A practical approach to the introduction of spectral imaging into a large UK acute care teaching hospital

Transcription

1 A practical approach to the introduction of spectral imaging into a large UK acute care teaching hospital Robert Loader Clinical Scientist Clinical & Radiation Physics Directorate of Healthcare Science & Technology Robert.loader@nhs.net

2

3 Viking:The name 'Viking' comes from a language called 'Old Norse' and means 'a pirate raid'. People who went off raiding in ships were said to be 'going Viking'.

4 Plymouth, Devon

5 Derriford Hospital Largest hospital Trust in South West Peninsular Budget of ~ 380M (5000M NOK) Deficit of ~ 40M 900+ Beds General hospital Services to a population of ~0.5M Specialist services up to 2M including Kidney transplant Pancreatic cancer surgery Neurosurgery Cardiothoracic surgery Bone marrow transplant Hepatobiliary surgery Neonatal intensive care and high risk obstetrics Plastic surgery Liver transplant evaluation Stereotactic Radiosurgery Radiology academy Dedicated Interventional Radiology theatres

6 Overview Introduction to problem What is Dual Energy Imaging? Material decomposition (the basics) Hardware solutions Practical evaluations on a GE HD750 CT scanner Iodine quantification Kidney stone analysis Cardiac perfusion maps Advice to the Dr s (Lead Radiologist for CT Sig)

7 Introduction to the Problem Oh Dear Rob. We are getting a new scanner next week, its got Dual Energy cool eh Plymouth 2009 Me Radiologist

8 What are we buying? How does it work? What are the claims? What training do Physicists/Radiologists/Radiographers need? What are the doses? What are the limitations? How will it inform patient management? Can we use it safely? How long do we need to commission this system? How do we optimise?

9 Can we use it yet??

10 What is Dual Energy imaging in CT? Attenuation of x-ray beam (linear attenuation coefficient, µ) within voxels Different µ generates contrast between tissue types Attenuation of contrast media and bone overlaps Variable density and composition of bone Variable concentration of contrast in vessels / organs How can we differentiate between them? Or other tissue types? The goal of dual energy methods is to simulate the energy dependence of attenuation to better describe it This not only has the potential to remove beam hardening artefacts, but more completely describe materials in the area of interest

11 Attenuation and X-ray energy

12 The method is not new!

13 Problem FBP I I exp x, y, z 0 dl Reality Eexp x, y, z E I I 0, dlde The line integral What we need What we have P x, y, z dl P x, y, z, E dl

14 Compton Effect Can be modelled by the Klein Nishina formulae ln ln c Attenuation data sourced from National Institute of Standards (NIST)

15 Photoelectric Effect Can be modelled by 1/E 3 p 1 E 3 Attenuation data sourced from National Institute of Standards (NIST)

16 Vector space a b c d f g A(E) B(E) C(E) D(E) F(E) G(E) h i j H(E) I(E) J(E) At energy E 0 For any energy Constants Functions

17 Vector Space Photoelectric coefficient, this is the energy independent photoelectric coefficient unique to each material and density Compton base, this is the Compton energy dependent function, it is independent of material Klein Nishina ( E) a a p p c c Photoelectric base, this is the photoelectric energy dependent function, it is independent of material p 1 E 3 Compton coefficient, this is the energy independent Compton coefficient unique to each material and density

18 Result (a p,a c ) 1,1 (a p,a c ) 2,1 (a p,a c ) 3,1 (a p,a c ) 1,2 (a p,a c ) 2,2 (a p,a c ) 3,2 (a p,a c ) 1,3 (a p,a c ) 2,3 (a p,a c ) 3,3

19 Basis materials c p I B A de A A E I I W W I I exp ) 0( 1 1 de A A E I I W W I I exp ) 0( 2 2 c p W D C Mass of iodine required to produce this attenuation Mass of water required to produce this attenuation

