banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Optimization of a tomosynthesis system for the detection of lung nodules
Rent this article for
View: Figures


Image of FIG. 1.
FIG. 1.

Diagram used for optimization. Note that the estimation and detection tasks are limited by the anatomical background not the SNR, and the SBDX system is fast enough to eliminate motion blur. Through this hierarchical approach we reduce the optimization of the system parameters to the effect of two surrogates of image quality on two clinical tasks.

Image of FIG. 2.
FIG. 2.

Base CT slice from lung phantom with a nodule added in multiple locations.

Image of FIG. 3.
FIG. 3.

Sample images from those used in a 2-AFC experiment for projections (top) and for a tomosynthetic angle of 3 degrees (bottom). In both cases the nodule is in the image on the left in the center of the image.

Image of FIG. 4.
FIG. 4.

Laguerre-Gauss basis and fit to a signal. We use only five Laguerre-Gauss basis functions with a scaling factor of 8.5 and obtain a reasonable approximation of the mean signal.

Image of FIG. 5.
FIG. 5.

Eye filters give different weights to the frequency components of the image. The Eye1 filter (, ) models the average human observer using a peak of 1.3 cycles per degree and the Eye4 filter (, ) using a peak of four cycles per degree. The best filter uses (, ) used to maximize the percent correct.

Image of FIG. 6.
FIG. 6.

Results from 2-AFC studies. (a) Distribution of the percent correct for the human observers using the high resolution dog-lung phantom (150 microns), (b) binned to a resolution of 600 microns.

Image of FIG. 7.
FIG. 7.

Estimation of diameter using images with nodules used in the 2-AFC detection study.

Image of FIG. 8.
FIG. 8.

Results from mathematical observers that (a) try to achieve the optimal performance and (b) track average human performance.

Image of FIG. 9.
FIG. 9.

Results from a mathematical observer which uses the nodule as a template with no eye filter (NPW) and another which uses all of the data to find a close to perfect fit to the human data (NPWE1).


Article metrics loading...


Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Optimization of a tomosynthesis system for the detection of lung nodules