NATIONAL UNIVERSITY OF HO CHI MINH CITY

HO CHI MINH CITY UNIVERSITY OF

TECHNOLOGY
···☼···

CALCULUS 2
REPORT ON MAJOR PROJECT

TOPIC: INTRODUCTION TO THE

MATHEMATICS OF MEDICAL IMAGING
Lecturer: Dr. Phung Trong Thuc
Class: CC02
Nguyễn Tuấn Minh 2352757 [email protected]

Phan Đức Việt 2353323 [email protected]

Đoàn Gia Quyền 1852707 [email protected]

Ngô Minh Khoa 2352567 [email protected]

Contents
ACKNOWLEDGEMENT...................................................................................................3
1. INTRODUCTION...........................................................................................................4
2. BASIC THEORY............................................................................................................4
2.1. The space of line in the plane....................................................................................5
2.2. Reconstructing an object from its shadows...............................................................7
2.3. Exercise 1.2.14..........................................................................................................9
ACKNOWLEDGEMENT

We extend our gratitude to Mr. Phung Trong Thuc from Ho Chi Minh City University of
Technology - Ho Chi Minh City National University for providing us with fundamental
knowledge essential for accomplishing our project. While working on this significant task
as a group, we acknowledge that due to our limited understanding, we encountered
various challenges in learning, assessing, and presenting our work. We eagerly await
your guidance and recommendations to further refine and enhance our skills. Thank you
sincerely for your support!
1. INTRODUCTION
Mathematics is a crucial tool not only in theory but also in practical applications,
especially in the realm of medical imaging. In the field of X-ray imaging, the integration
of mathematics and technology has led to significant advancements in improving the
quality and accuracy of medical images. This technology not only provides a deeper
insight into the human body but also plays a vital role in disease diagnosis and treatment
planning. By applying mathematical methods such as Fourier transforms, tomography,
and plane geometry, X-ray images can be generated accurately and efficiently. This not
only opens up new avenues for medical advancements but also contributes to enhancing
the quality of human life. For an assignment of the topic, we would like to introduce the
basic application of the mathematics for the reconstruction of an object from it shadow
and all the information and formulas listed in the report will primarily be sourced from
the book “Introduction to the Mathematics of Medical Imaging” of Charles L. Epstein.

2. BASIC THEORY
The challenge of reconstructing images in X-ray tomography is often explained as
reconstructing an object based on its "projections," akin to shadows cast under X-ray
illumination. In this section, we simplify the problem by considering the task of outlining
an object solely from its shadows. This two-dimensional scenario mirrors situations in
medical imaging. Let's denote D as the convex region in the plane. If we position a
distant light source, the emitted rays become essentially parallel. These rays act as a
collection of parallel lines. Our objective is to observe the shadow cast by D from various
positions of the light source. To capture these measurements, envision a screen placed on
the opposite side of D, perpendicular to the light rays' direction, as illustrated below. This
screen serves as the detector; in practical setups, sensors would detect the shadow's
boundaries on the screen.
 Figure 1. The shadow of convex region

The region D obstructs a specific set of light rays while allowing others to pass through.
Therefore, measuring the shadow entails identifying the "first" and "last" lines within this
set of parallel lines that intersect D. To fully characterize the object, we must rotate both
the source and detector through 180 degrees, recording the points where the shadow
begins and ends at each angle. The lines that intersect a region for the first and last time
precisely meet its boundary. These lines are tangent to the boundary of D. Consequently,
reconstructing a region from its shadows is mathematically equivalent to reconstructing it
from the knowledge of its tangent lines. To embark on this endeavor, we require an
effective method to organize our measurements. To that end, we present a framework for
describing the space of all lines in the plane.

2.1. The space of line in the plane.

A line in the plane is the set of points which satisfy an equation of the form:

a x +b y =c

Where a 2+ b2 ≠ 0. We get the same set of points if we replace this equation by:

a b c
x+ y=
√ a +b
2 2
√ a +b
2 2
√ a + b2
2

a b
The coefficients, ( , ) define a point ω=(ω ¿ ¿ 1 , ω2 )¿ on the unit circle and
√ a +b √ a +b2
2 2 2

c
the constant = t. The lines in the plane are parametrized by a function below
√ a +b2
2

⟨ ( x , y ) , ω ⟩=t

Where ⟨ ( x , y ) , ω ⟩ is the dot product.

