoRay – National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Helmholtz-Zentrum Dresden – Rossendorf, Dresden, Germany

    Helmholtz-Zentrum Dresden – Rossendorf, Institute of Radiooncology – OncoRay, Dresden, Germany
    Search for articles by this author
  • Christian Hofmann
    Affiliations
    Siemens Healthineers, Forchheim, Germany
    Search for articles by this author
  • Christian Richter
    Affiliations
    OncoRay – National Center for Radiation Research in Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Helmholtz-Zentrum Dresden – Rossendorf, Dresden, Germany

    Helmholtz-Zentrum Dresden – Rossendorf, Institute of Radiooncology – OncoRay, Dresden, Germany

    Department of Radiotherapy and Radiation Oncology, Faculty of Medicine and University Hospital Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany

    German Cancer Consortium (DKTK), partner site Dresden, and German Cancer Research Center (DKFZ), Heidelberg, Germany
    Search for articles by this author
  • Author Footnotes
    1 Now with Siemens Healthineers, Forchheim, Germany.
Open AccessPublished:August 06, 2022DOI:https://doi.org/10.1016/j.radonc.2022.07.021

      Highlights

      • First comprehensive DECT-based in vivo tissue parameter analysis in a large patient cohort.
      • Expansion of current standard by intra- and inter patient parameter variation.
      • Large deviations from ICRU46 ex vivo tissue parameters observed for kidney and liver.

      Abstract

      Purpose/objective

      Experimental in vivo determination of radiological tissue parameters of organs in the head and pelvis within a large patient cohort, expanding on the current standard human tissue database summarized in ICRU46.

      Material/methods

      Relative electron density (RED), effective atomic number (EAN) and stopping-power ratio (SPR) were obtained from clinical dual-energy CT scans using a clinically validated DirectSPR implementation and organ segmentations of 107 brain-tumor (brain, brainstem, spinal cord, chiasm, optical nerve, lens) and 120 pelvic cancer patients (prostate, kidney, liver, bladder). The impact of contamination by surrounding tissues on the tissue parameters was reduced with a dedicated contour adaption routine. Tissue parameters were characterized regarding the cohort mean value as well as the variation within each patient (2σintra) and between patients (2σinter). For the brain, age-dependent differences were determined.

      Results

      For 10 organs, including 4 structures not listed in ICRU46, the mean RED, EAN and SPR as well as their respective intra- and inter-patient variation were determined. SPR intra-patient variation was higher than 1.3% (1.3–4.6%) in all organs and always exceeded the inter-patient variation of the organ mean SPR (0.6–2.1%). For the brain, a significant SPR variation between pediatric and non-pediatric patients was determined.

      Conclusion

      Radiological tissue parameters in the head and pelvis were characterized in vivo for a large patient cohort using dual-energy CT. This reassesses parts of the current standard database defined in ICRU46, furthermore complementing the data described in literature by smaller substructures in the brain as well as by the quantification of organ-specific inter- and intra-patient variation.

