CT Image Reconstruction

Question 1

what are some parameters of image quality

Answer 1

spatial resolution, image noise, image artifacts.

Question 2

what is the radiation dose based on

Answer 2

effective usage of x-ray photons

Question 3

what is FBP

Answer 3

fan beam reconstruction

Question 4

what are the two types of reconstruction algorithms

Answer 4

analytical and iterative

Question 5

what is analytical reconstruction

Answer 5

closed form and one step

Question 6

what is iterative reconstruction

Answer 6

iteratively solve an optimization problem

Question 7

how does CT attenutation work

Answer 7

as each thing goes through its a summation of the length and density of each block

Question 8

forward projection

Answer 8

the 1 D map of the person or thing

Question 9

back projection

Answer 9

is the 1-D maps over each other to make an image

Question 10

projection slice theorem

Answer 10

Fourier transform of a parallel beam projection yields the Fourier coefficients of the image object along a straight line in 2-D Fourier space

Question 11

Limitations of Fourier Space Reconstruction

Answer 11

REgrid data from polar to cartersian coordinate sytem, and the interpolation error in k space is non local. Difficulty for region of interest reconstruction

Question 12

what is the goal of parallel beam filtered back projection

Answer 12

to develop an algorithm that can reconstruct CT images directly from measured projection data

Question 13

data redundancy

Answer 13

more data than you need so you can halve the dose- ish

Question 14

filtered back projection

Answer 14

this is the method of taking the 1D images and summating them into a form

Question 15

parallel beam FBP

Answer 15

aquire data in the slices and in order to capture all of the information, you can’t just do half.because of the difference in filed of view

Question 16

ramp filter

Answer 16

this is for the frequency domain so that the low frequency signals are not as effective as high frequency signals. So you do it in real space by convolution and made the edges more prominent. Not complete removal of low frequency, but a sharp edge would be more prominent. A super blurry image could be from this so if you don’r ramp filter then there are not sharp edges

Question 17

kernals

Answer 17

changing the form of the ramp filter

Question 18

when do you use a sharp kernal

Answer 18

bone

Question 19

when do you use a soft kernal

Answer 19

for soft tissue

Question 20

fan beam

Answer 20

Ct is acquired data with a divergent beam

Question 21

fan parallel rebinning

Answer 21

if you are doing a short scan, all of the angles in the views you want and you have to have more than 180+2 times the fan angle

Question 22

what is short scan FBP

Answer 22

it is a

Question 23

the concept of sinograms

Answer 23

it is defined as parallel beam projections, where the key idea is to follow the trace of a spceific data point during a paralllel beam data aquisition procedure. IN fan-beam case, we have already generalized the concept

Question 24

how to handle non-uniform redundancy

Answer 24

a simple scheme to handle data redundancy is to pick up one of the redundant rays and discard the other one, however this is not the optimal way to use x-ray photons (radiation dose). Our goal is to use all measured data, use every phton delivered to the patient

Question 25

why do you weight the data

Answer 25

you weight it for the data redundancy, so this is just one scan going around one are, so thsi is not moving scan. This is as its moving around. You get double of some and you weight it for trustworthiness

Question 26

helical reconstruction

Answer 26

it interpolates the data in between the slices. You guess at what the signal is beteween the two. There is 360 and 180. SO are you going t ointerpolate when you do a full movement arpound or half the way around

Question 27

Mutli-slice helical

Answer 27

it is a cone beam has artifacts because there are more detectors remove some of the collimation

Question 28

is 180 or 360

Answer 28

180 is better because there is less interpolation between the points because you are getting double the points because its closer to the true curve