Diagnostic unenhanced clinical thoracic CT examinations were performed utilizing 64-slice or 128-slice multidetector computed tomography scanners (Definition, Definition AS+ or Definition Flash, Siemens AS, Malvern, PA) with protocols using 120 to 140 kVp, gantry rotation time of 0.33 to 0.5 seconds, 512×512 matrix, and reference tube current of 70 to 160 mAs. The images were reconstructed with either b46 or b60 kernels and a section thickness ranging from 1 to 1.5 mm at 0.8 to 1 mm intervals.
Patients were instructed to fast starting midnight before intravenous administration of 15±1.5 mCi of 18F-FDG. PET imaging was performed 60 to 70 minutes after FDG administration with integrated PET and 64-slice or 16-slice CT devices (Discovery 690 or Discovery RX; GE Medical Systems). All images were acquired during shallow breathing. The acquisition time for PET in 3D and time-of-flight mode was 3 to 5 minutes per table position. PET data were resampled from either a 128×128 or a 192×192 matrix to a 512×512 matrix to match the CT data to allow attenuation correction, and a CT transmission map was generated. PET image data were reconstructed iteratively using the ordered subsets expectation maximization algorithm with segmented attenuation correction (2 iterations, 35 to 36 subsets) and the CT data. The CT protocol used 120 kVp, gantry rotation time of 0.5 seconds, a pitch of 1 (1.5 for the 16-slice CT), and adjustable reference tube current depending on the patient’s body mass index, ranging from 20 to 90 mAs (maximum 120 mAs for the 16-slice CT) for a standard patient. A de-noising algorithm was applied to CT images of patients with body mass index >35. The transaxial CT slice thickness was 3.75 mm with an interval of 3.27 mm for fusion with the transaxial PET images. The PET, CT, and fused images were available for review in the axial, coronal, and sagittal planes by means of software (Advantage Windows; GE Medical Systems) on a computer workstation.
MR examinations were uploaded onto a stand-alone workstation (Leonardo VE30A, Siemens Healthcare) for analysis. Regions of interest (ROIs) were independently placed by a second-year medical student (V.T.) after a training session of 5 cases and by 1 thoracic radiologist with 7 years of thoracic imaging experience (C.W.K.), both blinded to clinical data at the time of evaluation. The ROIs encompassed the entire lesion with the highest signal intensity for the T1, T2, T2*, and diffusion-weighted sequences (Fig. 2A). ROI size was kept constant over the range of b values for the DWIs. Signal intensities derived from the ROIs as a function of b value were fitted to a monoexponential decay model whose decay rate was the ADC as derived from the following equation, with ADC and S0 as free parameters: Sb=S0 exp (−b×ADC), where Sb is the signal intensity for a given b factor, and S0 is the signal intensity without diffusion weighting. For T2* values, ROIs were drawn over each lesion to determine the signal intensity for each TE. Signal intensity at each echo time, SE, is given as follows: SE=S0 exp (−T2*/TE), where S0 is the signal at zero TE. T2* value was calculated by nonlinear least-squares fitting of the above equation to measured values of SE, with T2* and S0 as free parameters. ROIs were redrawn at least 2 weeks later by each reader and the analysis repeated.
A single reader (C.W.K.) performed all CT assessments. For CT Hounsfield unit (HU) measurements, ROIs were placed once to encompass the entire lesion (Fig. 2B). Similarly, an ROI was drawn to encompass the solid component, which was defined as the part of the lesion where the normal pulmonary architectures (vessels, airways, and alveoli) were completely obscured. The maximal diameter on axial images in which the nodule appeared the largest was determined. The longest linear dimension of the solid component was also recorded on the axial section in which the solid component appeared the largest. Percentage solid component was visually stratified into 4 groups: <25%, 25% to 49%, 50% to 75%, and >75%.
Statistical analysis, detailed below, was performed to assess the relationship between ADC, T2*, and measured signal intensity values on T1-weighted and T2-weighted images and benignity or malignancy. For patients with pathologically proven adenocarcinoma, MRI parameters were correlated to degree of invasiveness and histologic subtype. Statistical correlation between the ADC, T2*, T1, and T2 values and maximum standard uptake values (SUVmax) was performed in patients with preoperative PET or PET/CT examinations. MRI parameters were assessed for correlation with HU measurements as well. Lesion visibility, defined as whether or not a lesion was seen on a particular MRI sequence, was also correlated with solid component size and percentage of solid component on CT.
