Click here to load reader

Computed Tomography IIComputed Tomography II C · PDF fileConventional CT Dosimetry • Computed Tomography Dose Index (CTDI) •Developed in the context of axial CT-Averaggp pe multiple

  • View
    222

  • Download
    1

Embed Size (px)

Text of Computed Tomography IIComputed Tomography II C · PDF fileConventional CT Dosimetry •...

  • Computed Tomography IIComputed Tomography IIComputed Tomography IIC-Arm Cone-Beam CT:

    Computed Tomography IIC-Arm Cone-Beam CT:

    Principles and ApplicationsPrinciples and Applications

    Jeff Siewerdsen1 and Guang-Hong Chen21. Department of Biomedical Engineering, Johns Hopkins University

    2. Department of Medical Physics, University of Wisconsin

    Johns Hopkins UniversitySchools of Medicine and Engineering

    University of WisconsinInstitutes for Medical Research

  • Overview

    Part 1: (Siewerdsen)- Cone-beam CT image quality- Radiation doseRadiation dose- Applications (non-vascular)- Sustained applause

    Part 2: (Chen)- 3D CBCT reconstruction- Artifacts- Artifacts- Applications (cardiovascular)- Thunderous ovation

  • Not Your Mamas C-ArmNot Your Mamas C-ArmSome Essential Science and Practicalities

    for the New Generation of Cone-Beam CT-Capable CsSome Essential Science and Practicalities

    for the New Generation of Cone-Beam CT-Capable Cs

    Jeff Siewerdsen, PhDDepartment of Biomedical EngineeringDepartment of Biomedical Engineering

    Johns Hopkins University

    Johns Hopkins UniversitySchools of Medicine and Engineering

  • The New C-Arm

    Fluoroscopy + Cone-Beam CT3D imaging capabilit- 3D imaging capability

    3D filtered backprojection (FDK)FOV ~(20x20x20) cm3from a single half-rotationg

    Flat-Panel Detector- Replacement to XRII

    Larger FOVLarger FOVBetter 2D image qualityDistortionless

    - High-performance CBCTHigh performance CBCTSub-mm spatial resolutionSoft-tissue visibility

  • C-Arms for IGIKey Characteristics Real-time

    (or near-real-time)(or near real time)

    Radiation dose~1/10 1/2 of Dx CT

    Sub-mm resolution

    Soft-tissue visibility

  • Mobile Isocentric C-Arm

    Siemens PowerMobil

    MotorizedOrbit Replace XRII withFlat-Panel Detector

    GeometricControl System

    Calibration

    Tube + CollimatorModification (FOV)

    Image Acquisition3D Reconstruction

  • Cone-Beam CT

    Projection data Volume reconstructionjMultiple projections

    over ~180oSub-mm spatial resolution

    + soft tissue visibility

  • Image Quality:Key Characteristics

    Large volumetric FOV Single orbit about the patient

    Sub-Millimeter Spatial Resolution Sub Millimeter Spatial Resolution Soft-Tissue Visibility

  • Image Quality

    C-arm System Parameters Key Image Quality Metrics C-arm System Parameters- System configuration

    Geometry, grid, bowtieFPD readout mode

    - Geometric calibration

    - Image uniformity / stationarityShading, view aliasing

    - CT # accuracyHU calibration, shading artifacts

    - System configurationGeometry, grid, bowtieFPD readout mode

    - Geometric calibrationMechanical flex, reproducibilityDegrees of freedom

    - Acquisition parametersNumber of projections

    HU calibration, shading artifacts- Spatial resolution

    LP/mm, FWHM wire, MTF- Contrast

    Signal difference (HU) SDNR

    Mechanical flex, reproducibilityDegrees of freedom

    - Acquisition parametersNumber of projectionsNumber of projectionskVp, mAsDose

    - Reconstruction parametersReconstruction filter

    Signal difference (HU), SDNR- Noise

    Voxel noise, NPS- SNR

    N i i l t t (NEQ)

    Number of projectionskVp, mAsDose

    - Reconstruction parametersReconstruction filterReconstruction filterVoxel size (axy and az)2D/3D sampling

    Noise equivalent quanta (NEQ)- Artifacts

    Truncation, scatter, metal, etc.

