State of the art and future development for standardized estimation of organ doses in CT March 2015 William J. O Connel, Dr. Ph, Senior Medical Physicist Imagination at work.
Agenda Introduction Duke Florida UCLA / MD Anderson RPI AAPM TG 246 Conclusion 2
Introduction 3
Introduction What is the radiation dose? Multi-faceted calculation requiring knowledge of energy deposited in defined mass (organ / tissue) Is the procedure safe? (risk) Growing interest in role of organ dose as descriptor of risk 4
National Radiological Protection Board Monte Carlo simulations of calculated x-ray spectra in an adult, hermaphrodite, mathematical model (MIRD) 75 scanners (out of 200) operating in the UK at the time. 23 data sets produced for surveyed scanner models Original data from contiguous axial scans 5
National Radiological Protection Board 6
ImPACT Estimating patient dose on current CT scanners: Results of the ImPACT* CT dose survey M.A. Lewis, S. Edyvean, S.A. Sassi, H. Kiremidjian, N. Keat and A.J. Britten. ImPACT, Medical Physics, St. George's Hospital, London 7
Introduction Estimated effective doses are not patient specific DLP / k-factor method (k is not scanner specific) Effective Dose in obese patients is problematic 8
Overview 9
Introduction Limitations to existing patient dose metrics CTDI VOL is a useful benchmarking tool but is not ideal indicator of organ dose and radiation risk Broader beam widths Bow-tie filters Variable Pitch Non-contiguous slices 10
Introduction Researchers are looking at many aspects of organ dose estimation Validated Monte Carlo modeling Computational Phantoms Tube Current Modulation Obese Patients Roadmap to Organ Dose in computed tomography 11
Florida 12
Florida WE Bolch AAPM Imaging Symposium July 2014 13
Florida WE Bolch AAPM Imaging Symposium July 2014 Stylized (Mathematical) Phantom ORNL stylized adult phantom Flexible Anatomically unrealistic Errors 14
Florida WE Bolch AAPM Imaging Symposium July 2014 Voxel (Tomographic) Phantom Constructed from patient acquisitions Anatomically realistic Not flexible 15
Florida WE Bolch AAPM Imaging Symposium July 2014 Hybrid Phantom Nurbs (non-uniform rational B-spline) mathematical model used for generating / representing curves and surfaces. Realistic Flexible 16
Florida WE Bolch AAPM Imaging Symposium July 2014 Hierarchy of phantom morphometric categories Patient-Specific patient match: individual patient morphology Patient-Sculpted patient match: height, weight, body contour Patient-Dependent patient match: nearest height and weight Reference 50 th percentile individual, patient matching by age only 17
Florida - Long et al, Med. Phys. 40 (1), January 2013 18
Florida - Long et al, Med. Phys. 40 (1), January 2013 Benchmark Monte Carlo simulations against anthropomorphic phantoms SOMATOM Sensation 16 multidetector CT scanner Multiple axial and helical acquisitions UF computational adult male reference hybrid phantom 19
Florida - Long et al, Med. Phys. 40 (1), January 2013 Monte Carlo radiation transport code, MCNPX version 2.6. SPEC78 spectrum generation program Bow-tie filter and over-ranging UF Series-B 9-month-old voxel phantom fiber-optic coupled plastic scintillator dosimetry (PSD) system 20
Florida - Long et al, Med. Phys. 40 (1), January 2013 21
Florida - Long et al, Med. Phys. 40 (1), January 2013 22
Florida - Long et al, Med. Phys. 40 (1), January 2013 On average, organ doses from the Monte Carlo simulations agreed with physically measured doses within 8%-9% for axial and helical imaging of the reference adult phantom Agreement is within 6%-7% for the 9-month old child Individual organ doses were found to be within 15% of measurements of organ dose for both phantoms 23
Duke 24
Duke Zhang et al, Med. Phys. 39 (6), June 2012 25
Duke Zhang et al, Med. Phys. 39 (6), June 2012 How are dose results affected by choice of computational anthropomorphic phantom? What uncertainties exist in the estimation of dose with different types of phantoms? Organ doses, effective doses, risk indices, and conversion coefficients to effective dose and risk index were estimated for ten body and three neurological examination categories 26
Duke Zhang et al, Med. Phys. 39 (6), June 2012 1. Male and Female Extended Cardiac-Torso (XCAT) 2. ICRP No. 110 reference male and female phantoms 3. Impact Group phantoms 4. CT-Expo 27
Duke Zhang et al, Med. Phys. 39 (6), June 2012 28
Duke Zhang et al, Med. Phys. 39 (6), June 2012 29
Duke Zhang et al, Med. Phys. 39 (6), June 2012 XCAT Hybrid Phantoms Visible Human anatomical data National Library of Medicine NURBS based phantoms modified to match ICRP 89 reference values Brains modeled separately on MRI models 30
Duke Zhang et al, Med. Phys. 39 (6), June 2012 ICRP 110 Voxelized Phantoms Tomographic data of individuals whose body height matched reference values in ICRP Publication 89 Radiosensitive organs were directly segmented from tomographic data 31
Duke Zhang et al, Med. Phys. 39 (6), June 2012 ImPACT Phantoms Stylized mathematical phantom 208 contiguous 5 cm slabs extending from upper legs to head NRPB-R186 32
Duke Zhang et al, Med. Phys. 39 (6), June 2012 CT-Expo Phantoms Stylized mathematical phantom Design characteristics of MIRD- 5 phantom ADAM and EVA GSF ICRP Publication 23 33
Duke Zhang et al, Med. Phys. 39 (6), June 2012 34
Duke Zhang et al, Med. Phys. 39 (6), June 2012 35
Duke - Tian et al, Radiology 270 (2), February 2013 36