Statistical analyses were performed with the SAS System (version 9.4; Cary, NC). The analysis was conducted on a per-nodule basis under the assumption of statistical independence, accounting for a limited number of nodules nested within cases (32 nodules in 28 cases). Given multiple measurements that resulted from 2 measurements by each of 2 readers, the mean of the 4 values was utilized for diagnostic performance and correlation analysis. Continuous variables were reported as mean (SD) and categorical variables as frequency (percentage). The Spearman rank order correlation coefficient was utilized to assess for degree of correlation. Receiver operating characteristic curve analysis was performed to predict the association between malignancy and MRI parameters and to obtain sensitivity, specificity, and accuracy measurements. Intraclass correlations (ICC) were calculated to validate the intrareader and interreader reliabilities between readings for each rater. An overall ICC was also computed to summarize the variability of the 4 measurements within each nodule using a random-effects model with a random nodule effect. As proposed by Landis and Koch,37 ICC values >0.8 were regarded as “almost perfect,” those between 0.61 and 0.8 as “substantial,” those between 0.41 and 0.6 as “moderate,” and those between 0.21 and 0.4 as “fair.” Statistical tests with a 2-sided P≤0.05 were considered significant.
The mean ADC for benign nodules ranged from 1.1 to 1.3 μm2/ms, whereas malignant lesions ranged from 1.6 to 2.3 μm2/ms (Table 3). Mean T2* values for benign nodules ranged from 13.2 to 16.4 ms, and those for malignant nodules ranged from 12.2 to 24.2 ms. Mean T1 signal intensities for benign nodules ranged from 75.9 to 87.5, and those for malignant lesions ranged from 78.1 to 87.6. Mean T2 signal intensities for benign lesions ranged from 91.9 to 104.5, and those for malignant lesions ranged from 99.6 to 110. Of the 32 nodules, 32 were visualized on T1-weighted, 31 on T2-weighted, 30 on DWI, and 29 on T2*-weighted sequences.
The mean CT density for benign lesions was −396.6 HU, and that for malignant lesions was −424.0 HU. The mean CT density of the solid component for benign nodules was −226.9 HU, and that for malignant nodules was −248.6 HU. Data for solid component on CT were available for 30 lesions, as 2 outside CT results were unavailable at the time of data analysis. Four lesions contained <25% solid components, 7 contained 25% to 49% solid components, 9 contained 50% to 75% solid components, and 10 had >75% solid components. The mean diameter (SD) of the solid component for benign nodules was 1.3 cm (0.8), and that for malignant nodules was 1.5 cm (0.9). The maximum standardized uptake values were available for 4 benign lesions with a mean (SD) of 3.0 (2.3) and for 9 malignant lesions with a mean of 2.8 (1.7).
ADC significantly correlated with malignancy (P=0.005) with high discrimination (area under the curve 0.79; 95% confidence interval [CI], 0.63-0.96). An ADC≥1.28 μm2/ms predicted malignity with a pooled sensitivity of 83.3% (95% CI, 60.8%-94.2%), specificity of 66.7% (95% CI: 39.1%-86.2%), and accuracy of 76.7% (95% CI, 59.1%-88.2%). Neither T2*, T1, and T2 values, SUVmax, HU of the solid components, nor HU of the entire nodule correlated with malignancy (P>0.28). For adenocarcinomas, ADC inversely correlated with lepidic subtype (P<0.03) but not with acinar predominance (P>0.71, Table 4). Conversely, T2* correlated with the acinar (P=0.013) but not with the lepidic subtype (P=0.71). T1 and T2 did not correlate with histopathology subtypes. None of the MRI parameters significantly predicted tumor differentiation (P>0.11). For patients who had undergone PET imaging, we did not find significant correlation between SUVmax and any of the MRI parameters (P>0.56). T1 values showed significant correlation with HU measurements of the entire nodule (P<0.001) and with HU measurements of solid components (P=0.031), whereas ADC and T2* did not. T2 values showed significant correlation with HU of the entire nodule (P=0.024) and with HU of the solid components (P=0.008). The greatest diameter of the solid component correlated with lesion visibility on DWI and T2* images (P<0.04, Table 4). There was a significant correlation between percentage of solid component and lesion visibility on T1 (P=0.018).