20 KeV Images I I ( E ) exp 0 0 A A p p c c Source GSI White Paper

21 Applications Base pair images (e.g. iodine/water, water/calcium, iodine/calcium) Synthesised monochromatic images Spectral curves/scatter plots/concentration analysis Material subtraction

22 Some potential clinical uses Virtual non contrast studies (iodine mathematically subtracted) Metal artefact reduction (by selection of KeV synthesised image) ROI analysis is this lesion blood/cyst/iodinated? What's the composition of this renal stone? Perfusion maps (Heart and lung)

23 Examples

24 kev imaging 40keV 55keV 75keV 100keV

25 Multiple scans Hardware solutions (1)

26 Two Tubes Hardware solutions (2)

27 McCollough C et al. Dual and multi-energy CT: Principles, technical approaches, and clinical applications. Radiology 2015;276: RSNA publications 2015 Two tube systems

28 Rapid KVp switching Hardware solutions (3)

29 Energy sensitive detectors Hardware solutions (4)

30 Photon counting detectors Hardware solutions (5)

31 McCollough C et al. Dual and multi-energy CT: Principles, technical approaches, and clinical applications. Radiology 2015;276: RSNA publications 2015

32 McCollough C et al. Dual and multi-energy CT: Principles, technical approaches, and clinical applications. Radiology 2015;276: RSNA publications (SIMPLIFIED) Solution + - Dual Spiral Sequence Dual Source scanner Rapid KVp switching Multilayer detector Photon Counting No special hardware required (can be performed on any CT scanner) Tube current and filtration optimised to maximise CNR and spectrum separation Near simultaneous data acquisition between low and high KVp datasets simultaneous data acquisition between low and high KVp datasets Permits energy specific noise rejection. Facilitates new imaging approaches (e.g. k-edge subtraction) Patient motion between scans may cause severe degradation of images 90 phase shift between low and high datasets. Detector cross talk requiring special scatter rejection correction Relatively high overlap of the energy spectra Relatively high overlap of the energy spectra. Difference in noise levels between datasets? Technological realisation

33 Practical implementation Test basic function before use (e.g. radiation dose) Determination of effective atomic number Kidney stone analysis Iodine concentration for perfusion maps

34 How does the Scanner calculate the effective atomic number? The effective atomic number is defined in terms of the attenuation signature for a material. That is, if the material that we scan behaves like a periodic element when an atomic number X, then we say that the effective atomic number of that material is X

35 How does the Scanner calculate the effective atomic number? Scanner uses the ratio of total attenuation at 70KeV and 120KeV. It matches this identified ratio to a point on the curve of known elements of known atomic number. This is then used as a calibration curve to calculate the effective atomic number of unknown compounds, elements and mixtures.

36 Scanner tells us Z, use NIST to get the attenuation ratio of 70KeV/120KeV

37 Example (for Iodine mixtures)

38 Example (for Iodine mixtures)

39 Example (for Iodine mixtures)

40 The materials Water Carbon (graphite) Air Aluminium Chalk Iodine concentrations Our Chemistry set

41 Help from the Chemists!

42 Elements and mixtures Materials: 1. Water 2. Iodine 3. Carbon 4. Chalk

43 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Hf= 65 W = 70 Re = 72 Os = 74 Pt = 78 Au = 81 Pb = 88 K-edge of element disrupting attenuation ratio Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements

44 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Water (H2O)

45 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Carbon (graphite bush)

46 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Chalk (CaCO3)

47 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Aluminium

48 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Air (78N21O1AR)

49 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Water (H2O) Carbon (graphite bush) Chalk (CaCO3) Aluminium Air (78N21O1AR)

50 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Air 0 Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Water (H2O) Carbon (graphite bush) Chalk (CaCO3) Aluminium Air (78N21O1AR)

51 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Chemistry mixtures (Various)

52 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) dz~2.2? Scanner measurement dz~4.5? Our error (% Ioversol by weight) Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Ioversol (opiray) C18H24I3N3O4

