Computer-assisted lung nodule volumetry from multi-detector row CT: Influence of image reconstruction parameters
Introduction
Small pulmonary nodules are frequently discovered by computed tomography (CT) of the chest, and the majority of these nodules turn out to be benign [1], [2]. Naturally, differentiation between benign and malignant nodules is of paramount importance, but this is difficult, and many studies have focused on resolving this problem. These studies have examined the morphology involved by evaluating the edge or internal characteristics of the nodules [3], [4], the relationships of the nodules to normal anatomical structures such as bronchi [5] and the enhancement pattern on CT [6], [7]. One of the most reliable criteria indicating benign lesions is the presence of calcification [3], [8]. But many nodules cannot be determined as benign lesions by CT, or by positron emission tomography or biopsy. Consequently, if the potential for malignancy cannot be ruled out, these nodules are usually followed in size by CT.
It is generally agreed that stability in the size of a nodule for more than 2 years is a good predictor of nodule dignity [8], [9]. Although the size of pulmonary nodules is usually determined and compared using two-dimensional CT evaluation, such measurements are not reliable for small nodules [10]. More reliable estimation of pulmonary nodule volume can now be obtained using three-dimensional volumetric measurement by computer software, with users recommending a thin slice-thickness and small field-of-view (FOV) [11], [12], [13]. CT images can be acquired using different parameters, and thus we investigated the effect of changing the reconstruction parameters on computer-aided volumetric measurements.
Section snippets
Materials and methods
The study involved 50 nonconsecutive patients (27 men, 23 women; average age, 58.3 years; range, 16–82 years) who were examined between September 2004 and February 2005. CT examinations were performed for clinical indication in all patients. If the applicable nodules were found after CT examinations, the data were used for this retrospective study. Patients with solid pulmonary nodules less than 2 cm in diameter, and with clear and regular margins were examined, while those with ground-glass
Unsuccessful volume measurement
All 50 nodules were solid, and located in the peripheral part of the lungs. The diameter of the 50 nodules was 9.1 ± 3.4 mm on average, ranging from 4.1 to 19.2 mm. Volume measurement for all 48 reconstruction parameters could be successfully performed in 39 of 50 nodules, and the diameter of these 39 nodules was 8.7 ± 2.7 mm on average, ranging from 4.3 to 16.4 mm. The number of unsuccessful volume measurements for each reconstruction parameter set, of the 50 nodules, is summarized in Table 1. For 11
Discussion
Recently developed computer software that automatically calculates three-dimensional nodule volume is now available. The software extracts the pulmonary nodules by recognizing density, gradient strength and a shape constraint of the nodule in a slice-by-slice manner, and there is no statistically significant difference between the nodule areas delineated by a radiologist and those obtained by the software [15]. The repeatability of three-dimensional volume measurement by automatic volumetric
Conclusion
Volumetric measurements of pulmonary nodules can change by changing the reconstruction parameters. In the present study, the measured volume of a nodule obtained using the high-spatial frequency algorithm and non-overlapping reconstruction was significantly larger than that using the low-frequency spatial algorithm and overlapping reconstruction. Although differences in slice thickness and FOV did not affect volume measurements, the frequencies of the maximum measured volume occurring with a 5
References (23)
- et al.
Early lung cancer action project: overall design and findings from baseline screening
Lancet
(1999) - et al.
Pulmonary nodule characterization: a comparison of conventional with quantitative and visual semi-quantitative analyses using contrast enhancement maps
Eur J Radiol
(2006) - et al.
The solitary pulmonary nodule
Eur J Radiol
(2003) Technical aspects of helical (spiral) CT
Radiol Clin North Am
(1995)- et al.
Results of three-year mass screening programme for lung cancer using mobile low-dose spiral computed tomography scanner
Br J Cancer
(2001) - et al.
CT of the solitary pulmonary nodule
AJR Am J Roentgenol
(1980) - et al.
Pulmonary lesions detected in population-based CT screening for lung cancer: reliable findings of benign lesions
Radiat Med
(2004) - et al.
Prevalence of air bronchograms in small peripheral carcinomas of the lung on thin-section CT: comparison with benign tumors
AJR Am J Roentgenol
(1991) - et al.
Lung nodule enhancement at CT: prospective findings
Radiology
(1996) Radiologic evaluation of the solitary pulmonary nodule
AJR Am J Roentgenol
(1990)
Are two-dimensional CT measurements of small noncalcified pulmonary nodules reliable?
Radiology
Cited by (32)
Management strategy of pulmonary nodule in 2013
2013, Diagnostic and Interventional ImagingCitation Excerpt :The variability between volume measurements using these algorithms may be as high as 25% [82] and it is therefore important to use the same software for the initial and repeat scans. The two scans also need to have the same acquisition parameters (level of inspiration, mAs, kV, collimation) and reconstruction (section thickness, filter) settings [83,84]. Overlapping millimeter sections are ideal, particularly for the smallest nodules [85].
Comparison of Semiautomated and Manual Measurements for Simulated Hypo- and Hyper-attenuating Hepatic Tumors on MDCT. Effect of Slice Thickness and Reconstruction Increment on Their Accuracy
2011, Academic RadiologyCitation Excerpt :The acceptable limit of ST might be dependent on the minimum tumor size, because the volume-averaging artifact is introduced when the ST approaches or exceeds the tumor size. Therefore, it is natural that smaller tumors are more vulnerable for measurement error with increasing ST. Our results are in agreement with the results reported in previous measurement studies in which measurement errors for pulmonary nodules or phantoms were assessed on CT images acquired with different z-axis spatial resolutions (7,10,12). Although we prepared somewhat large tumor models (13.3–50.7 mm), there were significant differences of accuracy among the ST and RI settings used.
Incidental lung nodules on CT examinations of the abdomen: Prevalence and reporting rates in the PACS era
2010, European Journal of RadiologyCitation Excerpt :Unfortunately nodule volumetry software is not implemented in every PACS workstations including ours. Moreover the accuracy and reproducibility of nodules volumes and doubling times calculations would have been affected by the use of different reconstruction parameters [14] and different contrast protocols in the CT scans we reviewed; software-based volume measurements of pulmonary nodules after administration of contrast medium are indeed higher than those of unenhanced scans [15]. In conclusion, radiologists should take time to review lung bases on abdominal CT studies taking advantage of PACS capabilities since lung pathology may be incidentally visible in it.
Effect of reconstruction algorithm on image quality and identification of ground-glass opacities and partly solid nodules on low-dose thin-section CT: Experimental study using chest phantom
2010, European Journal of RadiologyCitation Excerpt :It would therefore be preferable to determine the CT protocol for detection of small pulmonary cancers, including lung metastases, with an optimum radiation dose. In addition, several factors, including reconstruction algorithms, have an indirect effect on CT images, and their selection influences the operator [16–18]. In spite of these considerations, however, to the best of our knowledge no major studies have been reported that evaluated the influence of a reconstruction algorithm on detection of GGO and partly solid nodule on low-dose chest CT examination.
Morphological Analysis of Pancreatic Cystic Masses
2010, Academic RadiologyCitation Excerpt :We used the smallest slice thickness that we use in our clinical practice to segment these lesions. We did not evaluate thinner or thicker slices, although one previous study evaluated the effect of slice thickness on volumetry of small lesions (15). We did not evaluate the follow-up imaging studies to assess changes in the morphology of these masses over time.