      Keywords

      Reliable knowledge of tissue properties is essential for a multitude of radiotherapeutic applications, such as Monte Carlo transport simulations in external radiotherapy and brachytherapy [
      • Rogers D.W.O.
      Fifty years of Monte Carlo simulations for medical physics.
      ,
      • Bazalova M.
      • Graves E.E.
      The importance of tissue segmentation for dose calculations for kilovoltage radiation therapy.
      ,
      • Beaulieu L.
      • Carlsson Tedgren Å.
      • 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.
      ,
      • Mann-Krzisnik D.
      • Verhaegen F.
      • Enger S.A.
      The influence of tissue composition uncertainty on dose distributions in brachytherapy.
      ], stoichiometric CT calibration [
      • Schaffner B.
      • Pedroni E.
      The precision of proton range calculations in proton radiotherapy treatment planning: experimental verification of the relation between CT-HU and proton stopping power.
      ,
      • Yang M.
      • Zhu X.R.
      • Park P.C.
      • Titt U.
      • Mohan R.
      • Virshup G.
      • et al.
      Comprehensive analysis of proton range uncertainties related to patient stopping-power-ratio estimation using the stoichiometric calibration.
      ,
      • Schneider W.
      • Bortfeld T.
      • Schlegel W.
      Correlation between CT numbers and tissue parameters needed for Monte Carlo simulations of clinical dose distributions.
      ] and the creation of anthropomorphic phantoms [
      • Jones D.G.
      A realistic anthropomorphic phantom for calculating organ doses arising from external photon irradiation.
      ]. Based on their elemental composition, the interaction between ionizing radiation and patients’ tissue is calculated. In particle therapy, tissue properties are used in dual-energy CT for range prediction [
      • Hünemohr N.
      • Krauss B.
      • Tremmel C.
      • Ackermann B.
      • Jäkel O.
      • Greilich S.
      Experimental verification of ion stopping power prediction from dual energy CT data in tissue surrogates.
      ,
      • Möhler C.
      • Wohlfahrt P.
      • Richter C.
      • Greilich S.
      Range prediction for tissue mixtures based on dual-energy CT.
      ] as well as in range verification via PET [
      • Parodi K.
      • Pönisch F.
      • Enghardt W.
      Experimental study on the feasibility of in-beam PET for accurate monitoring of proton therapy.
      ] or prompt-gamma emission [
      • Polf J.C.
      • Peterson S.
      • Ciangaru G.
      • Gillin M.
      • Beddar S.
      Prompt gamma-ray emission from biological tissues during proton irradiation: a preliminary study.
      ]. Tissue parameters are furthermore used for improving other imaging modalities such as MRI [
      • Korhonen J.
      • Kapanen M.
      • Keyriläinen J.
      • Seppälä T.
      • Tenhunen M.
      A dual model HU conversion from MRI intensity values within and outside of bone segment for MRI-based radiotherapy treatment planning of prostate cancer.
      ] and cone-beam CT [
      • Sisniega A.
      • Zbijewski W.
      • Badal A.
      • Kyprianou I.S.
      • Stayman J.W.
      • Vaquero J.J.
      • et al.
      Monte Carlo study of the effects of system geometry and antiscatter grids on cone-beam CT scatter distributions.
      ].
      The most-established reference dataset of tissue parameters is based on the work of Woodard and White from the 1980s, summarizing and harmonizing results from other groups as well as amending them in several parts [
      • Woodard H.Q.
      • White D.R.
      The composition of body tissues.
      ]. Their data was used in the International Commission on Radiation Units and Measurements (ICRU) report 44 and 46 [
      • White D.R.
      • Booz J.
      • Griffith R.V.
      • Spokas J.J.
      • Wilson I.J.
      ICRU report 44.
      ,
      • White D.R.
      • Griffith R.V.
      • Wilson I.J.
      ICRU report 46.
      ] as well as in the International Commission on Radiological Protection report ICRP89 [
      • Valentin J.
      Basic anatomical and physiological data for use in radiological protection: reference values.
      ]. In their work, Woodard and White analyzed the content of water, lipid, protein, carbohydrates and high-atomic elements (Z8) for 56 body tissues from adult humans, calculated the corresponding elemental composition and estimated the resulting mass- and electron density.
      While the tissue database by Woodard and White is comprehensive and widely used, there are several limitations already pointed out by the authors when collecting the data [
      • Woodard H.Q.
      • White D.R.
      The composition of body tissues.
      ,
      • White D.R.
      • Griffith R.V.
      • Wilson I.J.
      ICRU report 46.
      ]: Information was gathered from different sources without a common standard in the experimental handling of the investigated tissues. Some were examined fresh after procurement, others were dehydrated, defatted, or otherwise manipulated before examination. Furthermore, as all tissues were investigated ex vivo, blood flow was not fully considered, and decay was inevitable.
      In an extension of their original work [
      • White D.R.
      • Widdowson E.M.
      • Woodard H.Q.
      • Dickerson J.W.T.
      The composition of body tissues. (II) Fetus to young adult.
      ] as well as in ICRU 46, the authors distinguish tissue parameters for different age groups (fetus, infant, child and adult). However, due to the tremendous experimental effort needed for analysis, tissues were obtained only from a small number of individuals. Described variations are therefore primarily case reports. Factors specific to the individual, such as health status and diseases, physical activity and, closely connected, hydration as well as diet are not sufficiently reflected in the data.
      Recent innovations in medical imaging allow for a highly precise voxel-wise determination of radiological parameters from patient CT scans [
      • Van Elmpt W.
      • Landry G.
      • Das M.
      • Verhaegen F.
      Dual energy CT in radiotherapy: current applications and future outlook.
      ,
      • Paganetti H.
      • Beltran C.J.
      • Both S.
      • Dong L.
      • Flanz J.B.
      • Furutani K.M.
      • et al.
      Roadmap: proton therapy physics and biology.
      ]. In combination with clinical organ delineation, this enables a tissue-specific characterization of different organs and substructures. The benefits are evident: Tissues can be investigated in vivo in a standardized procedure, increasing accuracy and allowing for a straightforward assessment of error sources, as all error sources can easily be traced back to the prediction model, limiting the experimental influence. Furthermore, as data is already gathered in clinical routine, large patient cohorts can be analyzed. The impact of individual-related influences (health, hydration, diet) on the average tissue parameters is therefore reduced, while variations over the population can be quantified adequately.
      Here, we present a comprehensive patient-cohort analysis of the radiological tissue parameters relative electron density (RED), effective atomic number (EAN) and stopping-power ratio (SPR) for tissues in the head and pelvis, utilizing a clinically implemented tissue characterization tool. The tool comprises a DirectSPR implementation, using dual-energy CT (DECT) for parameter determination. For the brain, age dependency of the SPR is analyzed and quantified.

      Materials and methods

      Patient cohort

      CT scans and organ contours from 107 patients with a brain tumor and 120 patients with cancer in the pelvic region were selected from the clinical database considering a balanced distribution of sex and age (Table 1 and Supplement EA). All patients were treated at the University Proton Therapy Dresden (Dresden, Germany) with either protons or photons. The retrospective study was approved by the local ethics committee (EK535122015).
      Table 1Mean tissue parameters and their relative inter- and intra-patient variation relative to water for organs in the head and pelvic region. Variation of the effective atomic number is depicted in absolute numbers.
      Relative electron densityEffective atomic numberStopping-power ratio
      Number of contoursAge in years (Avg. ± 2σ)Mean2σintra in %2σinter in %Mean2σintra2σinterMean2σintra in %2σinter in %
      Brain10742 ± 221.0311.60.87.680.340.101.0291.80.8
      Brainstem10742 ± 221.0321.20.97.570.230.101.0321.41.0
      Spinal cord5341 ± 221.0321.51.07.570.320.121.0321.71.1
      Optical nerve10742 ± 221.0202.71.17.610.650.451.0192.91.4
      Chiasm10342 ± 221.0181.91.27.680.350.181.0162.01.3
      Lens10542 ± 221.0682.62.07.490.320.491.0692.72.1
      Prostate5067 ± 111.0312.30.67.640.540.111.0302.50.6
      Kidney3049 ± 201.0214.31.17.381.650.731.0234.91.1
      Kidney without calyces3049 ± 201.0262.90.87.401.590.781.0273.81.2
      Liver1542 ± 221.0502.91.37.531.070.341.0503.41.3
      Urine10664 ± 131.0081.91.47.640.460.411.0082.11.5
      For the brain-tumor patients, clinical contours of the brain, brainstem, optical nerve, lens, chiasm, lacrimal glands and cervical spinal cord; in the pelvis, contours of the prostate, kidney, liver and bladder (thus urine) were analyzed, in each case excluding the delineated tumor volume and the overlap with other contours. Paired organs were combined if both sides were delineated. While in the head most contours were present in most of the patients, number of contours in the pelvis varied widely, as organ segmentation depends heavily on tumor location. The number of contours analyzed for each organ as well as the average patient age are listed in Table 1.