However, we can parameterize ω by the angle θ, because the hypotenuse is always 1, so

that ¿), sin(θ ¿) is equal to (ω ¿ ¿ 1 ,ω 2)¿.

ω (θ)=¿), sin(θ ¿)

Therefore, we have:
 ⟨ ( x , y ) ,(cos ⁡(θ), sin(θ))⟩ =t

The pair ( t , ω ¿ actually specifies and oriented line. That is, we can use these data to
define the positive direction along the line. The vector

^ =(−ω2 , ω1 )
ω

We form the matrix:

(ωω 12 −ω 2
ω1 )
The determinant of this matrix is given by ω 12+ ω22

Since we are given that the determinant is +1, we have ω 12+ ω22=1which implies that lies
on the unit circle in the 2D plane. Therefore, ω ^ defines the positive direction or
^ ). This proposition ensures that the orientation of the line is
orientation of the line l(t,ω
consistent and determined by the direction vector with a determinant of 1, ensuring a
consistent orientation regardless of the choice of parameter t.

Figure 2. Parameterization of oriented lines in plane

The line l t , ω is the common boundary of these half-planes. Facing along the line l t , ω in the
−¿ ¿
direction specified by ω ^ , the half-plane H t ,ω lies to the left.
Given the pair (t,ω), where t is the affine parameter and ω is the direction orthogonal to the
line, we define two half-planes:

+¿= {( x ϵ R |⟨ x ,ω ⟩ >t ) }∧H t ,ω

−¿= {( x ϵ R |⟨ x, ω⟩< t )} ¿
2
2
¿
H t ,ω

The line l(t,ω) serves as the common boundary of these half-planes. When facing along l(t,ω) in
−¿ ¿
the direction specified by H t ,ω lies to the left.

2.2. Reconstructing an object from its shadows

h D ( θ )=t 1 and h D ( θ+ π )=−t 0

t 1∧−t 0 represent tangent vectors to the boundary of the region D, associated with the
orientation of the boundary and the direction in which it is traversed

t 1: represents a tangent vector to the boundary that aligns with the chosen orientation,
indicates the direction in which the boundary is traversed positively (typically
counterclockwise)

−t 0: represents a tangent vector to the boundary that aligns with the opposite orientation,
indicates the direction opposite to the chosen orientation, often traversed clockwise

The choice between t 1 and −t 0 depends on the selected orientation of the boundary. t 1 is
chosen when following the positive orientation, while −t 0 is chosen when following the
opposite orientation ( counterclockwise ).

{ hD ( θ ) (cos ⁡(θ),sin (θ))+ s (−sin ( θ ) , cos ( θ ) )|s ∈(−∞ , ∞ )}

s is a constant, which runs from negative infinity to positive infinity

(−sin ( θ ) , cos (θ ) )=ω

^ (θ)

In summary, this is a cross product between the two vectors ω and ω ^ , with every values
of s, we will have 1 point, so that s runs from neagtive to positive infinity. It will form a
set of point, which will create the line l h (θ) , ω(θ) D

( x ( θ ) , y ( θ ) ) ¿ h D ( θ ) (cos ⁡(θ), sin(θ))+s (θ ) ( −sin ( θ ) , cos ( θ ) )

s ( θ ) is a function, which is respected to θ, so for each value of θ , we can find a point on
the boundary of D

At the point of tangency, the tangent line direction is ω ( θ ). Differentating ω respect ¿ θ = ω

^
^.
. ω is perpendicular with ω

^ ( θ ), so that h'D ( θ )−s ( θ )=0 ( the formula below )

The tangent line is parallel to ω

( x ' ( θ ) , y ' (θ ) )=( h'D ( θ )−s ( θ )) ω ( θ ) + ( hD ( θ )+ s ' ( θ ) ) ω^ (θ )

Since the tangent line at( x ( θ ) , y ( θ ) ) is parallel to ω

^ ( θ ) it follows from that

' '
=> ^ (θ)
x ( θ ) , y ( θ ) must be equal to k . ω therefore,

( h'D ( θ )−S ( θ ) ¿ ω (θ) = 0 thereforeh'D ( θ )−S ( θ )=0.