    Reconstruction filterVoxel size (axy and az)2D/3D sampling

  • Cone-Beam GeometryS t t di t t d b th li tiSystem geometry dictated by the application

    Geometry affects every aspect of image quality

  • Uniformity / Stationarity

    Signal Uniformity- Stationarity of the mean

    (3.8 4.2)

    Shading artifactsBeam-hardeningTruncation

    (4.6 3.2)(5.6 2.4) (-1.3 6.2)

    HU = (4.6-1.3) HU= 3.3 HU

    Noise Uniformity- Stationarity of the noise- WSS of second-order statistics

    Physical effects:

    (4.6 3.2)( ) ( )

    (4.4 4.2)

    0.20

    Physical effects:Quantum noiseBowtie filter

    Sampling effects:Intrinsic to FBP

    SPR ~0

    al (

    /mm

    )

    Intrinsic to FBPNumber of projectionsView aliasing

    Mea

    n S

    igna

    SPR ~100%

    0.00

    Distance (mm)-10 0 +10

  • Uniformity / Stationarity2

    Signal Uniformity- Stationarity of the mean

    Variance Maps 2(x,y)2

    (/mm)2

    Shading artifactsBeam-hardeningTruncation

    Noise Uniformity- Stationarity of the noise- WSS of second-order statistics

    Physical effects:

    Water CylinderCylinder + Bowtie

    Physical effects:Quantum noiseBowtie filter

    Sampling effects:Intrinsic to FBP Water Cylinder

    Cylinder + Bowtie

    aria

    nce

    Intrinsic to FBPNumber of projectionsView aliasing

    Air

    Air

    Va

    -10 0 +10Distance (mm)

    10 0 +10

  • Spatial Resolution

    Factors affecting spatial resolution Focal spot sizeFocal spot size System geometry

    Magnification Detector configuration

    X-ray converterSAD

    Pixel pitch Recon parameters

    Recon filterSDD

    Recon filter Voxel size

  • Spatial Resolution

    Factors affecting spatial resolution Focal spot sizeFocal spot size System geometry

    Magnification

    SAD Detector configuration X-ray converter

    SDD

    Pixel pitch Recon parameters

    Recon filter Recon filter Voxel size

    C tConverter

    Pixel Matrixapix

  • Spatial Resolution

    Factors affecting spatial resolution Focal spot sizeFocal spot size System geometry

    Magnification Detector configuration

    X-ray converter Pixel pitch

    Recon parametersRecon filter Recon filter

    Voxel size

  • Spatial Resolution( H f h PS )(FWHM of the PSF)

    m)

    HM

    (mm

    FWH

    Sm

    ooth

    Shar

    p

    S S

    Filter Param (hwin)

  • Spatial Resolution(li i )(line-pairs per mm)

    Minimum resolvableline-pair group

  • Spatial Resolution( d l i T f i )

    127 m Wire in H2O

    (Modulation Transfer Function)

    JJ

    JJ

    JJ

    JJ

    JJJJ

    0.8

    1.0

    Steel Wire

    -1)

    2

    JJ

    JJ

    JJ

    JJ

    JJ

    0.4

    0.6System MTF

    nal

    (m

    m

    JJJJJJJ

    JJJJ

    JJJJ

    JJJJ

    JJJ

    JJ

    J

    0.2

    0.4

    Measured

    Sig

    JJJJJJJJJJJ

    J

    0.00.0 0.5 1.0 1.5 2.0

    Spatial Frequency (mm-1)