Reader 1 demonstrated almost perfect intraobserver correlation for T1 values, substantial correlation for T2 and T2* values, and moderate correlation for ADC (Table 5). There was near-perfect intraobserver correlation for ADC, T1, and T2 values and moderate correlation for T2* evaluations for reader 2. We demonstrated almost perfect interobserver correlation for T1 and T2 values, substantial correlation for ADC, and fair correlation for T2* evaluations.
In summary, we have demonstrated that ADC significantly correlated with malignancy with high discrimination. ADC and T2* correlated with adenocarcinoma subtypes. No MRI parameters predicted tumor differentiation. SUVmax did not correlate with any MRI parameters. Visibility on T1-weighted images correlated with percentage of solid components. T1 and T2 values showed significant correlation with HU measurements of the entire nodule and with HU measurements of solid components.
Our results confirmed prior reports promoting MRI as a useful noninvasive means for imaging pulmonary nodules without the use of ionizing radiation.27–42 Although these studies used DWI with a 1.5 T system, our study represents one of the few studies utilizing a 3.0 T system. According to Ohba et al,39 both 1.5 and 3.0 T DWI modalities are equally effective for assessing pulmonary nodules. We also expected the higher field system provided by a 3.0 T system to generate a better signal to noise ratio and faster scan times compared with a 1.5 T system. The increased signal to noise ratio was helpful for the low-intensity PSNs, whereas the faster scanning time enabled easier breath-holding for the patients with pulmonary disease. Our study utilized a single-shot, twice-refocused spin echo sequence with bipolar diffusion gradients and centric-ordered TSE instead of EPI readouts to avoid susceptibility-induced distortions that worsen at higher applied fields and are especially pronounced in the lungs. According to Zhou et al,43 EPI is sensitive to main magnetic field heterogeneity, especially in the phase encoding direction, because of relatively low bandwidth. The use of spin echoes in the TSE sequence rather than gradient echoes in an EPI readout alleviates such an issue given the lessened phase encoding.
Although the numerical value of the ADC threshold demonstrated in this study is similar to previously reported thresholds,34–36,39,40,42 unlike most previous studies we found the mean ADC of malignant lesions to be higher than that of benign lesions. According to Liu et al,40 the mean ADC of malignant pulmonary lesions was significantly lower than that of benign pulmonary lesions in a cohort of solid nodules. This might be explained by the fact that our benign lesions tended to be very cellular, consisting of lymphoid hyperplasic, fibrotic, granulomatous, and other inflammatory nodules. In contrast, our malignant lesions consisted of PSNs that are not as cellular as the solid nodules discussed in prior studies. This is also supported by the fact that the mean HU of benign lesions was greater than that of malignant ones, indicating that the benign lesions were denser than the malignant ones, hence with increased barriers to water diffusion. In addition, most of our malignant lesions were high-grade adenocarcinomas, which might have necrotic components. Herneth et al44 reported that the ADC values of necrotic tumors are significantly higher than that of hypercellular neoplasia and benign lesions in mice. Similar to prior studies, our ADC threshold also demonstrated a high sensitivity of 88.9% with good accuracy (area under the curve 0.79), and the ADC correlated with adenocarcinoma histopathology subtype.41
To our knowledge this is the first study to examine the relationship between T2* and pulmonary nodule malignant potential. As expected, the mean T2* values and T2 signal intensities are higher for malignant lesions. This could be due to less susceptibility and increased water content in the malignant lesions, respectively. Our study showed that T2* correlated with acinar subtype of adenocarcinomas, which might be related to increased susceptibility and field inhomogeneity associated with this gland-forming adenocarcinoma subtype. As different adenocarcinoma subtypes have different prognostic significance levels,9 such a finding is clinically significant. We were unable to show that MRI parameters significantly predicted aggressiveness, but that could be because nearly all (15 of 16) malignant lesions were poorly differentiated.
To our knowledge, no study to date has associated T1 and T2 signal intensities and density measurements of PSNs. Given that nodule visibility is likely related to lesion cellularity, it is not surprising that the T1 and T2 signal intensities showed significant correlation with CT density of the entire PSN and CT density of solid components. Intuitively, one might think that solid component size might be related to lesion density; thus, visibility of these lesions on MRI sequences, such as on T1-weighted and T2-weighted images, might correlate with the solid component diameter; however, this was not the case in our study. This might simply indicate that nodule densities are more complex. However, nodule visibility, at least on T1-weighted images, correlated well with the percentage of solid components.
