Computer-Tomography II · 1 Seite 1 Uwe Oelfke Computer-Tomography II: Image reconstruction and applications Prof. Dr. U. Oelfke DKFZ Heidelberg Department of Medical Physics (E040)

1

Seite 1

Uwe Oelfke

Computer-Tomography II:

Image reconstruction and applications

Prof. Dr. U. OelfkeDKFZ HeidelbergDepartment of Medical Physics (E040)Im Neuenheimer Feld 28069120 Heidelberg, [email protected]

Uwe OelfkeContents

Motivation for CT

B i Ph i H fi ld U it Basic Physics – Hounsfield Units

Technical aspects: CT scanner

CT reconstruction

Spiral- and multi-slice CT

Image quality Image quality

Clinical applications

2

Seite 2

Uwe Oelfke

CT- Reconstruction

Radon-transform

Backprojection

Filtered backprojection

Uwe OelfkeRadon-transform

Measurement: Line-integrals of attenuation coefficient distribution

A

rdnrprdlyxp ²)ˆ()(),()(

Sinogram: All projections as function of Φ,p:

)]([),( rRp

)()( ),(),( pyx

3

Seite 3

Uwe OelfkeBackprojection:1st attempt of image reconstruction

)ˆ()( nrrf

)()()(0

rBdrfrfb

Summation of projection values for each voxel over all projection angles projection angles

Uwe OelfkeBackprojection

Object Backprojected of profiles

= reconstructed image )(rfb

4

Seite 4

Uwe OelfkeBackprojection = Convolution with PSF

'²|'|

1)'()( rd

rrrrf

A

b Convolution:

1)(

rrhKernel, PSF

)()()( rBRrBrfb

))(()( rhrfb Deconvolution ?

))(()(f b

Uwe OelfkeImage reconstruction via deconvolution

Convolution theorem:

)()())(( 222 rhFrFrhF 222

1)()(

HrhF

1)(

rrhKernel, PSF

2 )()( HrhF

5

Seite 5

Uwe OelfkeTheory:Image reconstruction via deconvolution

)()( 21

2 rBRFFrf

1. Backprojection

2. 2D-Fourier-transform

3. Multiplication with filter

4 In erse 2D Fo rier transform4. Inverse 2D-Fourier-transform

Uwe OelfkeCentral Slice-Theorem

Disadvantage of deconvolution approach: alle projections have to be known

Central-Slice-Theorem allows image reconstruction during image acquisition

21 FRF

6

Seite 6

Uwe OelfkeImage reconstruction: Filtered backprojection

11

11 FBFR

1. 1d-Fourier transform of projection profile

2. Multiplication with filter (kernel)

3. 1d-inverse Fourier transform

4. Backprojection

Uwe OelfkeFiltered backprojection

Object Filtered backprojection

7

Seite 7

Uwe Oelfke

Ungefilterte RückprojektionBackprojection of measured profiles

Uwe Oelfke

Filter, kernel

=

Measured profile

Filtered profile

8

Seite 8

Uwe Oelfke

Gefilterte RückprojektionBackprojection of filtered profiles

Uwe OelfkeFiltered backprojection

SinogramObject

Image acquisition

FilteredSinogram

filteringTomogramm

Back-projection Projektionswinkel

Fächerwinkel

9

Seite 9

Uwe OelfkeSpiral-CT (1989)

Continuosly rotating CT-gantry and continous patient transport spiral CT

Uwe OelfkeSpiral-CT reconstruction

Requires interpolation of data

additional reconstruction algorithms

10

Seite 10

Uwe OelfkeSpiral-CT – Additional x-ray power required

Extended scanning times, high scanning speed

20 60 k V 80 140 k V 20-60 keV, 80-140 keV

1% x-ray efficiency, 99% heat production at the anode

new x-ray tube design:

Uwe Oelfke

STRATON-x-ray source (2005)

Direct cooling of the anode (convection, oil)

Increases cooling by a factor of 3

11

Seite 11

Uwe OelfkeSTRATON-x-ray source

z-Sharp-Technology: B-field focusses / guides e--beam

2 alternating focal spots at the anode

< 0,4 mm resolution at 0,6 mm detector width

Uwe OelfkeMulti-slice CT-CT (1998)

Several detector rows

simultaneous acquisition of several CT-slices

Fast volume scans: whole body scans in 25s!

I t i l Isometric voxel

< 1 mm < 1 mm

12

Seite 12

Uwe OelfkeCone beam-CT

Uwe OelfkeAdaptive array detector

13

Seite 13

Technical developments 1972 - 2000

1972 1980 1990 2000

min. scan time

data per 360 scan

data per spiral scan

image matrix

tube power

300 s 5 s 1 s 0.5 s

58 kB 1 MB 2 MB 12 MB

- - < 50 MB < 500 MB

80 x 80 256 x 256 512 x 512 512 x 512

2 kW 10 kW 40 kW 60 kW

slice thickness

spatial resolution

contrast resolution

13 mm 2 - 10 mm 1 - 10 mm 0.5 - 5 mm

3 lp/cm 12 lp/cm 15 lp/cm 24 lp/cm

5 mm / 5 HU / 50 mGy

3 mm / 3 HU / 30 mGy

3 mm / 3 HU / 30 mGy

3 mm / 3 HU / 30 mGy

Uwe Oelfke

Image quality

14

Seite 14

Uwe OelfkeImage quality indicators

Indicators:

ti l l ti spatial resolution

contrast

noise, homogeneity

artefacts

Uwe OelfkeSpatial resolution

There are 7 or 8 mice...

