Iterative Reconstruction Algorithms

Question 1

A class of
methods that start from an initial estimate of the
image, then repeatedly improve it by comparing its
simulated projections (from the current estimate) to
the measured data, adjusting the image to reduce
error, often incorporating statistical / physical
models, regularization, etc

Answer 1

Iterative reconstruction (IR)

Question 2

Generally yields better suppression of noise and
artifacts, allows dose reduction, handles
incomplete or noisy data better.

Answer 2

Iterative reconstruction (IR)

Question 3

Lower radiation dose (up to 50% or more reduction).

Answer 3

Iterative reconstruction (IR), Advantage

Question 4

Improved image quality, especially in low-dose scans.

Answer 4

Iterative reconstruction (IR), Advantage

Question 5

Reduced image noise and streak artifacts

Answer 5

Iterative reconstruction (IR), Advantage

Question 6

Better contrast resolution

Answer 6

Iterative reconstruction (IR), Advantage

Question 7

Longer reconstruction time (though newer hardware and algorithms have reduced this).

Answer 7

Iterative reconstruction (IR), Disadvantage

Question 8

Higher computational demand

Answer 8

Iterative reconstruction (IR), Disadvantage

Question 9

May produce a “plastic-like” or overly smooth image appearance if over-regularized.

Answer 9

Iterative reconstruction (IR), Disadvantage

Question 10

Types of Iterative reconstruction (Depending on Vendor and Complexity)

Answer 10

1.Statistical Iterative Reconstruction (SIR)
2.Model-Based Iterative Reconstruction (MBIR)

3.Hybrid IR (HIR)

Question 11

Models noise statistics

Answer 11

Statistical Iterative Reconstruction (SIR)

Question 12

Includes both system geometry and noise modeling for high precision.

Answer 12

Model-Based Iterative Reconstruction (MBIR)

Question 13

Combines FBP and IR to balance
speed and quality.

Answer 13

Hybrid IR (HIR)

Question 14

Works directly with the raw
projection data, not just the
image.

Answer 14

SIR

Question 15

It incorporates statistical noise
models to improve the
accuracy of the reconstruction.

Answer 15

SIR

Question 16

Process:
A. Starts with an initial image estimate.
B. Uses Poisson statistics (to model X-ray photon noise).
C. Iteratively compares the estimated projections to the actual raw data.
D. Updates the image to minimize the statistical difference between them.

Answer 16

SIR

Question 17

SIR, Examples by Vendor
GE=

Answer 17

Veo

Question 18

SIR, Examples by Vendor
Siemens=

Answer 18

SAFIRE (Sinogram Affirmed Iterative Reconstruction)

Question 19

SIR, Examples by Vendor
Philips=

Answer 19

iDose⁴ Premium

Question 20

SIR, Examples by Vendor
Canon

Answer 20

AIDR 3D Enhanced

Question 21

Better noise suppression

Answer 21

SIR, Advantage

Question 22

More accurate handling of low-dose data.

Answer 22

SIR, Advantage

Question 23

Reduces streak artifacts and improves contrast.

Answer 23

SIR, Advantage

Question 24

More computationally intensive.

Answer 24

SIR, Disadvantage

Question 25

Reconstruction time is longer

Answer 25

SIR, Disadvantage

Question 26

May cause a slightly smoother or artificial texture if over-iterated

Answer 26

SIR, Disadvantage

Question 27

Most advanced and computationally complex form.

Answer 27

MBIR

Question 28

It models everything that affects image formation — including scanner geometry, X-ray physics, noise statistics, and detector response.

Answer 28

MBIR

Question 29

Process: A. Simulates how X-rays travel through the body and interact with tissues. B. Incorporates models for beam hardening, scatter, and system blur. C. Iteratively refines the image until simulated and measured data match within very tight tolerances.

Answer 29

MBIR, Process

Question 30

MBIR, Examples by Vendor GE=

Answer 30

Veo MBIR

Question 31

MBIR, Examples by Vendor Siemens=

Answer 31

ADMIRE (Advanced Model Iterative Reconstruction

Question 32

MBIR, Examples by Vendor Philips=

Answer 32

IMR (Iterative Model Reconstruction)

Question 33

MBIR, Examples by Vendor Canon=

Answer 33

FIRST (Forward-projected Model-based Iterative Reconstruction SoluTion)

Question 34

Best image quality and lowest possible noise.

Answer 34

MBIR, Advantage

Question 35

Enables very low radiation doses (up to 80% reduction).

Answer 35

MBIR, Advantage

Question 36

Excellent edge preservation and artifact reduction.

Answer 36

MBIR, Advantage

Question 37

Extremely computationally demanding — may take several minutes to reconstruct a single image set (though newer hardware accelerates this)

Answer 37

MBIR, Disadvantage

Question 38

Images may appear overly smooth or “plastic” if overregularized

Answer 38

MBIR, Disadvantage

Question 39

Hybrid Iterative Reconstruction (HIR) aka

Answer 39

Adaptive or Partial Iterative Reconstruction

Question 40

Combines the speed of Filtered Back Projection (FBP) with some of the noise- and artifact treducing capabilities of iterative methods

Answer 40

HIR

Question 41

It’s a “bridge” between traditional and fully iterative method

Answer 41

HIR

Question 42

Process: A. The CT image is first reconstructed using FBP. B. An iterative process then refines the image, usually in the image domain (not the raw data domain). C. The algorithm reduces image noise by comparing the image to statistical models and smoothing out inconsistencies.

Answer 42

HIR

Question 43

Faster than full iterative reconstruction

Answer 43

HIR, Advantage

Question 44

Moderate noise reduction and dose savings (typically 30–40%

Answer 44

HIR, Advantage

Question 45

Compatible with existing CT hardware

Answer 45

HIR, Advantage

Question 46

Doesn’t fully model the physics of the scanner (like detector geometry or photon statistics)

Answer 46

HIR, Diasadvantage

Question 47

Less accurate than model-based IR

Answer 47

HIR, Diasadvantage