by Robert Frysch, Richard Bismark, Andreas Maier, Georg Rose
Abstract:
In computed tomography (CT), it is desirable to scan only a limited field of view in order to reduce the overall patient dose, when there is only a small region inside the patient body that is of interest. The CT reconstruction problem gets more challenging in that circumstance and typical truncation artifacts occur in the reconstructed volume , even for iterative reconstruction techniques. Given such a volume-of-interest problem, we achieve an improvement of image quality compared to the standard algebraic reconstruction technique, when applying Kaczmarz's algorithm to seek the optimal truncated forward projection for the normal equation instead of the whole volume vector x for the system Ax = b directly. In this work, we derive and motivate this method as well as show a proof of concept on simulated data.
Reference:
Ray-Density Weighted Algebraic Reconstruction for Volume-of-Interest CT (Robert Frysch, Richard Bismark, Andreas Maier, Georg Rose), In Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine, volume 14, 2017.
Bibtex Entry:
@inproceedings{frysch_ray-density_2017,
	address = {Xi'an, China},
	title = {Ray-{Density} {Weighted} {Algebraic} {Reconstruction} for {Volume}-of-{Interest} {CT}},
	volume = {14},
	abstract = {In computed tomography (CT), it is desirable to scan only a limited field of view in order to reduce the overall patient dose, when there is only a small region inside the patient body that is of interest. The CT reconstruction problem gets more challenging in that circumstance and typical truncation artifacts occur in the reconstructed volume , even for iterative reconstruction techniques. Given such a volume-of-interest problem, we achieve an improvement of image quality compared to the standard algebraic reconstruction technique, when applying Kaczmarz's algorithm to seek the optimal truncated forward projection for the normal equation instead of the whole volume vector x for the system Ax = b directly. In this work, we derive and motivate this method as well as show a proof of concept on simulated data.},
	booktitle = {Fully {Three}-{Dimensional} {Image} {Reconstruction} in {Radiology} and {Nuclear} {Medicine}},
	author = {Frysch, Robert and Bismark, Richard and Maier, Andreas and Rose, Georg},
	month = jun,
	year = {2017}
}