EPISODE 6

THE TUFFEST STUFF CT REGISTRY

REVIEW SEMINAR SOLUTION
TYPES OF DATA REVIEW
IMAGE PROCESSING
SECTION 1, PRESENTATION F
GENERIC CT SYSTEM RECONSTRUCTION

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 2

TYPES OF DATA REVIEW
• MEASUREMENT DATA = SCAN DATA
• Arises from ALL of the exit beam photon quanta incident on the detectors
• RAW DATA
• The data manipulated in both PROSPECTIVE & RETROSPECTIVE RECONSTRUCTION to
make CT images
• CONVOLVED DATA
• This is FILTERED DATA via the application of a KERNEL ALGORITHM
• FILTERED BACK PROJECTION DATA
• The CT NUMBER data received by a matrix after RECONSTRUCTION
• As Filtered Back Projection INCREASES, Image quality INCREASES
• IMAGE DATA
• What is viewed on the monitor in the form of a CT IMAGE
• The monitor is composed of a matrix of pixels shaded in their individual shade of gray
based on their related CT Number
THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 3
 TYPES OF DATA REVIEW
• MEASUREMENT DATA = SCAN DATA
• Measurements that arise from the detectors
• Subject to Preprocessing to correct for data set errors
otherwise poor image quality, may generate image
artifacts. Correct errors due to:
• Beam hardening
• Bad detector reading adjustment
• Scatter radiation
THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 4
 TYPES OF DATA REVIEW
• RAW DATA
• The results of the Preprocessing of scan data
• Erroneous data have been removed from the data set
• Scatter radiation
• Bad detector readings
• Beam hardening
• Subjected to the individual scanner image
reconstruction algorithms in both prospective and
retrospective reconstruction
THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 5
 TYPES OF DATA REVIEW

• CONVOLVED DATA
• The removal of blur through the application of the
CONVOLUTION TECHNIQUE
• Application of a kernel mathematical filter
• Convolution kernels can only be applied to RAW DATA

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 6

 A Convolution “kernel”
The averaging of 9 values to create 1 value
As applied to raw data

0.70

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 7

 TYPES OF DATA REVIEW

By a process known as FILTERED BACK PROJECTION,

attenuation profile data are converted to a matrix

The greater the number of profiles back-projected the

sharper the image

During back-projection
CT NUMBERS are being electronically distributed onto a
matrix that has depth

THE TUFFEST STUFF CT LIVE LECTURE SEMINAR

 TYPES OF DATA REVIEW
• IMAGE DATA = RECONSTRUCTED DATA
• Convolved data that have been filter back-projected onto an
image matrix to create the CT images displayed on a monitor

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 9

 TYPES OF DATA REVIEW
PROJECTION & BACK-PROJECTION
• OBJECT/PROJECTION
• 1. The scanner provides the RAW DATA which is converted to electrical impulses
• 2. The electrical impulses are received by the ADC where they are sampled, quantified &
coded as BINARY DATA

• RECONSTRUCTION/BACKPROJECTION
• 1. In the CT COMPUTER the BINARY DATA is filtered & reconstructed. The BINARY DATA is
then BACK PROJECTED onto a matrix to eventually be viewed as IMAGE DATA.

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 10

 TYPES OF DATA REVIEW
THE ARRAY PROCESSOR
Converting digitized Scan Data to digitized Image Data
• IMAGE PROCESSING
• RECONSTRUCTION ALGORITHMS
• Filtered back projection reconstruction
• Fourier Transform
• Iterative reconstruction
• Interpolation
• Other functions
• From raw data to image data
• Prospective and retrospective reconstruction
• Reconstruction interval
• Post-processing
• Multi-planar Reformation (MPR)
• 3D Rendering (MIP, SSD, VR)
• Quantitative analysis
• E.g. distance, diameter, calcium scoring, ejection fraction
THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 11
 IMAGE PROCESSING
• ESSENTIAL ELEMENTS OF IMAGE PROCESSING
• PERFORMED BY THE CT COMPUTER, AKA, ARRAY
PROCESSOR
• Receives the digital signal from the ADC, and applies
the respective reconstruction technique
• Ie. FILTERED BACK PROJECTION, FOURIER TRANSFORM,
ITERATIVE RECONSTRUCTION, etc.
• Performed to produce a desired output digital image
in binary digit form
• Program to reduce input image noise
• Program to enhance input image sharpness
• Alter input image contrast
THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 12
 IMAGE PROCESSING
• ESSENTIAL ELEMENTS OF IMAGE PROCESSING
• Digital images are numerical representatives or images of objects
• The ADC converts continuous analog signals to discrete signals or
digital data (BINARY DATA)

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 13

 INTERPOLATION
“Take the SLANT out of the SLICE”
• DEFINITION
• A mathematical technique used to estimate the
value of a function from known values on either
side of the function
• Used in the image reconstruction process
• Used in the determination of helical CT slices

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 14

 Reconstruction Terminology

• Interpolation is a mathematical method of estimating the

value of an unknown function using the known value on
either side of the function

Given the values listed, calculate the

estimated values of X & Y

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 15

 RECONSTRUCTION ALGORITHM
• DEFINITION OF RECONSTRUCTION ALGORITHM
• Mathematical procedure applied to convert
attenuation readings into a CT image
• Available as reconstruction computer software
• Methods include:
• Filtered back projection
• Iterative methods
• Fourier Transform

