BACK TO
SEARCH
CODE:
Volume Quantitation of Small Pulmonary Nodules on
Low-dose Chest CT: A Phantom Study
DATE: 11/26/2001 - 11/26/2001
LOCATION: Room E450B
(C17)
Chest (Nodule Detection and Analysis)
(C17-374)
Detection of Calcium in Small Pulmonary Nodules Using Multi-slice CT
Scanners: Implications for Lung Cancer Screening
(C17-383)
Low-Dose Lung Cancer Screening Using Multidetector-row CT System:
Utility of the MIP Image for Improvement of Detectability
(C17-375)
Wavelet Compression of Low-dose Chest CT: Effect on Nodule Detection
(C17-376)
Missed Lung Cancers in Low-dose Helical CT Screening Obtained from a
General Population
(C17-377)
Volumetric Growth Index (VGI) for Small Pulmonary Nodules: Development
and Validation
(C17-378)
Automated Segmentation and Volumetry of Small Pulmonary Nodules:
Comparison of Low Dose and Standard Dose 1-mm Multidetector-Row CT
Studies
(C17-379)
Volume Quantitation of Small Pulmonary Nodules on Low-dose Chest CT: A
Phantom Study
(C17-380)
Performance of Automated CT Lung Nodule Detection on Missed Cancers
(C17-381)
Patient-specific Models for Lung Nodule Detection and Surveillance in
CT
(C17-382)
Maximum Intensity Projection (MIP) Processing in the Detection of Small
Pulmonary Nodules: Incremental Benefit Using Multi-detector Computed
Tomography (CT) Data Sets
PARTICIPANTS:
Jane
Ko MD
Henry
Rusinek PhD
Ramesh
Chandra PhD
Georgeann
McGuinness MD
Margrit
Betke PhD
David
Naidich MD
Abstract:
Purpose: To evaluate 3D algebraic and threshold
methods for volumetric quantitation of small pulmonary nodules on
low-dose multi-detector computed tomography (CT) that is used in
lung-cancer screening.
Methods and Materials: 40 plastic nodules (20 ground glass and 20
soft-tissue density) were embedded in a chest phantom comprised of
material with the attenuation of lung parenchyma (-790 HU). Four groups
of nodules contained 5 nodules each, all <5 mm in diameter. Nodule
mass and specific gravity were used to calculate the true volume (V).
The phantom was imaged using a multi-detector CT scanner (0.5
sec/rotation, 140 kVp, 1.0 mm collimation) using low-dose (20 mAs)
technique. Data were reconstructed in 1.25 mm sections at 1.0 mm
intervals, FOV 38 cm using a high-frequency reconstruction algorithm.
Imaged nodule volumes were calculated using three 3-D quantitative
techniques: (a) an algebraic method that uses density in a region of
interest to correct for partial volume effect, (b) a fixed threshold
method (FT) that uses a threshold of -500 HU, and (c) a variable
threshold method (VT) that uses a threshold calculated from the HU of
pure nodule and adjacent lung. Measured volumes were correlated to Vs,
and residual volume errors were calculated. Any effect of nodule
attenuation, size, and location within the lung on volume error was
identified.
Results: The mean Vs of the 4 nodule size categories were 63.2 +/- 4.5
mm3, 36.3+/- 2.8 mm3, 18.6 +/- 1.4 mm3,
8.0 +/- 0.6 mm3 correlating with diameters of 4.9, 4.1, 3.3,
2.5 mm. The algebraic method's mean residual volume error (3.7 mm3,
r=0.97) was significantly lower (P=0.024) than the volume error of the
VT (5.7 mm3, r=0.94) and FT (5.6 mm3, r=0.93)
methods. The algebraic method had a significant (P<0.001) increase in
residual error for ground-glass nodules (5.37 mm3) as
compared to solid nodules (1.99 mm3) while VT and FT did not.
The mean residual volume error for Vs of 63.2, 36.3, 18.6, 8.0 mm3
were 7.06, 5.08, 4.33, 3.48 mm3, respectively. For nodule
location, mean residual volume errors for anterior, medial, central,
posterior, and lateral positions were 5.39, 5.60, 4.57, 5.72, 3.26 mm3,
respectively.
Conclusion: A 3-D algebraic method can improve nodule quantitation by
decreasing the partial volume effect and can potentially detect changes
in nodule volume in patients who undergo low-dose multi-detector
screening CT.
|