      DECT scans and tissue characterization

      Dual-spiral DECT scans (two consecutive CT scans at 80 and 140 kVp) were acquired on a single-source CT scanner SOMATOM Definition AS (Siemens Healthineers, Forchheim, Germany) with automatic exposure control (CareDose 4D, CTDIvol, 32cm15mGy) in the pelvis and a fixed exposure (CTDIvol, 16cm44mGy) in the head. Slice collimation was 1.2 mm. Image noise was reduced by using an iterative reconstruction kernel Q34f/5 (SAFIRE at maximum strength) with 0.6 × 0.6 × 2 mm3 and 1 × 1 × 2 mm3 voxel spacing in the head and pelvis, respectively. The kernel comprised an iterative beam hardening correction for bone to improve CT number stability [
      • Wohlfahrt P.
      • Möhler C.
      • Hietschold V.
      • Menkel S.
      • Greilich S.
      • Krause M.
      • et al.
      Clinical implementation of dual-energy CT for proton treatment planning on pseudo-monoenergetic CT scans.
      ]. Patient motion between two scans were limited by immobilization devices (head cushion and thermoplastic masks for the head and vacuum cushion for the pelvis). Remaining minimal organ movement was addressed by deformable image registration.
      For radiological tissue characterization, a DirectSPR prototype implementation (Siemens Healthineers, Forchheim, Germany) with an in-house size-dependent calibration was used [
      • Peters N.
      • Wohlfahrt P.
      • Hofmann C.
      • Möhler C.
      • Menkel S.
      • Tschiche M.
      • et al.
      Reduction of clinical safety margins in proton therapy enabled by the clinical implementation of dual-energy CT for direct stopping-power prediction.
      ]. In previous work, the RED and SPR prediction accuracy in soft tissues was experimentally determined to be 0.1±0.2% [
      • Möhler C.
      • Russ T.
      • Wohlfahrt P.
      • Elter A.
      • Runz A.
      • Richter C.
      • et al.
      Experimental verification of stopping-power prediction from single- and dual-energy computed tomography in biological tissues.
      ]. The patient size is approximated by the effective water-equivalent thickness obtained slice-wise from the CT scans to increase calibration accuracy. For additional noise reduction in the tissue parameter calculation, spatial frequency filtering was applied [
      • Grant K.L.
      • Flohr T.G.
      • Krauss B.
      • Sedlmair M.
      • Thomas C.
      • Schmidt B.
      Assessment of an advanced image-based technique to calculate virtual monoenergetic computed tomographic images from a dual-energy examination to improve contrast-to-noise ratio in examinations using iodinated contrast media.
      ]. Remaining noise was quantified for soft tissues in a large body and small head phantom geometry. In the head, remaining noise σnoise was below 0.4% for SPR and RED and below 10% for EAN; in the pelvis below 0.8% for SPR and RED and below 20% for EAN. The conversion from CT number to the percentage level follows from the respective DICOM saving format, where for RED and SPR Δ1HU corresponds to ΔRED/SPR=0.001 and ΔEAN = 0.1 for EAN. The DirectSPR implementation was comprehensively validated and its accuracy quantified in previous work [
      • Peters N.
      • Wohlfahrt P.
      • Hofmann C.
      • Möhler C.
      • Menkel S.
      • Tschiche M.
      • et al.
      Reduction of clinical safety margins in proton therapy enabled by the clinical implementation of dual-energy CT for direct stopping-power prediction.
      ].
      Tissue parameters RED and EAN are derived for each voxel from the superposition of the two CT scans Hlow and HHigh [
      • Hünemohr N.
      • Krauss B.
      • Tremmel C.
      • Ackermann B.
      • Jäkel O.
      • Greilich S.
      Experimental verification of ion stopping power prediction from dual energy CT data in tissue surrogates.
      ]:
      (RED-1)·1000HU=αREDHlow+(1-αRED)Hhighand


      EAN3.1=1REDαEANHlow1000HU+1+EANwater3.1-αEANHhigh1000HU+1,


      with αRED and αEAN as size-specific calibration factors. The subsequent SPR is calculated as the product of RED and relative stopping number (RSN) according to the Bethe equation. RSN was estimated from EAN3.1 using a heuristic calibration, in which pairs of EAN and RSN from tabulated human tissues are fitted, as depicted in previous work [
      • Möhler C.
      • Wohlfahrt P.
      • Richter C.
      • Greilich S.
      Range prediction for tissue mixtures based on dual-energy CT.
      ,
      • Peters N.
      • Wohlfahrt P.
      • Hofmann C.
      • Möhler C.
      • Menkel S.
      • Tschiche M.
      • et al.
      Reduction of clinical safety margins in proton therapy enabled by the clinical implementation of dual-energy CT for direct stopping-power prediction.
      ].
      Reference tissue parameters for Woodard and White were calculated from the respective elemental composition [
      • Woodard H.Q.
      • White D.R.
      The composition of body tissues.
      ] following the methodology described in [
      • Möhler C.
      • Wohlfahrt P.
      • Richter C.
      • Greilich S.
      Range prediction for tissue mixtures based on dual-energy CT.
      ] for a nominal proton beam energy of 100 MeV (see Supplement EB for full methodology and calculated tissue parameters). Because an accurate differentiation between grey matter (GM), white matter (VM) and spinal fluid (CSF) is challenging based on CT, the respective tissue parameters from Woodard and White were weighted according to their volumetric fraction: VGM=55%, VWM=27%, VSF=18% [
      • Lüders E.
      • Steinmetz H.
      • Jäncke L.
      Brain size and grey matter volume in the healthy human brain.
      ].