Duke - Tian et al, Radiology 270 (2), February 2013 Validated Monte Carlo simulations performed on 42 pediatric patient models (normals) Organ dose estimates for routine chest and abdominopelvic examinations Feasible to estimate patientspecific organ dose with knowledge of patient size and CTDI VOL 37
Duke - Tian et al, Radiology 270 (2), February 2013 38
Duke - Tian et al, Radiology 270 (2), February 2013 Multi scanner study Lightspeed VCT and SOMATOM Definition Flash CTDI VOL is used as an index of scanner radiation output Calculate CTDI VOL conversion factor (h O, S, P ) specific to each organ, scanner and patient model CTDI VOL determined with 100 mm chamber and 16-cm phantom CTDI VOL conversion factor showed exponential relationship with average patient diameter 39
Duke - Tian et al, Radiology 270 (2), February 2013 40
Duke - Tian et al, Radiology 270 (2), February 2013 41
Duke - Tian et al, Radiology 270 (2), February 2013 42
Duke - Tian et al, Radiology 270 (2), February 2013 43
Duke - Tian et al, Radiology 270 (2), February 2013 44
Duke - Tian et al, Radiology 270 (2), February 2013 45
UCLA / MD Anderson 46
UCLA - Khatonabadi et al, Med. Phys. 39 (8), August 2012 47
UCLA - Khatonabadi et al, Med. Phys. 39 (8), August 2012 Most methods estimating patient dose from computed tomography are based on fixed tube current scans A growing number of CT scans are performed with tube current modulation (TCM) Detailed TCM data is difficult to obtain What is accuracy of organ dose estimates obtained using methods that approximate detailed TCM function? 48
UCLA - Khatonabadi et al, Med. Phys. 39 (8), August 2012 MCNPX (Monte Carlo N- Particle extended v2.6.0) Two MDCT scanners: Sensation 64 and LightSpeed 16 Twenty adult female chest voxelized models Twenty pediatric female models (whole body) 49
UCLA - Khatonabadi et al, Med. Phys. 39 (8), August 2012 For each patient model, detailed TCM function was extracted from the raw projection data Over-ranging region can be determined from start and end locations of the image data and locations of x-ray beam on and x-ray beam off 50
UCLA - Khatonabadi et al, Med. Phys. 39 (8), August 2012 51
UCLA - Khatonabadi et al, Med. Phys. 39 (8), August 2012 Longitudinal approximated TCM function obtained from the image data is reasonable surrogate to detailed TCM function for use in Monte Carlo dose simulations. Longitudinal approximated TCM function only represents the z-axis modulation of the TCM algorithm and it does not capture the over-ranging information that the detailed TCM function Results suggest angular modulation has a stronger effect on smaller peripheral organs (breasts) compared to larger and more central organs (lungs). 52
RPI 53
RPI - Ding et al, Phys. Med. Biol. 57 (9), May 2012 54
RPI - Ding et al, Phys. Med. Biol. 57 (9), May 2012 Study the effect of obesity on the calculated radiation dose to organs and tissues Visceral adipose tissue (VAT) Developed BMI-adjustable phantoms ( Range 23.5 46.4) Subcutaneous adipose tissue (SAT) 55
RPI - Ding et al, Phys. Med. Biol. 57 (9), May 2012 SAT layer is added to phantom in the space between the body surface and internal organs Thickness of adipose tissue is only considered for attenuation properties Dose to adipose tissue is not estimated 56
RPI - Ding et al, Phys. Med. Biol. 57 (9), May 2012 No data to estimate the effect of VAT on internal organ placement / deformity Internal organ size and VAT volume held constant for all BMI settings VAT density is corrected for obesity with waist circumference (WC) 57
RPI 3D - BMI Adjustable Phantoms 58
RPI - Ding et al, Phys. Med. Biol. 57 (9), May 2012 Data validated against anthropomorphic data from literature (Ogden 2004) AP and LAT measurements performed at mid-chest and mid-abdomen 59
RPI - Ding et al, Phys. Med. Biol. 57 (9), May 2012 Good correlation between AP and LAT measurements and Ogden data BMI-adjustable phantoms are realistic representation of overweight and obese patients for the purpose of estimating CT imaging doses 60
RPI - Ding et al, Phys. Med. Biol. 57 (9), May 2012 61
RPI - Ding et al, Phys. Med. Biol. 57 (9), May 2012 62
AAPM Task Group 246 63
AAPM TG 246 William Pavlicek Chair Daniel Bednarek Wesley Bolch Dianna Cody Frank Dong Sue Edyvan Aaron Jones Cynthia McCollough Ed McDonagh Michael McNitt-Gray Donald Miller Donald Peck Madan Rehani Ehsan Samei Mark Supanich 64
AAPM TG 246 Phantom Organ Dose True Patient Dose Enhanced SSDE SSDE Dose Page CTDI VOL DLP 65
AAPM TG 246 MC calculation tools have been validated numerous times with physical measurements and are considered capable of accuracy equal to measured values Possible to assemble tables of dose coefficients to convert individual episodes of patient exposures to dose Approach would reduce need for long duration MC computations for each patient MC with a near matched scanning device and patient matched phantom computation 66
Conclusion 67
Conclusion Dose to radiosensitive organs is a useful basis for estimating metrics related to risk Organ dose is more informative than CTDI VOL, DLP or Effective Dose Better accounts for scanner differences Better accounts for variability in patient size Better accounts for changes in target region Better accounts for tube current modulation 68
Conclusion Not quite ready for implementation How many computational phantoms are required? What level of accuracy is required? ± 20%, ± 35% THANK YOU! 69