Thus, h'D ( θ )=S ( θ )

If h'D ( θ )−S ( θ ) ≠ 0

( x ' ( θ ) , y ' (θ ) )=( h'D ( θ )−S ( θ )) ω(θ)+ ( hD ( θ )+ s ' ( θ ) ) ω^ ( θ )

^ (θ )
would turn into another vector and couldn’t be parallel to ω

This gives a parametric representation for the boundary of a convex region in terms of its
shadow function: If the shadow function is h D ( θ ) then the boundary of D is given
parametrically by

( x ( θ ) , y ( θ ) ) ¿ h D ( θ ) ω ( θ ) + h'D ( θ ) ω
^ ( θ ).

Let D denote a convex region and hD its shadow function. We can think of D → hD as a
mapping from convex regions in the plane to 2π periodic functions. It is reasonable to
enquire if every 2π periodic function is the shadow function of a convex region. The
answer to this question is no. For strictly convex regions with smooth boundaries we are
able to characterize the range of this mapping. If h is twice differentiable then the tangent
vector to the curve defined by
 ( x ( θ ) , y ( θ ) ) ¿ h D ( θ ) ω ( θ ) + h'D ( θ ) ω
^ (θ )

is given by
'' '
( x ' ( θ ) , y ' (θ ) )=(h ( θ ) +h ( θ ) ¿ ) ω
^ (θ)
¿

we observed that the tangent vector to the curve at x (θ ) , y ( θ ) and ω

^ ( θ ) point in the same
direction. From our formula for the tangent vector we see that this implies that

h ( θ ) +h ( θ ) ¿ 0for all θ ∈[0 ; 2 π ] (*)

'' '

h ( θ ) +h ( θ ) ≠ 0because ( x ( θ ) , y (θ ) ) must have the same direction as ω

'' ' ' '
^ (θ )

h' ' ( θ ) +h ' ( θ ) is postive because ( x' ( θ ) , y ' ( θ ) ) must be collinear with ω
^ (θ )

( The positive direction on the boundary is selected so that, when facing in that direction
the region lies to the left; the counterclockwise direction is, by convention the positive
direction.)

2.3. Exercise 1.2.14

Suppose that h Dis a function satisfying (*). Show that the area enclosed by Γ h is given
by:
2π
1
Area ( Dh) = ∫ ¿ ¿
20

Green’s thereom:

∂ M ∂h
∮ Lⅆx + Mⅆy=¿ ∬ ¿(¿ − )dxⅆy ¿
D
∂x ∂ y

* M =x , L=− y

¿>∬ 1−(−1 ) dA=2 ∬ dA = 2A ( A=∬ ⅆA )

1
A= ∮ ¿)
2
 1
¿ ∮(xⅆγ − ydx)
2

Lⅇt x=x ( θ ) , y= y ( θ )

ⅆx =x ' ( θ ) d ( θ ) , ⅆy = y ' ( θ ) dθ
π
2
1
 ∫ ¿ ¿ –(h sin θ+h ' cos θ ' ¿ ( h' ' + h ) (−sin θ ) dθ
20

2π
1
¿ ∫ ¿¿
20

2π
1 1
20
∫ '' 2
( h. h ¿ +h )dθ= ¿ ¿
2

''
u=h ⅆv =h dθ

ⅆu=h' dθ v=h '

1
¿ ¿¿
2

Therefore we have:
2π
1
¿ – h(0). h' (0) +∫ ( hθ )2 −(h¿ ¿' (θ))2 ⅆθ ¿ ]
2 0

Because we have, h(θ+2 π )=h(θ)

=> h' (θ)=h' (2 π +0) => h' (θ)=h' (2 π )

=> h(0)= h(2 π ) , h' (0)=h' ¿ )

Thus, h ( 2 π )−h' ( 2 π ) – h(0). h' (0)=0

So we have:
 2π
1
Area ( Dh) = ∫ ¿ ¿
20

*Explain why this implies that a function satisfying (*) also satisfies the estimate

2π 2π

∫ ¿ ¿ <∫ [¿ ( h (θ))2]dθ ¿
0 0

-Because Area have to be positive therefore, the statement above is true

2π
1
Area ( Dh) = ∫ ¿ ¿
20

2π 2π
 ∫ ¿ ¿ <∫ [¿ ( h (θ)) ]dθ ¿
2

0 0