      Refinement of clinical organ delineation

      Tissues as well as target volumes were delineated in clinical routine by experienced radiation oncologists in the treatment planning system RayStation (RaySearch, Stockholm, Sweden) or XIO (Elekta AB, Stockholm, Sweden) on 79 keV pseudo-monoenergetic CT (MonoCT) datasets, corresponding to an effective X-ray attenuation of 140 kVp CT scans. MonoCT datasets were obtained from DECT using the module Syngo.CT DE Monoenergetic Plus in the syngo.via environment (Siemens Healthineers, Forchheim, Germany).
      Clinical contours of the tumour and organs at risk (OAR) were created for treatment planning purposes. While dose calculation is not affected by small overlaps of contours with surrounding tissues, in tissue characterization any overlap would effectively compromise the results. Therefore, a multi-step automated contour adaptation was performed for each axial CT slice to reduce potential contamination from surrounding tissues as well as inter-observer variation in OAR delineation [
      • Vinod S.K.
      • Jameson M.G.
      • Min M.
      • Holloway L.C.
      Uncertainties in volume delineation in radiation oncology: a systematic review and recommendations for future studies.
      ]. It included the following steps, with each step finetuned to the specific adaption needs of the respective organ:
      • (1)
        Excluding voxels with CT numbers that can be attributed to surrounding tissues (e.g. <−125 HU for air-tissue mixtures, >75 HU for bone-tissue mixtures).
      • (2)
        Filling small holes (<5 voxels) within the contour resulting from voxels that were removed in the first step due to noise.
      • (3)
        Smoothing contours with a morphological opening procedure.
      • (4)
        Removing contour overlaps (e.g. brain and brainstem, brain and target volume).
      • (5)
        Shrinking contours by specified number of voxels.
      • (6)
        Excluding CT slices from analysis that include artifacts potentially spreading into the contour (e.g. due to metal implants or marker seeds).
      During contour adaptation, all patient data were checked for consistency but not further adapted individually to avoid bias. The effect of the contour adaptation for representative organ segmentations of the brain, prostate, kidney (with and without calyces) as well as liver is displayed in Fig. 1. A full list of the organ segmentation adaption parameters as well as of the influence of the shrinking on the smaller brain structures is given in Supplement EC.
      Figure thumbnail gr1
      Fig. 1Exemplary CT slices illustrating the clinical contour adaptation. Position of the CT slices are indicated by the blue lines, clinical and adapted contours are shown in red and orange, respectively. For the kidney, exclusion of renal calyces is indicated in purple. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article).

      Analysis parameters

      Tissue parameters were quantified regarding the patient-specific average as the mean value over all CT voxels within each contour, and the cohort average calculated as the grand mean (thus the mean of the patient-specific average values) over all patients. Organ-specific inter-patient variations 2σinter, describing the variability within the patient, correspond to two standard deviations of the patient-specific average.
      Determined intra-patient standard deviations σintra,measured include both the biological tissue variation as well as the noise distribution σnoise. The tissue variation σintra thus follows from σintra=σintra,measured2-σnoise2. Tissue intra-patient variations were then averaged over the patient cohort as 2σintra¯.
      To assess the age dependence of the tissue parameters, the patient cohort was categorized in different age groups for which a change in tissue composition was expected: pediatrics (0-20y), adults (20-60y), and seniors (above 60y). Significance of differences in mean SPR as well as standard deviation between pediatric and non-pediatric patients were assessed by two-sample t tests with significance criterion of 5%. Results are visualized for the brain, where relevant changes were observed. Changes in the brain’s tissue type distribution were visualized in a histogram. To avoid bias from bin size and -position, a kernel density estimation with a linear kernel was fitted to the data, corresponding to an average shifted histogram [
      • Averaged S.DW.
      • Histogram S.
      Averaged shifted histogram. Wiley StatsRef stat ref online.
      ].

      Results

      The average RED, EAN and SPR as well as their respective intra- and inter-patient variation are listed in Table 1. In the following, results are illustrated exemplarily for the SPR as the clinically relevant parameter for particle therapy.
      In all tissues, SPR variation within the patients (2σintra¯) exceeded variation between them (2σinter), with 2σintra¯ larger than 2.1% and 1.3% in all organs in the pelvis and head, respectively. Largest intra-patient variation was observed in the kidney including the renal calyces (4.6%). Without the calyces, variation in the kidney was reduced to 3.6%. 2σinter exceeded 1% in all organs except for brain (0.8%) and prostate (0.6%). Averaged over all investigated organs and patients, intra- and inter-patient variation were 2.5% and 1.2%, respectively.
      For brain, lens, prostate and urine, the SPR distribution observed in the patient data incorporated the SPR obtained from ICRU46 (Fig. 2). In the brain, ICRU46 matches the median well, while for the lens, prostate and urine it is located at the periphery of the distribution. In both kidney and liver, ICRU36 exceed the distribution. For the kidney without renal calyces, median SPR determined here is 1.5% below ICRU46, while for the liver it is 0.7% higher.
      Figure thumbnail gr2
      Fig. 2Patient-specific mean stopping-power ratio (top) and its respective inter- and intra-patient variation 2σ (bottom) for different organs in the head and pelvic region. Kidney* corresponds to the kidney contours without renal calyces. Tissue information concluded from ICRU46 are depicted in blue. Whiskers in boxplot are defined by last values within 1.5x interquartile range.
      Differences between the different age groups for the brain are one magnitude below the observed age-specific variations and therefore minor (Table 2). An increase in SPR with adolescence was observed (p=0.05), whereas SPR decreased once patients reached seniority (p=0.04, Fig. 3). Patient-individual SPR variation (2σ) increased with age, differing significantly between pediatric and non-pediatric patients (p0.001).
      Table 2Brain mean tissue parameters and their relative inter- and intra-patient variation relative to water for different age groups. Variation of the effective atomic number is depicted in absolute numbers.
      Age groupRelative electron densityEffective atomic numberStopping-power ratio
      Mean2σintra in %2σinter in %Mean2σintra2σinterMean2σintra in %2σinter in %
      Young (<20 y)1.0301.30.97.650.280.091.0281.40.9
      Adult (20–60 y)1.0321.70.67.690.380.091.0301.80.7
      Senior(>60 y)1.0301.80.87.680.340.111.0281.90.9
      Figure thumbnail gr3
      Fig. 3Patient-individual mean stopping-power ratio (SPR) in the brain within the patient cohort (top) as well as the corresponding intra-patient SPR variation (2σ, bottom), summarized for the three age groups pediatrics (<20y), adults (20-60y) and seniors (>60y). Significant differences determined with a two-sample t-test are indicated above the boxplots. n.s.: not significant. Boxplot definition follows .

