MEDRADSC 3K03 Lecture 9 - Image Display

Back projection

• Process of converting data from the attenuation profile to a matrix

• Compiles all info from all attenuation profiles to create image

Problems with back projection

Does not produce a sharp image (star pattern)

Back projection is a ___ method

Summation

Filtered back projection is a ___ method

Convolution

Filtered back projection steps

1. All projection profiles gathered

2. Log of data obtained

3. Log values multiplied by a convolution filter to generate a set of filtered profiles

4. Filtered profiles are back-projected

5. Filtered profiles are summed. -ve and +ve components canceled → image free of blurring

ASIR stands for

Adaptive statistical iterative recon

ASIR

Start w assumed image → compare measurements/projections of actual data with expected → update image based on differences

Advantages of using ASIR

Can reduce noise and pt dose up to 50% currently

ASIR steps

1. Input

• Std filtered back projection algorithm utilized

2. Image reconstruction loop

3. Output

• Final back projection image prod

What is a recon filter?

Mathematical pre-processing function applied to each detector value in each attenuation profile, involving intentional distortion of 1 value by adding weighted neighbours’ values

Function of recon filter

Deblurring filter to undo blur effect during back projection; converts raw data into image data

2 types of recon filter

Smoothing and enhancing

Smoothing recon filter

Dec artifacts but dec spatial resol

Enhancing recon filter

Inc spatial resol but dec contrast resol

Another word for filters

Kernels

Recon algorithms 

1. Soft

2. Std

3. Lung

4. Detail

5. Bone

6. Edge

7. Bone plus

Soft recon algorithm

Useful for tissues w similar densities w contrast

Std recon algorithm

For routine exams like chest, abdo, pelvis

Lung recon algorithm

Interstitial lung pathology

Detail recon algorithm

For post myelograms, where hybrid tissue density detail and bone edges are impt

Bone recon algorithm

For high resol exams and sharp bone detail

Edge recon algorithm

For small bone work in head, as well as high resol scans

Bone plus recon algorithm

For sub mm detailed head work (e.g internal auditory canal)

Is window and algorithm the same?

No

CT number formula

CT number is based on

LAC of H2O

What does a CT number correspond to?

Signal intensity/brightness value

What if the CT numbers are inaccurate?

Misdiagnosis

In an acquired CT image, the CT number of muscle was measured to be 53. What is the LAC for this tissue given an LAC of water is 0.280 at beam energy?

0.295

ROI stands for

Region of interest

What is an ROI?

Provides an averaged measurement of all of the pixels within the ROI

When an ROI is used, the ___ is also displayed

Standard deviation

What does the cursor (+) display?

Measurement of the HU of the pixel that the cursor covers

HU measurements may be affected by

Volume averaging or image noise

Why is this ROI not appropriate?

Piece of Ca in ROI → SD  = 47.25 → make circle smaller so it’s not covering Ca

Why is this ROI not appropriate?

Beam hardening incl in ROI → SD = 202.01

Why is this ROI not appropriate?

Measuring contrast enhanced kidney with non-contrast enhanced tissue → SD = 53.77

When is an ROI used?

Measure HU for contrast running through an area. Monitor ideal peak enhancement to start the scan

Why do we measure the distance in images?

• Reporting size of an abnormality

• Placement of biopsy needle

• Drainage apparatus

Why is it better to use an ROI than the cursor tool when measuring HU?

ROI is less susceptible to variation

How can we measure the distance?

• Grid for CT guided procedures

  • Radio opaque markers that can be seen in images

• Scale alongside images

• Size reference

• Ruler

Magnification vs DFOV

Magnification	DFOV
No improvement in spatial resolution	Improves spatial resolution
No change in pixel size	Changes pixel size

Other display options

• Image annotations

• Multiple image display

• Histogram

• MPRs and 3D formatting