You are on page 1of 31

ES97D/ES3H5 Biomedical Imaging and Medical Devices

Lecture 13 X-rays and computed tomography Part 3


Module leader Prof Joanna Collingwood, School of Engineering
J.F.Collingwood@warwick.ac.uk

Slides marked CK-UHCW are from/adapted from slides by Chris Koller, Head of Radiology Physics,
UHCW NHS Trust, Chris.Koller@uhcw.nhs.uk
WHAT NEXT? – FROM END OF LECTURE 12
Having looked at the principles of x-ray interactions with matter, we will
continue in Lecture 13 to look at the process of making images, and of the
role of computed tomography in x-ray imaging.

In advance of this, please read Essentials of in vivo biomedical imaging


Chapter 2, remainder of Section 2.3, and sections 2.4 and 2.5. E-book
available via Library website
http://webcat.warwick.ac.uk/record=b2851815~S1
Learning Outcomes

1. UNDERSTAND AT A CONCEPTUAL LEVEL the acquisition and reconstruction of the


computed tomography image, including FBP and the main forms of iterative
reconstruction
2. BE FAMILIAR with factors affecting image quality (including artefacts and exposure
compensation), and the purpose of helical and multi-slice CT
3. EXPLAIN the concept of spectral CT and how it differs to (ordinary) CT
4. PERFORM SIMPLE CALCULATIONS of x-ray beam attenuation and CT number


You will also have read Physics for Diagnostic Radiology – Chapter 1 Fundamentals of
Radiation Physics and Radioactivity in Week 1, but in case you want to refresh your
knowledge on this topic, the chapter (where Chapter 1 Sections 1.1 and 1.10-1.16
are important background for this part of the module) is available as a PDF in the
Week 1 downloads folder, it’s also hard-copy in the library and available online at
http://encore.lib.warwick.ac.uk/iii/encore/record/C__Rb3155843
CK-UHCW

Overview
• Principles of CT
• Image Reconstruction
• Dose Optimisation
• Spectral CT
CK-UHCW

CT Basic Principles

Ni N0

Ni N0

N
Ni0
=ln
 
NN0 
CK-UHCW

CT Basic Principles 2

µ8

µ2

µ1
CK-UHCW

CT Number
• CT number (Hounsfield scale) is
calculated from X-ray attenuation of
object:

CT no = 1000  ( - w)/w
w = attenuation of water

CT air = -1000
CT fat = -120 to -90
CT water = 0
CT bone = +300 to +1900 Examples of measured values from Kamalian et al, Handbook of Clinical Neurology, 2016
https://www.sciencedirect.com/topics/medicine-and-dentistry/hounsfield-scale
(cancellous to cortical)

• CT numbers are then assigned to a grey


scale using a window width and level.
• Image may be processed, manipulated,
other planes reconstructed etc.
CK-UHCW (Here c is the equivalent of x from slide 5, i.e. c1 is the equivalent of Δx)
If you have not already watched Supporting Videos 2 & 3 (Week 8), please now
pause Lecture 13 so that you can do so. You can also view them directly from this
slide if you are using the Powerpoint version of the slides.

Fourier slice theorem Filtered back projection


Supporting video 2 Supporting video 3
CK-UHCW

Generating the Image

• Filtered Back Projection: (Example)


– Radon Transformation, Fourier Slice Theorem
– Accentuates noise and generate artefacts
• Iterative Reconstruction:
– Algebraic Reconstruction Technique,
• Given sinogram y and system model A,
reconstruct x by solving y=Ax
– Statistical Iterative Process
CK-UHCW

Filtered Back Projection Rf(2,s)

Rf(1,s)

Object f(x,y)

https://www.osapublishing.org/oe/abstract.cfm?uri=oe-17-25-22320
CK-UHCW

Source objects

Number of back projections shown (2, 4, 8, 16, 32)


CK-UHCW
CK-UHCW
CK-UHCW

Object & Sinogram


For an object, the projection data from a
tomographic scan of that object is contained in a
number of sine waves with different amplitudes
and phases, a ‘sinogram’.

Hint – it might help you to re-watch Supporting videos 2 & 3 at this point!
From the Fourier Slice Theorem:
A Fourier transform of an X-ray
image is a slice through the Fourier
transform of the 3D density of the
imaged object. Collecting multiple
projections produces an increasingly
complete Fourier transform of the
density. The inverse Fourier
transform then reveals the 3D
density of the object and is back-
projected over the domain.
➔ Formation of the sinogram is animated at
https://youtu.be/gQUOBM5Hon4
CK-UHCW
CK-UHCW

Iterative Reconstruction
CK-UHCW

Iterative Reconstruction
• Start with initial estimate of object
• Forward project object to obtain sinogram
• Compare sinogram and measured data.
• Difference is back-projected and result used to update the
original estimate
• Repeat many times
• Convergence occurs when change between successive
iterations is small

• Model based techniques


CK-UHCW

Model Based Iterative Reconstruction


• Start with initial estimate of object
• Forward project object to obtain sinogram
• Compare sinogram and measured data.
• Difference is back-projected and result used to update the
original estimate
• Repeat many times
• Convergence occurs when change between successive
iterations is small
CK-UHCW

Iterative Reconstruction

• ASiR – Adaptive Statistical Iterative Reconstruction


• Uses FBP as first estimate for IR
• Control how much IR is applied to image (around 30-40%)
• VEO – Model Based Iterative Reconstruction
• Uses full model of CT system – Physics (spectra / beam hardening /
scatter) and optics (focal spot size, detector spatial resolution)
• ASIR-V hybrid reconstruction technique
• Just uses physics model
CK-UHCW

Examples
CK-UHCW

CT Basic Principles - 2
• Remember that linear attenuation coefficient µ = f (material and
Energy)
• Polychromatic beam from X-ray tube
• Increase filtration to make X-ray more monochromatic (up to 8mmAl)

Beam hardening – where the


properties of the x-ray beam
spectrum changes during its
passage through the object
CK-UHCW

CT Artefacts
CK-UHCW

Previously CT (axial) scanning was slow.


Helical CT Now helical scanning is well-established and
enables images to be acquired much faster.

• Continuous gantry motion with table motion


• Path of X-ray beam around patient follows a helical path
• No waiting between slices for table to move
CK-UHCW

What happens when the scanners don’t behave as specified?


See https://www.ctug.org.uk/meet02/extrarotations.pdf
CK-UHCW

Multi Slice CT
• Two dimensional
array of detectors

• Acquire multiple
slices simultaneously

• Greatly decrease
scanning time
CK-UHCW


CK-UHCW
Automatic Exposure Control:
provides a modulation of the tube current for consistent
image quality

➢Patient size: tube current is fixed as a


function of patient size

➢Z-axis AEC: current varies for each


rotation as a function of patient size and
material density

➢Rotational AEC: current varies across


a rotation (accounts for elliptical shape
of human body)
CK-UHCW

Spectral CT
• Linear Attenuation Coefficient µ = f(E)

• As X-ray beam passes through object get the lower energy


photons preferentially absorbed – known as Beam Hardening at
CT energies

• Is there a way to get more useful information out about μ for the
various tissues?
CK-UHCW

Spectral CT
CK-UHCW Ii = incident photons (from x-ray tube)
I0 = output signal (to detector)

Spectral CT

I0 = Ii 𝒆−(𝝁𝟏 𝑬 𝒙𝟏+𝝁𝟐 𝑬 𝒙𝟐)

You might also like