      Discussion

      Radiological tissue parameters SPR, RED and EAN were determined in vivo in the head and pelvis region of a large patient cohort. The results of this study expand and reassess the current standard data described in ICRU46. For all tissues analyzed, the average as well as the intra- and inter-patient variation was quantified. In the head, additional radiosensitive substructures of the brain, not specified by Woodard and White, were included. To our knowledge, this is the first organ-specific in vivo characterization performed in a large patient cohort using dual-energy CT.
      The determination of the variation in tissue parameters was the main objective of this work to extend on the ex vivo tissue characterization of Woodard and White. In our work, the influence of health, diet and hydration status is intrinsically minimized by analyzing a large number of patients.
      The inter- and intra-patient variation can also be considered in Monte Carlo transport simulations, where base materials are defined from tabulated human tissues. An improvement in accuracy of material classification can thus directly enhance the simulation quality [
      • Almeida I.P.
      • Schyns L.E.J.R.
      • Vaniqui A.
      • Van Der Heyden B.
      • Dedes G.
      • Resch A.F.
      • et al.
      Monte Carlo proton dose calculations using a radiotherapy specific dual-energy CT scanner for tissue segmentation and range assessment.
      ,
      • Permatasari F.F.
      • Eulitz J.
      • Richter C.
      • Wohlfahrt P.
      • Lühr A.
      Material assignment for proton range prediction in Monte Carlo patient simulations using stopping-power datasets.
      ]. The consideration of local interaction processes, such as in brachytherapy [
      • Mann-Krzisnik D.
      • Verhaegen F.
      • Enger S.A.
      The influence of tissue composition uncertainty on dose distributions in brachytherapy.
      ] as well as positron- [
      • Seravalli E.
      • Robert C.
      • Bauer J.
      • Stichelbaut F.
      • Kurz C.
      • Smeets J.
      • et al.
      Monte Carlo calculations of positron emitter yields in proton radiotherapy.
      ] and prompt-gamma emission [
      • Berthold J.
      • Khamfongkhruea C.
      • Petzoldt J.
      • Thiele J.
      • Hölscher T.
      • Wohlfahrt P.
      • et al.
      First-in-human validation of CT-based proton range prediction using prompt gamma imaging in prostate cancer treatments.
      ] during charged particle radiotherapy, is primarily depending on the tissue composition and therefore directly benefits from the improved accuracy. Furthermore, the parameter range may serve as a baseline for DECT-based quantification of damaged tissue, such as for the detection of liver fibrosis [
      • Lamb P.
      • Sahani D.V.
      • Fuentes-Orrego J.M.
      • Patino M.
      • Ghosh A.
      • Mendonca P.R.S.
      Stratification of patients with liver fibrosis using dual-energy CT.
      ,
      • Ohira S.
      • Kanayama N.
      • Toratani M.
      • Ueda Y.
      • Koike Y.
      • Karino T.
      • et al.
      Stereotactic body radiation therapy planning for liver tumors using functional images from dual-energy computed tomography.
      ] or cerebral edema [
      • van Ommen F.
      • Dankbaar J.W.
      • Zhu G.
      • Wolman D.N.
      • Heit J.J.
      • Kauw F.
      • et al.
      Virtual monochromatic dual-energy CT reconstructions improve detection of cerebral infarct in patients with suspicion of stroke.
      ].
      For both kidney and liver, large deviations between the derived parameter spread and those of Woodard and White were determined. For the kidney, ICRU46 states to have included renal cortex and medulla in their analysis [
      • White D.R.
      • Griffith R.V.
      • Wilson I.J.
      ICRU report 46.
      ], but leaving out the renal calyces, most likely due to the experiments being performed ex vivo. Here, the kidney was analyzed both with and without the urine-filled large calyces. As expected, the exclusion of the calyces increased the determined SPR towards the ICRU46 value (Fig. 2). The remaining overestimation in ICRU46 may be traced back to the sub-voxel-sized medullary collecting ducts containing urine in the patient, but not in the ex vivo analysis. As for the liver, ICRU46 explicitly states that a cirrhotic liver, appropriate for subjects suffering from alcoholism, was analyzed [
      • White D.R.
      • Griffith R.V.
      • Wilson I.J.
      ICRU report 46.
      ]. Cirrhosis is associated with a decrease in water and increase in lipid content [
      • Dju M.Y.
      • Mason K.E.
      • Filer L.J.
      Vitamin E (Tocopherol) in human tissues from birth to old age.
      ,
      • Reddy J.K.
      • Rao M.S.
      Lipid metabolism and liver inflammation. II. Fatty liver disease and fatty acid oxidation.
      ], resulting in a lower SPR. This is in agreement with the patient cohort having an increased SPR compared to ICRU46. The tissue parameters determined here for both kidney and liver can therefore be considered a more appropriate fit for alive patients.
      An analysis of SPR variation has been performed in a previous study by Wohlfahrt et al. [
      • Wohlfahrt P.
      • Möhler C.
      • Troost E.G.C.
      • Greilich S.
      • Richter C.
      Dual-energy computed tomography to assess intra- and inter-patient tissue variability for proton treatment planning of patients with brain tumor.
      ], applying an in-house DirectSPR implementation on DECT data from brain-tumor patients and using a contour covering the whole head. Bones and soft tissues were divided in analysis by applying specific CT number ranges. There, a trend regarding SPR variation with increasing age, similar to the one observed for the brain here, was observed.
      The determined age dependency of SPR in the brain is consistent with physiological changes reported in qualitative MRI review studies [