53 Effective Atomic Number (Z) ROI analysis Effective atomic number (Z) measures for a range of elements, compounds and mixtures utilising an HD750 discovery series CT scanner (GE Healthcare) Attenuation Coefficient Ratio (70keV/120keV) NIST NIST data Elements Chemistry mixtures (Various) Water (H2O) Carbon (graphite bush) Chalk (CaCO3) Aluminium Air (78N21O1AR) Ioversol (opiray) C18H24I3N3O4

54 Local evaluation of kidney stone separation Renal stone characterisation Z value: GSI kidney stone software Hounsfield unit: 80 kvp and 140kvp Two methods ROI ( region of interest): Effective Z and HU determined by ROI placement over largest area of stone RC (report cursor) WW set at 1 and measured effective Z and HU of the stone as the stone appeared when the WL was increased Reference standard Fourier transform Infrared analysis : 8 stone groups

55 Local evaluation of kidney stone separation

56 Z (ROI) The use? Local work Cystine Uric_acid Struvite_stone Calcium_Oxalate Calcium_Phosphate Predominant_Calcium_Oxalate Predominant_Calcium_Phosphate Brushite Type of Stone

57 RC z GSI effective Z analysis GSI effective z ROI z GSI Effective Z, separation of Uric acid stones Cystine Uric acid Struvite Calcium Oxalate Calcium Phsphate Predom Ca Oxalate Predom Ca Phosphate Brushite

58 RC 80KVp conventional Scan Stone characterisation 80KVp Cystine Uric Acid Struvite Ca Oxalate Less separation of Uric acid than GSI effective Z ROI HU Ca Phsphate Pedom Ca Oxalate Predom Ca Phosphate Brushite

59 RC HU Conventional Scan (140KVp) 140 KVp Cystine Uric acid Struvite Ca Oxalate Ca Phosphate Predom Ca Oxalate Predom Ca Phosphate Less separation of Uric acid at 140KVp cf 80KVp & GSI effective Z ROI HU

60 Kidney stone Conclusions Only Uric acid separation possible Effective Z appears to increase separation from HU at 80 and 140KV To bear in mind, the real difference in effective atomic number among the different non uric acid stone is small (especially for mixed stones), so can we accurately predict atomic number and the differences? Further large sample analysis needed.

61 Validation of Iodine concentration for cardiac perfusion maps

62 Discussion With the exception of air, for the elements scanned, the CT scanner s measure of atomic number was in reasonable agreement with known elements. Chemistry mixtures and iodine concentrations broadly followed the expected signature curve. Further work is required to assess a possible saturation in scanner measured Z for high iodine concentrations. Further work is required to calculate total measurement uncertainty before firm conclusions can be made.

63

64 Reversible ischaemia

65 Validation of Iodine concentration for cardiac perfusion maps

66

67 Results

68

69 Gemstone spectral imaging (GSI) accurately quantifies water/iodine mixtures within this phantom. GSI appears to be free from the effects of beam hardening when assessed in this specific phantom. Low kev monoenergetic images allow significant increases in contrast, however at reduced CNR when compared to conventional polychromatic imaging.

70 Can we use it yet??

71 Advice to the Dr s Of course you can use it providing.. 1 You are satisfied the existing research base is extensive enough to introduce as established use. 2 You understand the differential diagnosis and its influence on onward patient management. 3 Where you are unsure of the benefit to patient care, the certainty of result or the limitations of the application the proposed protocol is assessed as part of a research trial.

72 Local policy surrounding use

73 Acknowledgements Nicholas Keat (Imanova Ltd., London) Professor Carl Roobottom (Plymouth University, Plymouth Hospitals NHS Trust) Mr Matthew Dixon (Plymouth Hospitals NHS Trust) Dr Tinu Purayil (Plymouth Hospitals NHS Trust) Dr Benjamin Clayton (Plymouth Hospitals NHS Trust)