De Hoop et al18 reported that mass measurements allowed detection of nodule growth earlier than diameter, area, or volume measurements. HU was utilized in a formula to calculate mass. However, no study to date has directly correlated HU with malignancy, and we found that there is no direct association between HU and malignant potential. As expected, the greatest diameter of the solid component correlated with lesion visibility on DWI and T2* images.
Interestingly, the widely utilized SUVmax also did not correlate with malignancy. Although this could be due to the small sample size, inherent properties of PSNs differing from solid nodules is the more likely explanation, as Hattori et al45 demonstrated that PSNs had a mean SUVmax that was lower than the normally accepted SUVmax. None of the MRI parameters correlated with SUVmax, which is likely related to increased susceptibility effects, such as image artifacts and shortened T2* values, with a 3 T system, as discussed in the study by Ohba et al.39
Our study showed moderate to near-perfect intraobserver agreements for all MRI parameters and substantial to near-perfect interobserver agreement for ADC, T1, and T2 values only, which might be due to difficulty in visualizing these nodules on T2* images. The fact that T2* dephasing is more prominent at 3 T may have also contributed to technical difficulty in imaging these PSNs.
Our study had several limitations. Given that this is a pilot study, the small sample size limited our ability to demonstrate statistical significance for several relationships and precluded statistical adjustment for nesting of nodules within cases. As the sample size was considered too small to model the ROC using a multirater, multicase design, we elected to simplify the data by means of computing a composite score. Further study with a larger cohort size will eliminate these limitations. Second, respiratory and electrocardiography triggering were not utilized, which might have improved our image quality. However, we felt that the quality of breath-holding for the most important sequences was adequate, and the majority of our lesions were not adjacent to the heart or great vessels. Two of the 32 PSNs were not visualized on the DWI. One of these lesions was an elastic fibrous scar, and the other lesion was an aggregate of alveolar cells and macrophages, which likely lacked significant freedom for water diffusion to generate signals. Three of the 32 lesions were visualized on the T2* sequences by only 1 of the 2 readers. Two of the lesions were low-grade adenocarcinomas with lepidic growth, and the other nodule was an adenocarcinoma with significant pseudocavitation that might not have filled the air spaces with enough cells to reduce the field inhomogeneity enough for significant visibility on T2* imaging. Finally, we used b values for ADC measurements where the perfusion fraction was not excluded, and because only monoexponential fitting was performed the ADC values might have been slightly overestimated; however, the reproducibility should have remained the same.
In conclusion, 3 T MRI shows promise for PSN evaluation with ADC, providing a potential quantitative means for distinguishing benign from malignant lesions. ADC and T2* values can potentially be helpful for predicting adenocarcinoma subtypes. T1 and T2 values showed significant correlation with HU measurements of the entire nodule and of the solid components.
The authors acknowledge the following individuals for their contributions to this study: Christine U. Lee, MD, PhD, and David L. Levin, MD, PhD, for general oversight of the study; and Kimberly K. Amrami, MD, for administrative support.
1. Kaneko M, Eguchi K, Ohmatsu H, et al.. Peripheral lung cancer: screening and detection with low-dose spiral CT versus radiography. Radiology. 1996;201:798–802.
2. Henschke CI, Yankelevitz DF, Mirtcheva R, et al.. CT screening for lung cancer: frequency and significance of part-solid and nonsolid nodules. Am J Roentgenol. 2002;178:1053–1057.
3. Yang PS, Lee KS, Han J, et al.. Focal organizing pneumonia: CT and pathologic findings. J Korean Med Sci. 2001;16:573–578.
4. Kim HY, Shim YM, Lee KS, et al.. Persistent pulmonary nodular ground-glass opacity at thin-section CT: histopathologic comparisons. Radiology. 2007;245:267–275.
5. Austin JHM, Garg K, Aberle D, et al.. Radiologic implications of the 2011 classification of adenocarcinoma of the lung. Radiology. 2013;266:62–71.
6. Park CM, Goo JM, Lee HJ, et al.. Nodular ground-glass opacity at thin-section CT: histologic correlation and evaluation of change at follow-up. Radiographics. 2007;27:391–408.