      • Hedman A.M.
      • van Haren N.E.M.
      • Schnack H.G.
      • Kahn R.S.
      • Hulshoff Pol H.E.
      Human brain changes across the life span: a review of 56 longitudinal magnetic resonance imaging studies.
      ,
      • Fjell A.M.
      • Walhovd K.B.
      Structural brain changes in aging: Courses, causes and cognitive consequences.
      ]. While Fjell et al. describe the physiological patterns of change as highly heterogeneous, which may correspond to the large inter-patient spread visible here (Fig. 3), Hedman et al. observed a general increase in white and grey matter during adolescence followed by a decrease of both in senior patients, corresponding to an increase and decrease in SPR, respectively. A similar trend of the mean SPR was observed here. At the same time, Hedman et al. reported an increase of cerebral spinal fluid (CSF) volume with age throughout life, leading to an increased intra-patient variability, which matches the observation made here (Fig. 3). A change in CSF volume would correspond to an increase of lower SPR values in the histogram in older patients, leading to a general shift of the distribution towards lower values (Fig. 4).
      Figure thumbnail gr4
      Fig. 4Histogram of the stopping-power ratio (SPR) distribution in the head patient cohort for pediatrics (red) and senior (blue) patients. The lines indicate the corresponding kernel density estimation of the histogram. (For interpretation of the references to color in this figure, the reader is referred to the web version of this article).
      There are multiple limitations that need to be considered when applying the presented data: While for all tissues in the head, analysis on many patients could be performed, the number of available contours in the pelvis region varied greatly. All data was collected from patients undergoing treatment for tumor diseases, which may influence both diet and hydration of the patients. An extrapolation to a fully healthy population therefore may not be valid without further investigation. At the same time, the presented data is directly applicable to patients undergoing radiotherapy.
      Both these limitations could be addressed by extending the patient cohort. With an increased interest in DECT-based treatment planning [
      • Paganetti H.
      • Beltran C.J.
      • Both S.
      • Dong L.
      • Flanz J.B.
      • Furutani K.M.
      • et al.
      Roadmap: proton therapy physics and biology.
      ,
      • Wohlfahrt P.
      • Richter C.
      Status and innovations in pre-treatment CT imaging for proton therapy.
      ], analysis could be extended to include data from other cancer treatment centers – or even radiological institutes – using a CT scan protocol matching the DirectSPR calibration. However, a major limiting factor is the labor-intensive manual organ delineation. This is most relevant for tissues typically not traversed by the treatment beam (such as liver and kidney), for which only few clinical contours are available. Recent innovations in automated delineation, in combination with an organ-specific delineation adaptation routine as described here to ensure contour quality, may prove useful to extend the data basis [
      • Dai X.
      • Lei Y.
      • Wang T.
      • Dhabaan A.H.
      • McDonald M.
      • Beitler J.J.
      • et al.
      Head-and-neck organs-at-risk auto-delineation using dual pyramid networks for CBCT-guided adaptive radiotherapy.
      ,
      • Balagopal A.
      • Kazemifar S.
      • Nguyen D.
      • Lin M.-H.
      • Hannan R.
      • Owrangi A.
      • et al.
      Fully automated organ segmentation in male pelvic CT images.
      ].
      Parameters for the delineation adaptation routine used here were selected to exclude tissue mixtures and artifacts from the contours without substantial volume loss compared to the original clinical contours. This is especially important for structures in which tissue parameters vary greatly between the core and outer layers, such as the lens [
      • Cogan D.G.
      Anatomy of lens and pathology of cataracts.
      ]. The large brain structure was investigated as a whole. For a further refinement, an accurate separation of grey matter, white matter and spinal fluid could be performed for patients receiving additional magnetic-resonance imaging [
      • Gordillo N.
      • Montseny E.
      • Sobrevilla P.
      State of the art survey on MRI brain tumor segmentation.
      ].
      Imaging artifacts were reduced by the organ-specific adaptation routine. Noise was reduced using an iterative image reconstruction as well as spatial frequency filtering in the calculation of the tissue parameters. Mean tissue parameters and their respective inter-patient variation were determined with a high quantitative accuracy, not affected by the remaining low, symmetrical noise. For the intra-patient variation, noise was excluded from analysis using a conservative estimation in phantom setups. Remaining minimal beam hardening artifacts extending from smaller bone structures into neighboring soft tissue could not be fully precluded. This may lead to a slight overestimation of intra-patient variation for small structures close to bone, such as the optical nerves in the parts traversing the dense skull bone. For the brain itself, this effect is negligible due to the large total volume.
      Based on the RED and EAN determined in this work, a further calculation of elemental tissue composition is possible [
      • Hünemohr N.
      • Paganetti H.
      • Greilich S.
      • Jäkel O.
      • Seco J.
      Tissue decomposition from dual energy CT data for MC based dose calculation in particle therapy.
      ]. However, as additional uncertainties are introduced by the choice and parameterization of tissue decomposition, this step is omitted here as to instead provide a comprehensive database for further analysis.