    ( ) ( )[ ] ,, yxLSFFTffMTF yx =

  • Spatial Resolution

    Axial Stapes Crura

  • Image Noise

    CT image noise depends on Dose Detector efficiency

    V l i Voxel size Axial, axy Slice thickness a Slice thickness, az

    Reconstruction filter

    Barrett, Gordon, and Hershel (1976)

  • Image Noise

    Dose Reconstruction Filter

    Xba +~50

    60

    ) X

    30

    40

    se (

    CT#

    )

    Sm

    ooth

    Shar

    p

    10

    20Nois

    S S

    00 0.5 1.0 1.5 2.0 2.5 3.0

    Dose (mGy)Dose (mGy)

  • Noise-Power Spectrum

    The NPS describes Frequency content of the noise:

    Magnitude of the noise: Magnitude of the noise:

  • Noise-Power Spectrum

    Axial Plane (x,y)S(f f )

    Axial NPSS(fx, fy)

    m3 )

    0.4 mAs1 mAs2 mAs

    NPS

    (2 m

    m 2 mAs4 mAs

    N

    Spatial Frequency, fx (mm-1)y fx

  • Noise-Power Spectrum

    Sagittal Plane (x,z)

    S(f f )Sagittal NPS

    S(fx, fz)0.4 mAs

    1 mAs

    S (

    2 mm

    3 ) 2 mAs4 mAs

    NP

    Spatial Frequency, fz (mm-1)

  • Noise-Power Spectrum

    NPS(fx, fy, fz)

    Transverse domain:Filtered-rampGreen NPS

    Axial domain:Band-limitedRed NPSRed NPS

  • Contrast

    A large-area transfer characteristicDefined:Defined:

    As an absolute difference in mean pixel values:ROI #1

    ROI #2For example:C |0 18 cm-1 0 20 cm-1|C = |0.18 cm-1 0.20 cm-1|

    = 0.02 cm-2orC = |-100 HU 0 HU|

    100 HU

    As a relative difference in mean pixel values:

    = 100 HU

    For example:C = |0.18 cm-1 0.20 cm-1|

    0.19 cm-1~ 10% ~ 10%

  • Signal Difference-to-Noise Ratio

    3.5100 kV 103 HU

    23.3 mGySoft-Tissue-Simulating Spheres

    2.53.0 100 kVp

    88 HU

    103 HU

    1.52.0

    CN

    R

    66 HU

    9.6 mGy

    0 51.0.C

    45 HU

    25 HU

    y

    0.00.5

    11 HU

    22 HU2.9 mGy

    0 5 10 15 20 25Dose to Isocenter (mGy)

    0.6 mGy

  • 3D NEQ and DQENEQEffective number of quanta

    DQE

    Fraction of quanta used at each eachused at each spatial frequency(Efficiency x Fluence)

    Fraction of quanta used at each each frequency.

    Observations:3D DQE(0) ~ Projection DQE(0)

    (f) d d i3D DQE(f) dependent on reconstruction parameters

  • 3D NEQ

    Axial NEQ4 mAs2 mAs1 mAs

    Axial NEQ

    mm

    2 )

    y(m

    m-1

    )

    1 mAs0.4 mAs

    hoto

    ns/m

    uenc

    y, f y

    NEQ

    (ph

    tial F

    requ

    Spatial Frequency, fx (mm-1)

    N

    Spatial Frequency, fx (mm-1)

    Spat

  • 3D NEQ

    4 ASagittal NEQ 4 mAs2 mAs1 mAs

    Sagittal NEQ

    mm

    2 )

    z(m

    m-1

    )

    0.4 mAs

    hoto

    ns/m

    uenc

    y, f z

    NEQ

    (ph

    tial F

    requ

    N

    Spatial Frequency, fz (mm-1)Spatial Frequency, fx (mm-1)

    Spat

  • Artifacts

    Rings Shading MotionStreaks

    LagMetal Cone-BeamTruncation

  • Geome

Search related