Image Reconstruction from Projection

► Reconstruct an image from a series of projections

X-ray computed tomography (CT)

“Computed tomography is a medical imaging method employing

tomography where digital geometry processing is used to generate a
three-dimensional image of the internals of an object from a large
series of two-dimensional X-ray images taken around a single axis of
rotation.”

http://en.wikipedia.org/wiki/Computed_tomography

3/11/2014 97
 Backprojection

“ In computed tomography or other imaging techniques

requiring reconstruction from multiple projections, an
algorithm for calculating the contribution of each voxel of
the structure to the measured ray data, to generate an
image; the oldest and simplest method of image
reconstruction. “

http://www.medilexicon.com/medicaldictionary.php?t=9165

3/11/2014 98
 Image Reconstruction: Introduction
Intensity is
Soft, uniform proportional to
tissue absorption

Uniform with higher

absorption Tumor

3/11/2014 99
 Image Reconstruction: Introduction

3/11/2014 100
3/11/2014 101
3/11/2014 102
 Other CTs

► Electron beam CT (Fifth-generation CT)

Electron beam tomography (EBCT) was introduced in the early 1980s, by
medical physicist Andrew Castagnini, as a method of improving the temporal
resolution of CT scanners.
High cost of EBCT equipment, and poor flexibility

► Helical (or spiral) cone beam computed tomography (sixth-generation)

A type of three dimensional computed tomography (CT) in which the source
(usually of x-rays) describes a helical trajectory relative to the object while a
two dimensional array of detectors measures the transmitted radiation on part
of a cone of rays emitting from the source

http://en.wikipedia.org/wiki/Computed_tomography

3/11/2014 103
 Other CTs

► Multislice CT (seventh-generation)

► The major benefit of multi-slice CT

 Significant increase in detail

 Utilizes X-ray tubes more economically
 Reducing cost and potentially reducing dosage

3/11/2014 104
 Projections and the Radon Transform

3/11/2014 105
 Projections and the Radon Transform

∞ ∞
g ( ρ j ,θ k )
3/11/2014
∫ ∫
−∞ −∞
f ( x, y )δ ( x cos θ k + y sin θ k − ρ j )dxdy
106
 Projections and the Radon Transform

► Radon transform gives the projection (line integral) of

f(x,y) along an arbitrary line in the xy-plane

∞ ∞
ℜ { f=
} g ( ρ ,θ=) ∫ ∫ f ( x, y )δ ( x cos θ + y sin θ − ρ )dxdy
−∞ −∞

M −1 N −1
ℜ { f=
} g ( ρ ,θ=)
=
∑ ∑ f ( x, y)δ ( x cos θ + y sin θ − ρ )
x 0=y 0

3/11/2014 107
Example: Using the Radon transform to obtain
the projection of a circular region

► Assume that the circle is centered on the origin of the xy-plane.

Because the object is circularly symmetric, its projections are the
same for all angles, so we just check the projection for θ = 0


A x2 + y 2 ≤ r 2
f ( x, y ) = 
0 otherwise

3/11/2014 108
Example: Using the Radon transform to obtain
the projection of a circular region
∞ ∞
g ( ρ ,θ )
= ∫ ∫ f ( x, y)δ ( x cos θ + y sin θ − ρ )dxdy
−∞ −∞
∞ ∞
∫ ∫ f ( x, y)δ ( x − ρ )dxdy
−∞ −∞
∞
= ∫ f ( ρ , y )dy
−∞

r2 −ρ2
=∫ f ( ρ , y )dy
− r2 −ρ2

r2 −ρ2
=∫ Ady
− r −ρ 2 2

2 A r 2 − ρ 2 |ρ | ≤ r
=
 0 otherwise
3/11/2014 109
 2 A r 2 − ρ 2 |ρ | ≤ r
g (ρ ) = 
 0 otherwise

3/11/2014 110
 Sinogram: The Result of Radon Transform

► Sinogram: the
result of Radon
transform is
displayed as an
image with
ρ and θ as
rectilinear
coordinates