No, there are10 of them!

The ability to resolve High Contrast Objects,- High Contrast Resolution

15

Seite 15

Uwe OelfkeSpatial resolution

Edge enhancement filter Smoothing filter

Uwe Oelfke

When Small Contrast differences are questioned - Low Contrast Resolution

!

Contrast resolution

Is therea cat? ! There

could bea cat!

A cat !

16

Seite 16

Uwe OelfkeContrast resolution

Edge enhancement filter Smoothing filter

Uwe OelfkeContrast detail diagram

17

Seite 17

Uwe OelfkeNoise

Poisson statistics of photons

Depends on x-ray current I, slice thickness d

I = 50 mAd = 4 mm

I = 200 mAd = 4 mm

I = 200 mAd = 16 mm

Uwe OelfkeCARE-Technology

Volumes with low attenutation contribute less to image noise

potential reduction of dose without loss of image qualityp g q y

Dynamic adaptation of output of the x-ray source to the expected attenuation

dose reduction of up to 68%

18

Seite 18

Uwe OelfkeArtefacts

Uwe OelfkeArtifacts

Movement artifacts Partial volume effects

19

Seite 19

Uwe OelfkeInfluence of different reconstruction kernels

Uwe OelfkePartial volume effect

CT-slice

BoneSoft tissue

20

Seite 20

Uwe OelfkePartial volume effect

CT-slice

BoneSoft tissue

Uwe OelfkePartial volume effect

CT-slice

BoneSoft tissue

21

Seite 21

Uwe Oelfke

Clinical applications

Uwe OelfkeRoutine applications

Bones & Lungs

Soft Tissues

22

Seite 22

Uwe Oelfke

high contrast

Bone/Lung

Uwe Oelfke

Superior Vena Cava Thrombosis (Emotion)

130kV, 150mA, 0.8s5 Pit h 1 5 50% l5 mm, Pitch 1.5, 50% overlap

23

Seite 23

Uwe OelfkeLiver Hemangioma Study (Esprit)

4-Phase spiral studyExcellent Enhancement

• Precontrast

• Arterial

• Portal-venous

•Delayed phase

• 8 mm, Pitch 1.5

• 130kV, 120mA, 1.5s

• 120 ml / flow rate 3 ml/s

Uwe OelfkeLow dose imaging

24

Seite 24

Uwe OelfkeBroken heel – 1mm Spiral CT

Emotion - Sagittal Reformat Volume Access - VRT

Uwe Oelfke

Spine (Emotion/Emotion Duo)

• 2 mm, Pitch 2

• Low dose ( 90 mA)

• Low Dose of 100 mAs

• 2 x 1.5 mm collimation

• 2 mm slice width

99 i l d ti• 99 s spiral duration

25

Seite 25

Uwe Oelfke

Pediatrics Head (Emotion Duo)

2 x 1mm, 60 mAs

1.25mm axial slice

Uwe OelfkeCT Angiography

26

Seite 26

Uwe Oelfke

CT-Angiography

23 cm in 5 s23 cm in 5 s

< 0,4 mm Schichtdicke

0,5 s Rotation

Uwe Oelfke

Cerebrale arteries

1 mm, Pitch 2, Emotion 2 x 0.5mm, Pitch 3, Volume Zoom

27

Seite 27

Uwe Oelfke

Carotids (Emotion/Emotion Duo)

2 x 1 mm, Pitch 3, 1.25mm SW

130kV, 100mA, 0.8s, 2 mm, Pitch 2

Uwe Oelfke

Abdominal Aortic Stent Follow-up (Emotion Duo)

35cm

130kV, 143mAs, 0.8s, 2 x 1.5 mm, Pitch 4

28

Seite 28

Uwe OelfkeMehrschicht-CT

Uwe OelfkeIntrafractional Movement by 4D-MSCT on a Siemens Sensation open

Lung cancerT4N3T4N3

paralysis of rt diaphragm

29

Seite 29

Uwe OelfkeSiemens SOMATOM Definition Dual-CT

Dual CT

2 x-ray sources

2 detectors 2 detectors

Uwe OelfkeSiemens SOMATOM Definition Dual-CT

Slice thickness < 0,4 mm

32 row-detektor + z-sharp p

= 64 slices per rotation

Min. scanning time: 83ms (90° rotation)

CT of the moving heart

30

Seite 30

Uwe Oelfke

"Dual Energy"

Scanning with two x-ray sources at different energies

Energy dependence of attenuation coefficient

Siemens SOMATOM Definition Dual-CT

additional contrast

Uwe OelfkeToshiba 256-slices CT

128 mm scan-width at 0,5 mm slice thickness

whole heart imaging in one rotation

31

Seite 31

Uwe Oelfke4D-CT Toshiba 256-slice-CT