7. Park CM, Goo JM, Lee HJ, et al.. Focal interstitial fibrosis manifesting as nodular groundglass opacity: thin-section CT findings. Eur Radiol. 2007;17:2325–2331.
8. Kim H, Park CM, Koh JM, et al.. Pulmonary sub solid nodules: what radiologists need to know about the imaging features and management strategy. Diagn Interv Radiol. 2014;20:47–57.
9. Travis WD, Brambilla E, Noguchi M, et al.. International association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society international multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol. 2011;6:244–285.
10. Goo JM, Park CM, Lee HJ. Ground-glass nodules on chest CT as imaging biomarkers in the management of lung adenocarcinoma. Am J Roentgenol. 2011;196:533–543.
11. Lee HJ, Goo JM, Lee CH, et al.. Predictive CT findings of malignancy in groundglass nodules on thin-section chest CT: the effects on radiologist performance. Eur Radiol. 2009;19:552–560.
12. Takahashi S, Tanaka N, Okimoto T, et al.. Long term follow-up for small pure ground-glass nodules: implications of determining an optimum follow-up period and high-resolution CT findings to predict the growth of nodules. Jpn J Radiol. 2012;30:206–217.
13. Lee SM, Park CM, Goo JM, et al.. Invasive pulmonary adenocarcinomas versus preinvasive lesions appearing as ground-glass nodules: differentiation by using CT features. Radiology. 2013;268:265–273.
14. Nomori H, Ohtsuka T, Naruke T, et al.. Differentiating between atypical adenomatous hyperplasia and broncholoalveolar carcinoma using the computed tomography number histogram. Ann Thorac Surg. 2003;76:867–871.
15. Kawakami S, Sone S, Takashima S, et al.. Atypical adenomatous hyperplasia of the lung: correlation between high-resolution CT findings and histopathologic features. Eur Radiol. 2001;11:811–814.
16. Gaeta M, Caruso R, Barone M, et al.. Ground-glass attenuation in nodular bronchioloalveolar carcinoma: CT patterns and prognostic value. J Comput Assist Tomogr. 1998;22:215–219.
17. Oh JY, Kwon SY, Yoon HI, et al.. Clinical significance of a solitary ground-glass opacity (GGO) lesion of the lung detected by chest CT. Lung Cancer. 2007;55:67–73.
18. De Hoop B, Gietema H, Van de Vorst S, et al.. Pulmonary ground-glass nodules: increase in mass as an early indicator of growth. Radiology. 2010;255:199–206.
19. Patsios D, Roberts HC, Paul NS, et al.. Pictorial review of the many faces of bronchioloalveolar cell carcinoma. Br J Radiol. 2007;80:1015–1023.
20. Lee SM, Park CM, Goo JM, et al.. Transient part-solid nodules
detected on screening thin-section CT for lung cancer: comparison with persistent part-solid nodules
. Radiology. 2010;255:242–251.
21. Glynn C, Zakowski MF, Ginsberg MS. Are there imaging characteristics associated with epidermal growth factor receptor and KRAS mutations in patients with adenocarcinoma of the lung with bronchioloalveolar features? J Thorac Oncol. 2010;5:344–346.
22. Kim TJ, Park CM, Goo JM, et al.. Is there a role for FDG PET in the management of lung cancer manifesting predominantly as ground-glass opacity? Am J Roentgenol. 2012;198:83–88.
23. Chun EJ, Lee HJ, Kang WJ, et al.. Differentiation between malignancy and inflammation in pulmonary ground-glass nodules: the feasibility of integrated 18
F-FDG PET/CT. Lung Cancer. 2009;65:180–186.
24. Kubo T, Lin PJP, Stiller W, et al.. Radiation dose reduction in chest CT: a review. Am J Roentgenol. 2008;190:335–343.
25. Fazel R, Krumholz HM, Wang Y, et al.. Exposure to low-dose ionizing radiation from medical imaging procedures. N Engl J Med. 2009;361:849–857.
26. Wrixon AD. New ICRP recommendations. J Radiol Prot. 2008;28:161–168.
27. Henzler T, Dietrich O, Krissak R, et al.. Half-Fourier-Acquisition Single-Shot Turbo Spin-Echo (HASTE) MRI of the lung at 3 Tesla using parallel imaging with 32-receiver channel technology. J Magn Reson Imaging. 2009;30:541–546.