3/11/2014 111
 Image Reconstruction

fθ ( x, y ) g ( x cos θ + y sin θ , θ )
=
π
f ( x, y ) = ∫ fθ ( x, y )dθ
0
π
f ( x, y ) = ∑ fθ ( x, y )
θ =0

A back-projected image formed is referred to as a laminogram

3/11/2014 112
 Examples: Laminogram

3/11/2014 113
 The Fourier-Slice Theorem

For a given value of θ , the 1-D Fourier transform

of a projection with respect to ρ is
∞
G ( w, θ ) = ∫ g ( ρ ,θ )e − j 2πωρ d ρ
−∞

∞ ∞ ∞
G (ω , θ ) ∫ ∫ ∫
−∞ −∞ −∞
f ( x, y )δ ( x cos θ + y sin θ − ρ )e − j 2πωρ d ρ dxdy

f ( x, y ) ∫ δ ( x cos θ + y sin θ − ρ )e − j 2πωρ d ρ dxdy


∞ ∞ ∞
∫ ∫
−∞ −∞  −∞ 
∞ ∞
=∫ ∫ f ( x, y )e − j 2πω ( x cosθ + y sin θ ) dxdy
−∞ −∞

3/11/2014 114
 The Fourier-Slice Theorem
∞
G ( w, θ ) = ∫ g ( ρ ,θ )e − j 2πωρ d ρ
−∞
∞ ∞
G (ω , θ ) = ∫ ∫ f ( x, y )e − j 2πω ( x cosθ + y sin θ ) dxdy
−∞ −∞

= ∫ ∫ f ( x, y )e − j 2π (ux + vy ) dxdy 
 ∞ ∞

 −∞ −∞ =
 u cosθ , v w sin θ
w=

= [ F (u ,=
v) ]u cosθ , v w sin θ
w=

= F ( w cos θ , w sin θ )
Fourier-slice theorem: The Fourier tansform of a projection is a slice of
the 2-D Fourier transform of the region from which the projection was
obtained
3/11/2014 115
 Illustration of the Fourier-slice theorem

3/11/2014 116
 Reconstruction Using Parallel-Beam Filtered
Backprojections
∞ ∞
f ( x, y ) = ∫ ∫ F (u, v)e j 2π (ux + vy ) dudv
−∞ −∞

= cos θ , v w sin θ , then

Let u w= = dudv wdwdθ ,
2π ∞
f ( x, y ) = ∫
∫ F ( w cos θ , w sin θ )e j 2π w( x cosθ + y sin θ ) wdwdθ
0 0
π 2 ∞
=∫ ∫ G ( w, θ )e j 2π w( x cosθ + y sin θ ) wdwdθ
0 0

G ( w, θ + 180 ) =
G (− w,θ )
π ∞
f ( x, y ) = ∫ ∫ | w | G ( w, θ )e j 2π w( x cosθ + y sin θ ) dwdθ
0 −∞
3/11/2014 117
 Reconstruction Using Parallel-Beam Filtered
Backprojections

π ∞
f ( x, y ) = ∫ ∫ | w | G ( w, θ )e j 2π w ( x cosθ + y sin θ )
dwdθ
0 −∞
π
=∫  ∞ | w | G ( w, θ )e j 2π wρ dw
 ∫−∞
It’s not
dθ
integrable 0 =
 ρ x cosθ + y sin θ

Approach:
Window the ramp so it becomes zero outside of a defined frequency
interval. That is, a window band-limits the ramp filter.

