Abstract:
HTML
Purpose: Utilization of CT for three-dimensional lung lesion sizing is an
active area of research in recent years. Many studies have been conducted to
automatically identify and characterize lung nodules. In this paper, we perform
an analytical and experimental study to understand the impact of the CT
acquisition parameters and reconstruction algorithms on the accuracy of the
volumetric nodule measurement.
Methods and Materials: We examined two major factors in the data acquisition:
acquisition mode and slice aperture. Acquisition modes can be either
step-and-shoot or helical. For helical mode, different helical pitches are
used. We also analyzed two factors of the reconstruction algorithm:
reconstruction field-of-view and filter kernel selection. To understand their
impact, we first conducted a set of phantom experiments that isolate each
factor. Two phantoms were selected: plastic beads and chewing gums. The plastic
beads of known volumes are used to establish the ground truth, and the chewing
gums are used to examine the shape distortion. To further quantify the impact,
we derived a set of analytical models that provide transfer functions between
various acquisition and reconstruction parameters. We applied these analytical
models to predict the experimental results.
Results: Phantom experiments have shown that slice thickness plays an important
role in the nodule volume measurement. Our study shows that sub-mm slice
thickness (0.625mm) provides the most accurate volume measurement. When
measuring nodules of 6.7mm in diameter, a 2.5mm collimation can introduce an
error as large as 12%. When this collimation is combined with a high-pitch
helical (1.5:1), the error is increased to 21%. For the same object, the impact
of the reconstruction algorithm is equally significant. Compared to the
Standard algorithm, the volume measurements obtained with Lung and Bone
algorithms can differ by 14% and 12%, respectively.
Conclusion: Our study indicates that acquisition and reconstruction parameters
can significantly impact the accuracy of the nodule volume measurement.
Consequently, when conducting quantitative analysis on lung nodules, especially
in sequential growth studies, it is important to make appropriate adjustment
and correction to maintain the desired accuracy. In this study, we also
developed an analytical model to predict the parameter impact. The accuracy of
the model has been validated against experiments. (J.H., K.K. are employees of
GE Medical Systems.)
Hsieh PhD, J,
Impact of Acquisition and Reconstruction Parameters on the Accuracy of the Lung Nodule Measurement. Radiological Society of North America 2003 Scientific Assembly and Annual Meeting, November 30 - December 5, 2003 ,Chicago IL.
http://archive.rsna.org/2003/3102953.html
