Iterative reconstruction refers to an image reconstruction algorithm used in CT that begins with an image assumption, and compares it to real time measured values while making constant adjustments until the two are in agreement.
Computer technology limited early scanners in their ability to perform the iterative reconstruction. However, this image reconstruction algorithm is now widely used due to the improvement of computer technology over the past decade.
Its ability to overcome noise associated with filtered back projection without increasing radiation dose has had a significant impact on the computed tomography image reconstruction industry 1-4.
On this page:
Stages of iterative reconstruction
Iterative reconstruction has three distinct stages. This is, of course, a simplified overview of iterative reconstruction without statistical modelling 1,3.
Input
Using the raw data produced by the computed tomographic scanner a standard filtered back projection algorithm is utilized to create a primary image.
Image reconstruction loop
A sequence is then performed where
- a forward projection to the primary image creates artificial raw data,
- simulated data is then correlated to the measured raw data where an updated image is generated, and then
- a filtered back projection is used to back-project the updated image onto the new updated image; this is repeated until the differences in the images reach a preset value.
Output
The final back projection image that is produced.
Model-based iterative reconstruction
Iterative reconstruction has undergone notable improvement with the introduction of 'model-based iterative reconstruction'. Reconstructions models that can take into account optics, physics, scanner, or, noise statistics are proving to reduce image noise effectively while keeping the patient dose as low as reasonably achievable 1,2,5.
Unable to process the form. Check for errors and try again.