28. Vogt FM, Herborn CU, Hunold P, et al.. HASTE MRI versus chest radiography in the detection of pulmonary nodules: comparison with MDCT. Am J Roentgenol. 2004;183:71–78.
29. Schafer JF, Vollmar J, Schick F, et al.. Detection of pulmonary nodules with breath-hold magnetic resonance imaging
in comparison with computed tomography. Rofo. 2005;177:41–49.
30. Schroeder T, Rheum SG, Debatin JF, et al.. Detection of pulmonary nodules using a 2D HASTE MR sequence: comparison with MDCT. Am J Roentgenol. 2005;185:979–984.
31. Biederer J, Schoene A, Freitag S, et al.. Simulated pulmonary nodules implanted in a dedicated porcine chest phantom: sensitivity of MR imaging for detection. Radiology. 2003;227:475–483.
32. Both M, Schultze J, Reuter M, et al.. Fast T1- and T2-weighted pulmonary MR-imaging in patients with bronchial carcinoma. Eur J Radiol. 2005;53:478–488.
33. Koyama H, Ohno Y, Aoyama N, et al.. Comparison of STIR turbo SE imaging and diffusion-weighted imaging of the lung: capability for detection and subtype classification of pulmonary adenocarcinomas. Eur Radiol. 2010;20:790–800.
34. Uto T, Takehara Y, Nakamura Y, et al.. Higher sensitivity and specificity for diffusion-weighted imaging of malignant lung lesions without apparent diffusion coefficient
quantification. Radiology. 2009;252:247–254.
35. Ohba Y, Nomori H, Mori T, et al.. Is diffusion-weighted magnetic resonance imaging
superior to positron emission tomography with fludeoxyglucose F 18 in imaging non-small cell lung cancer. J Thorac Cardiovasc Surg. 2009;138:439–445.
36. Mori T, Nomori H, Ikeda K, et al.. Diffusion-weighted magnetic resonance imaging
for diagnosing malignant pulmonary nodules/masses. Comparison with positron emission tomography. J Thorac Oncol. 2008;3:358–364.
37. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174.
38. Regier M, Schwarz D, Henes FO, et al.. Diffusion-weighted MR-imaging for the detection of pulmonary nodules at 1.5 Tesla: intraindividual comparison with multidetector computed tomography. J Med Imaging Radiat Oncol. 2011;55:266–274.
39. Ohba Y, Nomori H, Mori T, et al.. Diffusion-weighted magnetic resonance for pulmonary nodules: 1.5 vs. 3 Tesla. Asian Cardiovasc Thorac Ann. 2011;19:108–114.
40. Liu H, Liu Y, Yu T, et al.. Usefulness of diffusion-weighted MR imaging in the evaluation of pulmonary lesions. Eur Radiol. 2010;20:807–815.
41. Tanaka R, Horikoshi H, Nakazato Y, et al.. Magnetic resonance imaging
in peripheral lung adenocarcinoma: correlation with histopathologic features. J Thorac Imaging. 2009;24:4–9.
42. Satoh S, Kitazume Y, Ohdama S, et al.. Can malignant and benign pulmonary nodules be differentiated with diffusion-weighted MRI? Am J Roentgenol. 2008;191:464–470.
43. Zhou XJ, Du YP, Bernstein MA, et al.. Concomitant magnetic-field-induced artifacts in axial echo planar imaging. Magn Reson Med. 1997;39:596–605.
44. Herneth AM, Guccione S, Bednarski M. Apparent diffusion coefficient
: a quantitative parameter for in vivo tumor characterization. Eur J Radiol. 2003;45:208–213.
45. Hattori A, Suzuki K, Matsunaga T, et al.. Tumour standardized uptake value on positron emission tomography is a novel predictor of adenocarcinoma in situ for c-Stage IA lung cancer patients with a part-solid nodule on thin-section computed tomography scan. Interact Cardiovasc Thorac Surg. 2014;18:329–334.
Keywords:Copyright © 2016 Wolters Kluwer Health, Inc. All rights reserved
magnetic resonance imaging; part-solid nodules; diffusion-weighted magnetic resonance imaging; apparent diffusion coefficient