      Conclusion

      We presented an assessment of radiological tissue parameters of 10 organs in the head and pelvis, using a clinically validated DirectSPR implementation on dual-energy CT datasets. The determined organ-specific mean values as well as inter- and intra-patient variation expand and reassess the data described in ICRU46. This includes four brain substructures not previously described. Relevant variation in tissue parameters were determined and can be considered in various radiotherapeutic applications.

      Conflict of interest

      P. Wohlfahrt and C. Richter received individual funding as lecturer from Siemens Healthineers (2018), which are not related to this research study. OncoRay has an institutional research agreement with Siemens Healthineers in the field of DECT for particle therapy (2016–2020). Furthermore, OncoRay has an institutional agreement as reference center for dual-energy CT in radiotherapy as well as a software evaluation contract with Siemens Healthineers. For the present study, the authors received no financial support involved in the study design or materials used, nor in the collection, analysis and interpretation of data nor in the writing of the publication. The other authors report no conflict of interest.

      Appendix A. Supplementary material

      The following are the Supplementary data to this article:

      References

        • Rogers D.W.O.
        Fifty years of Monte Carlo simulations for medical physics.
        Phys Med Biol. 2006; 51: R287-R301https://doi.org/10.1088/0031-9155/51/13/R17
        • Bazalova M.
        • Graves E.E.
        The importance of tissue segmentation for dose calculations for kilovoltage radiation therapy.
        Med Phys. 2011; 38: 3039-3049https://doi.org/10.1118/1.3589138
        • Beaulieu L.
        • Carlsson Tedgren Å.
        • 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: 6208-6236https://doi.org/10.1118/1.4747264
        • Mann-Krzisnik D.
        • Verhaegen F.
        • Enger S.A.
        The influence of tissue composition uncertainty on dose distributions in brachytherapy.
        Radiother Oncol. 2018; 126: 394-410https://doi.org/10.1016/j.radonc.2018.01.007
        • Schaffner B.
        • Pedroni E.
        The precision of proton range calculations in proton radiotherapy treatment planning: experimental verification of the relation between CT-HU and proton stopping power.
        Phys Med Biol. 1998; 43: 1579-1592https://doi.org/10.1088/0031-9155/43/6/016
        • Yang M.
        • Zhu X.R.
        • Park P.C.
        • Titt U.
        • Mohan R.
        • Virshup G.
        • et al.
        Comprehensive analysis of proton range uncertainties related to patient stopping-power-ratio estimation using the stoichiometric calibration.
        Phys Med Biol. 2012; 57: 4095-4115https://doi.org/10.1088/0031-9155/57/13/4095
        • Schneider W.
        • Bortfeld T.
        • Schlegel W.
        Correlation between CT numbers and tissue parameters needed for Monte Carlo simulations of clinical dose distributions.
        Phys Med Biol. 2000; 45: 459-478https://doi.org/10.1088/0031-9155/45/2/314
        • Jones D.G.
        A realistic anthropomorphic phantom for calculating organ doses arising from external photon irradiation.
        Radiat Prot Dosim. 1997; 72: 21-29https://doi.org/10.1093/oxfordjournals.rpd.a032072
        • Hünemohr N.
        • Krauss B.
        • Tremmel C.
        • Ackermann B.
        • Jäkel O.
        • Greilich S.
        Experimental verification of ion stopping power prediction from dual energy CT data in tissue surrogates.
        Phys Med Biol. 2014; 59: 83-96https://doi.org/10.1088/0031-9155/59/1/83
        • Möhler C.
        • Wohlfahrt P.
        • Richter C.
        • Greilich S.
        Range prediction for tissue mixtures based on dual-energy CT.
        Phys Med Biol. 2016; 61: N268-N275https://doi.org/10.1088/0031-9155/61/11/N268
        • Parodi K.
        • Pönisch F.
        • Enghardt W.
        Experimental study on the feasibility of in-beam PET for accurate monitoring of proton therapy.
        IEEE Trans Nucl Sci. 2005; 52: 778-786https://doi.org/10.1109/TNS.2005.850950
        • Polf J.C.
        • Peterson S.
        • Ciangaru G.
        • Gillin M.
        • Beddar S.
        Prompt gamma-ray emission from biological tissues during proton irradiation: a preliminary study.
        Phys Med Biol. 2009; 54: 731-743https://doi.org/10.1088/0031-9155/54/3/017
        • Korhonen J.
        • Kapanen M.
        • Keyriläinen J.
        • Seppälä T.
        • Tenhunen M.
        A dual model HU conversion from MRI intensity values within and outside of bone segment for MRI-based radiotherapy treatment planning of prostate cancer.
        Med Phys. 2013; 41: 011704https://doi.org/10.1118/1.4842575
        • Sisniega A.
        • Zbijewski W.
        • Badal A.
        • Kyprianou I.S.
        • Stayman J.W.
        • Vaquero J.J.
        • et al.
        Monte Carlo study of the effects of system geometry and antiscatter grids on cone-beam CT scatter distributions.
        Med Phys. 2013; 40: 051915-1-051915-18https://doi.org/10.1118/1.4801895
        • Woodard H.Q.
        • White D.R.
        The composition of body tissues.
        Br J Radiol. 1986; 59: 1209-1218https://doi.org/10.1259/0007-1285-59-708-1209
        • White D.R.
        • Booz J.
        • Griffith R.V.
        • Spokas J.J.
        • Wilson I.J.
        ICRU report 44.
        J Int Comm Radiat Units Meas. 1989; os23https://doi.org/10.1093/jicru/os23.1.Report44
        • White D.R.
        • Griffith R.V.
        • Wilson I.J.
        ICRU report 46.
        J Int Comm Radiat Units Meas. 1992; os24https://doi.org/10.1093/jicru/os24.1.Report46
        • Valentin J.
        Basic anatomical and physiological data for use in radiological protection: reference values.
        Ann ICRP. 2002; 32: 1-277https://doi.org/10.1016/S0146-6453(03)00002-2
        • White D.R.
        • Widdowson E.M.
        • Woodard H.Q.
        • Dickerson J.W.T.
        The composition of body tissues. (II) Fetus to young adult.
        Br J Radiol. 1991; 64: 149-159https://doi.org/10.1259/0007-1285-64-758-149
        • Van Elmpt W.
        • Landry G.
        • Das M.
        • Verhaegen F.
        Dual energy CT in radiotherapy: current applications and future outlook.
        Radiother Oncol. 2016; 119: 137-144https://doi.org/10.1016/j.radonc.2016.02.026
        • Paganetti H.
        • Beltran C.J.
        • Both S.
        • Dong L.
        • Flanz J.B.
        • Furutani K.M.
        • et al.
        Roadmap: proton therapy physics and biology.
        Phys Med Biol. 2020; https://doi.org/10.1088/1361-6560/abcd16
        • Wohlfahrt P.
        • Möhler C.
        • Hietschold V.
        • Menkel S.
        • Greilich S.
        • Krause M.
        • et al.
        Clinical implementation of dual-energy CT for proton treatment planning on pseudo-monoenergetic CT scans.
        Int J Radiat Oncol Biol Phys. 2017; 97: 427-434https://doi.org/10.1016/j.ijrobp.2016.10.022