3/11/2014 118
 Hamming / Hann Widow

 2π w
c + (c − 1) cos 0 ≤ w ≤ ( M − 1)
h( w) =  M −1
 0 otherwise

c = 0.54, the function is called the Hamming window

c = 0.5, the function is called the Han window

3/11/2014 119
 The Plot of Hamming Widow

3/11/2014 120
 Filtered Backprojection

The complete, filtered backprojection (to obtain the

reconstructed image f(x,y) ) is described as follows:

1. Compute the 1-D Fourier transform of each projection

2. Multiply each Fourier transform by the filter function |w|
which has been multiplied by a suitable (e.g., Hamming)
window
3. Obtain the inverse 1-D Fourier transform of each
resulting filtered transform
4. Integrate (sum) all the 1-D inverse transforms from step
3
3/11/2014 121
 Examples: Filtered Backprojection

3/11/2014 122
 Examples: Filtered Backprojection

3/11/2014 123
 Implementation of Filtered Backprojection in
Spatial Domain
► Fourier transform of the product of two frequency domain
functions is equal to the convolution of the spatial
representation
► Let s(p) denote the inverse Fourier transform of |w|
π
f ( x, y ) = ∫  ∞ | w | G ( w, θ )e j 2π wρ dw
 ∫−∞
dθ
0 =
 ρ x cosθ + y sin θ
π
= ∫ [ s( ρ ) g ( ρ ,θ )] dθ
0 =ρ x cosθ + y sin θ
π
 ∞ g ( ρ , θ ) s ( x cos θ + y sin θ − ρ )d ρ dθ
∫0  ∫−∞ 
3/11/2014 124
 Reconstruction Using Fan-Beam Filtered
Backprojections

θ= α + β
ρ = D sin α

3/11/2014 125
 Reconstruction Using Fan-Beam Filtered
Backprojections
Objects are encompassed within a circular area of radius T about
the origin of the plane, or g(ρ ,θ )=0 for |ρ |>T
π
 ∞ g ( ρ , θ ) s ( x cos θ + y sin θ − ρ )d ρ dθ
=f ( x, y ) ∫0  ∫−∞ 
1 2π T
2 ∫0 ∫− T
g ( ρ , θ ) s ( x cos θ + y sin θ − ρ )d ρ dθ

= cos ϕ ; y r sin ϕ
x r=
x cos θ + y sin
= θ r cos ϕ cos θ + r sin ϕ sin θ
= r cos(ϕ − θ )
3/11/2014 126
 Reconstruction Using Fan-Beam Filtered
Backprojections

= cos ϕ ; y r sin ϕ
x r=
x cos θ + y sin
= θ r cos ϕ cos θ + r sin ϕ sin θ
= r cos(ϕ − θ )
1 2π T
f ( x, y ) ∫ ∫ g ( ρ , θ ) s [ r cos(ϕ − θ ) − ρ ] d ρ dθ
2 0 − T

θ= α + β ρ = D sin α

d ρ dθ = D cos α dα d β

3/11/2014 127
Reconstruction Using Fan-Beam Filtered
Backprojections

d ρ dθ = D cos α dα d β
1 2π T
f ( x, y ) ∫ ∫ g ( ρ , θ ) s [ r cos(ϕ − θ ) − ρ ] d ρ dθ
2 0 − T

1 2π −α sin −1 (T / D )
∫ ∫ −1
g ( D sin α , α + β ) s [ r cos(α + β − ϕ ) − D sin α ] D cos α dα d β
2 − α − sin ( − T / D )

3/11/2014 128
Reconstruction Using Fan-Beam Filtered
Backprojections
1 2π T
f ( x, y ) ∫ ∫ g ( ρ , θ ) s [ r cos(ϕ − θ ) − ρ ] d ρ dθ
2 0 − T

1 2π −α sin −1 (T / D )
∫ ∫ −1
g ( D sin α , α + β ) s [ r cos(α + β − ϕ ) − D sin α ] D cos α dα d β
2 − α − sin ( − T / D )

1 2π α m
f (r , ϕ ) ∫ ∫ p (α , β ) s  R sin (α '− α ) D cos α dα d β
2 0 −α m
 α 
2

s ( R sin α ) =   s (α )
 R sin α 
2π 1  αm
( )  dβ
f (r , ϕ ) ∫0 R 2  ∫−α m
q (α , β ) h α '− α d α

1 α 
2

h(α ) =   s (α ), q (α , β ) p (α , β ) D cos α
3/11/2014 2  sin α  129
3/11/2014 130
3/11/2014 131