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 16

 PROSPECTIVE vs RETROSPECTIVE
RECONSTRUCTION
• PROSPECTIVE RECONSTRUCTION
• Occurs during the exam
• Refers to the ability to view images as the individual study is conducted
• RETROSPECTIVE RECONSTRUCTION
• Reconstruction that occurs upon completion of the individual study
• Reconstruction that involves the recollection of raw data to perform an
individual function

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 17

 RECONSTRUCTION INTERVAL

• WHAT AM I?
• THE RECONSTRUCTION INTERVAL defines the degree
of OVERLAP between reconstructed axial images
• No effect on scan time or patient exposure
• Works independent of x-ray beam collimation
• Works independent of image thickness
• However, increased overlap improves Z-axis resolution, especially
for 3D & MPR images

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 18

 RECONSTRUCTION INTERVAL
• MPR & 3D APPLICATIONS
• For routine MPR & 3D, a 30% overlap is sufficient generally
• 1-mm section thickness with 0.7-mm reconstruction
interval
• For cardiac, 50% overlap is desirable
• 0.5-mm section thickness with 0.25 reconstruction
interval
• display, image archiving expenses are increased

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 19

 RECONSTRUCTION INTERVAL

• Results of too much overlap

• Large number of images
• Increased reconstruction time
• Longer interpretation periods
• Increased image handling overhead costs without an
appreciable increase in quality
• Image transfer, image display, image archiving expenses

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 20

 MULTIPLE ROW DETECTOR
SPATIAL RESOLUTION
• IN THE AXIAL OR X-Y PLANE As any of the influencing factors
• In CT, SR is quite high increases, Spatial Resolution
• 10-20 line pairs per cm DEcreases
• IN THE LONGITUDINAL OR Z AXIS
PLANE
• 7 – 15 line pairs per cm
• INFLUENCING FACTORS
• Detector size
• Reconstruction thickness
• PITCH
THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 21
ANALYTIC RECONSTRUCTION METHODS
• FILTERED BACK PROJECTION, aka, CONVOLUTION BACK PROJECTION
• Most common reconstruction method used in modern
CT
• Convolution is a digital image processing technique that
employs the application of kernels to modify an image
• FOURIER TRANSFORM
• The most straight forward method for reconstruction
• Derives directly from the projection slice theorem

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 22

 FILTERED BACK PROJECTION
• AKA, THE CONVOLUTION METHOD
• STEPS:
• Obtain projection profiles
• Obtain data logarithm
• Multiply log values by a convolution filter to generate a set of
filtered profiles
• Convolution kernels
• Back project the filtered profiles
• When summing, the negative and positive components cancel
one another. A blur free image is the result.
THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 23
 FOURIER TRANSFORM

• Used to reconstruct images of a patients anatomy in CT &

MRI
• Mathematical function that converts a signal in the Spatial
Domain to a signal in the Frequency Domain
• With Fourier Transform a signal waveform is divided into
a series of sine and cosine functions that possess
different amplitudes and frequencies.

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 24

 FOURIER TRANSFORM
• Example:
• E.g. Sound waves that enter the
ear are separated into different
signals & intensities. Arriving at
the brain, these signals are
interpreted and are rearranged
(relative to amplitude &
frequency) to produce a
perception of the original sound.

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 25

 THE 6 STEPS OF RECONSTRUCTION
• FROM THE CT DETECTORS TO THE ADC TO THE CT COMPUTER…
• PREPROCESSING
• Filter SCAN DATA = MEASUREMENT DATA and correct for data set errors
due to beam hardening, bad detector reading, scatter radiation
• REFORMATTED RAW DATA
• This is scan data that has been corrected by preprocessing
• CONVOLUTION WITH FILTER
• IMAGE RECONSTRUCTION ALGORITHM
• Back projection of convolved data
THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 26
 THE 6 STEPS OF
RECONSTRUCTION

• RECONSTRUCTED IMAGES OF CT
NUMBERS
• IMAGE STORAGE, DISPLAY,
RECORDING, ARCHIVING

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 27

 A Convolution “kernel”
The averaging of 9 values to create 1
As applied to raw data

THE TUFFEST STUFF CT REGISTRY REVIEW SEMINAR 28

 TYPES OF DATA REVIEW

By a process known as FILTERED BACK PROJECTION,

attenuation profile data are converted to a matrix

The greater the number of profiles back-projected the

sharper the image

During back-projection
CT NUMBERS are being electronically distributed onto a
matrix that has depth

THE TUFFEST STUFF CT LIVE LECTURE SEMINAR

Things to study…
1. These are the most important data types in CT. What do you know
about them?
1. Measurement Data
2. Scan Data
3. Raw Data
4. Image Data

2. To complete the discussion, what are Convolved Data and Filtered

Back Projection Data?

THE TUFFEST STUFF CT REGISTRY REVIEW SOLUTION

Things to study…
3. So what is Interpolation? What is the function of interpolation?

4. Can you define the difference between Prospective and

Retrospective Reconstruction?

5. Reconstruction interval is a fancy term for __________________.

6. The Axial Plane consists of which orthogonal axes?

7. The Longitudinal Plane consists of which orthogonal axis?

THE TUFFEST STUFF CT REGISTRY REVIEW SOLUTION

Things to study…
8. Listed are the 6 steps of Reconstruction. Write the important details
of each step.
1. Preprocessing of Scan Data/Measurement Data
2. The Reformatting of Raw Data
3. Convolution with Filter
4. Image Reconstruction Algorithm application
a. The back projection of convolved data follows
5. Reconstructed images of CT Numbers
6. Image storage, display

THE TUFFEST STUFF CT REGISTRY REVIEW SOLUTION