        • Peters N.
        • Wohlfahrt P.
        • Hofmann C.
        • Möhler C.
        • Menkel S.
        • Tschiche M.
        • et al.
        Reduction of clinical safety margins in proton therapy enabled by the clinical implementation of dual-energy CT for direct stopping-power prediction.
        Radiother Oncol. 2022; 166: 71-78https://doi.org/10.1016/j.radonc.2021.11.002
        • Möhler C.
        • Russ T.
        • Wohlfahrt P.
        • Elter A.
        • Runz A.
        • Richter C.
        • et al.
        Experimental verification of stopping-power prediction from single- and dual-energy computed tomography in biological tissues.
        Phys Med Biol. 2018; 63: 025001https://doi.org/10.1088/1361-6560/aaa1c9
        • Grant K.L.
        • Flohr T.G.
        • Krauss B.
        • Sedlmair M.
        • Thomas C.
        • Schmidt B.
        Assessment of an advanced image-based technique to calculate virtual monoenergetic computed tomographic images from a dual-energy examination to improve contrast-to-noise ratio in examinations using iodinated contrast media.
        Invest Radiol. 2014; 49: 586-592https://doi.org/10.1097/RLI.0000000000000060
        • Lüders E.
        • Steinmetz H.
        • Jäncke L.
        Brain size and grey matter volume in the healthy human brain.
        NeuroReport. 2002; 13: 2371-2374https://doi.org/10.1097/00001756-200212030-00040
        • Vinod S.K.
        • Jameson M.G.
        • Min M.
        • Holloway L.C.
        Uncertainties in volume delineation in radiation oncology: a systematic review and recommendations for future studies.
        Radiother Oncol. 2016; 121: 169-179https://doi.org/10.1016/j.radonc.2016.09.009
        • Averaged S.DW.
        • Histogram S.
        Averaged shifted histogram. Wiley StatsRef stat ref online.
        Wiley, 2019: 1-9 (https://doi.org/10.1002/9781118445112.stat00406.pub2)
        • Almeida I.P.
        • Schyns L.E.J.R.
        • Vaniqui A.
        • Van Der Heyden B.
        • Dedes G.
        • Resch A.F.
        • et al.
        Monte Carlo proton dose calculations using a radiotherapy specific dual-energy CT scanner for tissue segmentation and range assessment.
        Phys Med Biol. 2018; 63https://doi.org/10.1088/1361-6560/aabb60
        • Permatasari F.F.
        • Eulitz J.
        • Richter C.
        • Wohlfahrt P.
        • Lühr A.
        Material assignment for proton range prediction in Monte Carlo patient simulations using stopping-power datasets.
        Phys Med Biol. 2020; 65https://doi.org/10.1088/1361-6560/ab9702
        • Seravalli E.
        • Robert C.
        • Bauer J.
        • Stichelbaut F.
        • Kurz C.
        • Smeets J.
        • et al.
        Monte Carlo calculations of positron emitter yields in proton radiotherapy.
        Phys Med Biol. 2012; 57: 1659-1673https://doi.org/10.1088/0031-9155/57/6/1659
        • Berthold J.
        • Khamfongkhruea C.
        • Petzoldt J.
        • Thiele J.
        • Hölscher T.
        • Wohlfahrt P.
        • et al.
        First-in-human validation of CT-based proton range prediction using prompt gamma imaging in prostate cancer treatments.
        Int J Radiat Oncol Biol Phys. 2021; 111: 1033-1043https://doi.org/10.1016/j.ijrobp.2021.06.036
        • Lamb P.
        • Sahani D.V.
        • Fuentes-Orrego J.M.
        • Patino M.
        • Ghosh A.
        • Mendonca P.R.S.
        Stratification of patients with liver fibrosis using dual-energy CT.
        IEEE Trans Med Imaging. 2015; 34: 807-815https://doi.org/10.1109/TMI.2014.2353044
        • Ohira S.
        • Kanayama N.
        • Toratani M.
        • Ueda Y.
        • Koike Y.
        • Karino T.
        • et al.
        Stereotactic body radiation therapy planning for liver tumors using functional images from dual-energy computed tomography.
        Radiother Oncol. 2020; 145: 56-62https://doi.org/10.1016/j.radonc.2019.12.002
        • van Ommen F.
        • Dankbaar J.W.
        • Zhu G.
        • Wolman D.N.
        • Heit J.J.
        • Kauw F.
        • et al.
        Virtual monochromatic dual-energy CT reconstructions improve detection of cerebral infarct in patients with suspicion of stroke.
        Neuroradiology. 2021; 63: 41-49https://doi.org/10.1007/s00234-020-02492-y
        • Dju M.Y.
        • Mason K.E.
        • Filer L.J.
        Vitamin E (Tocopherol) in human tissues from birth to old age.
        Am J Clin Nutr. 1958; 6: 50-60https://doi.org/10.1093/ajcn/6.1.50
        • Reddy J.K.
        • Rao M.S.
        Lipid metabolism and liver inflammation. II. Fatty liver disease and fatty acid oxidation.
        Am J Physiol – Gastrointest Liver Physiol. 2006; 290: 852-858https://doi.org/10.1152/ajpgi.00521.2005
        • Wohlfahrt P.
        • Möhler C.
        • Troost E.G.C.
        • Greilich S.
        • Richter C.
        Dual-energy computed tomography to assess intra- and inter-patient tissue variability for proton treatment planning of patients with brain tumor.
        Int J Radiat Oncol Biol Phys. 2019; : 1-10https://doi.org/10.1016/j.ijrobp.2019.06.2529
        • Hedman A.M.
        • van Haren N.E.M.
        • Schnack H.G.
        • Kahn R.S.
        • Hulshoff Pol H.E.
        Human brain changes across the life span: a review of 56 longitudinal magnetic resonance imaging studies.
        Hum Brain Mapp. 2012; 33: 1987-2002https://doi.org/10.1002/hbm.21334
        • Fjell A.M.
        • Walhovd K.B.
        Structural brain changes in aging: Courses, causes and cognitive consequences.
        Rev Neurosci. 2010; 21: 187-221https://doi.org/10.1515/REVNEURO.2010.21.3.187
        • Wohlfahrt P.
        • Richter C.
        Status and innovations in pre-treatment CT imaging for proton therapy.
        Br J Radiol. 2020; 93: 20190590https://doi.org/10.1259/bjr.20190590
        • Dai X.
        • Lei Y.
        • Wang T.
        • Dhabaan A.H.
        • McDonald M.
        • Beitler J.J.
        • et al.
        Head-and-neck organs-at-risk auto-delineation using dual pyramid networks for CBCT-guided adaptive radiotherapy.
        Phys Med Biol. 2021; 66https://doi.org/10.1088/1361-6560/abd953
        • Balagopal A.
        • Kazemifar S.
        • Nguyen D.
        • Lin M.-H.
        • Hannan R.
        • Owrangi A.
        • et al.
        Fully automated organ segmentation in male pelvic CT images.
        Phys Med Biol. 2018; 63: 245015https://doi.org/10.1088/1361-6560/aaf11c
        • Cogan D.G.
        Anatomy of lens and pathology of cataracts.
        Exp Eye Res. 1962; 1: 291-295https://doi.org/10.1016/S0014-4835(62)80013-X
        • Gordillo N.
        • Montseny E.
        • Sobrevilla P.
        State of the art survey on MRI brain tumor segmentation.
        Magn Reson Imaging. 2013; 31: 1426-1438https://doi.org/10.1016/j.mri.2013.05.002
        • Hünemohr N.
        • Paganetti H.
        • Greilich S.
        • Jäkel O.
        • Seco J.
        Tissue decomposition from dual energy CT data for MC based dose calculation in particle therapy.
        Med Phys. 2014; 41: 061714https://doi.org/10.1118/1.4875976