Abstract
Background
Pulmonary adenocarcinoma is the most common pathological type of malignant pulmonary nodules, of which, invasive adenocarcinoma (IAC) is associated with a risk of recurrence. Pulmonary tuberculoma can sometimes present as noncalcified, solid pulmonary nodule with imaging characteristics such as lobulation, spiculation, and pleural indentation, which is difficult to distinguish from IAC. Magnetic resonance imaging (MRI) as a non-ionizing modality can be complementary tool for nodules assessment. This study aimed to evaluate the potential of the combined conventional and modified anatomical MRI sequences for differential diagnosis of IAC and tuberculoma.
Methods
Sixty-seven patients with 82 noncalcified nodules underwent computed tomography (CT) and MRI (T1WI-starVIBE, T1WI-VIBE, T2WI-TSE-fBLADE). Two radiologists independently assessed nodule dimensions and morphologic features. The inter-method agreement of morphologic features assessment by CT and MRI sequences were compared using Kappa test. Multivariate logistic regression analyses were applied to identify independent predictors of IAC. Receiver operating characteristic (ROC) analysis was performed to investigate the differential diagnosis capability.
Results
Thirty-eight IACs and 44 tuberculomas were identified. Readers 1 and 2 underestimated the nodules mean diameter with T1WI-starVIBE (T1WI-VIBE, T2WI-TSE-fBLADE) by 0.86±1.71 mm (1.19±2.06, 0.15±1.96 mm) and 0.99±1.75 mm (1.27±2.04, 0.19±1.91 mm). The inter-method agreements between MRI and CT were “fair” to “excellent” in the evaluation of morphological features except for spiculation. Compared with the tuberculoma group, the IAC group was significant with unclear margin (T1WI-starVIBE, T1WI-VIBE), irregular morphology (CT, MRI), lobulation (CT, MRI), spiculation (T1WI-starVIBE, T2WI-TSE-fBLADE) and air bronchogram (CT, T1WI-starVIBE and T1WI-VIBE) (P<0.05). The area under the curve (AUC) values for the logistic model by the combination of CT and MRI were 0.867/0.877 (sensitivity 73.68%/76.32%, specificity 86.36%/86.36%) and were significantly higher than that by T1WI-starVIBE (P=0.002) and T1WI-TSE-fBLADE (P=0.03) (reader 1), as well as higher than that by CT (P=0.045) and T1WI-starVIBE (P=0.003) (reader 2).
Conclusions
The combined conventional and modified anatomical MRI sequences has diagnostic potential in distinguishing pulmonary IAC from tuberculoma.
Keywords: Invasive adenocarcinoma (IAC), tuberculoma, magnetic resonance imaging (MRI)
Highlight box.
Key findings
• We evaluated the potential of the combined conventional and modified anatomical magnetic resonance imaging (MRI) sequences for differential diagnosis of invasive adenocarcinoma (IAC) and tuberculoma. Compared with the tuberculoma group, the IAC group was significant with unclear margin (T1WI-starVIBE, T1WI-VIBE), irregular morphology [computed tomography (CT), MRI], lobulation (CT, MRI), spiculation (T1WI-starVIBE, T2WI-TSE-fBLADE) and air bronchogram (CT, T1WI-starVIBE and T1WI-VIBE) (P<0.05). The area under the curve values for the logistic model by the combination of CT and MRI were significantly higher than that by T1WI-starVIBE (P=0.002) and T1WI-TSE-fBLADE (P=0.03) (reader 1), as well as higher than that by CT (P=0.045) and T1WI-starVIBE (P=0.003) (reader 2).
What is known and what is new?
• It is known that pulmonary tuberculoma can sometimes present as noncalcified, solid pulmonary nodule with imaging characteristics such as lobulation, spiculation, and pleural indentation, which is difficult to distinguish from IAC. Most of the previous studies concerning distinguishing IAC from tuberculoma are related to positron emission tomography/CT (PET/CT) and contrast-enhanced dynamic CT with some limitations. MRI is a non-ionizing modality for lung imaging. With progress in MRI techniques, state-of-the-art thoracic MRI plays a complementary role in the management of patients with various chest diseases and especially in the detection and evaluation of pulmonary nodules or masses which potentially indicate to cancer.
• This study is novel in applying conventional and modified anatomical MRI sequences for differential diagnosis of IAC and tuberculoma.
What is the implication, and what should change now?
• The findings imply that the combined conventional and modified anatomical MRI sequences has diagnostic potential in distinguishing pulmonary IAC from tuberculoma.
Introduction
Cancer is a leading cause of death worldwide, accounting for nearly 10 million deaths in 2020 and lung cancer ranks as the most common cause of cancer death (1.8 million deaths) (1). Lung cancer burden can be reduced through early detection and appropriate treatment (1). Early-stage lung cancer is usually presented as pulmonary nodules, which are defined as the round or round-like opacity circumscribed by the pulmonary parenchyma and measured less than 3 cm in diameter (2). Pulmonary adenocarcinoma is the most common pathological type of malignant pulmonary nodules, of which, invasive adenocarcinoma (IAC) is associated with a risk of recurrence (3-5). Tuberculosis (TB) remains the top infectious killer worldwide, accounting for nearly 1.5 million deaths each year (6,7). TB most often affects the lung. Pulmonary tuberculoma sometimes presents as noncalcified, solid pulmonary nodule with imaging characteristics such as lobulation, spiculation, and pleural indentation, and is difficult to distinguish from IAC. Accurate differentiation of IAC from pulmonary tuberculoma is vital for clinicians to develop an appropriate management plan. Most of the previous studies on distinguishing IAC from tuberculoma are related to positron emission tomography/computed tomography (PET/CT) and contrast-enhanced dynamic CT with some limitations (3,8,9). CT remains a primary first-line imaging modality for characterizing pulmonary disease to date. A previous study using clinical and CT radiomics nomogram model to differentiate IAC from tuberculoma achieved a favorable prediction efficacy (10).
Magnetic resonance imaging (MRI) is a non-ionizing modality for lung imaging. Advancements in MRI techniques have enabled the visualization of pathological pulmonary lesions and thus can be used as a complementary diagnostic tool for pulmonary disease (11-14). Short-tau inversion recovery (STIR) T2-weighted turbo-spin echo (TSE) imaging has been reported as a better strategy for pulmonary nodules detection (14,15). T2-weighted TSE with periodically rotated overlapping parallel lines with enhanced reconstruction (PROPELLER), called fBLADE technique (Siemens, Erlangen, Germany), is insensitive to motion and magnetic susceptibility artifacts, which can improve image quality (16,17). Conventional T1-weighted three dimensions (3D) gradient-echo (GRE) volumetric interpolated breath-hold examination (T1WI-VIBE) shows acceptable sensitivity in detecting pulmonary lesions (14), although it is sensitive to motion. Some patients with compromised pulmonary function cannot hold their breath during images acquisition, which results in artifacts. Radial technique is based on acquiring k-space data along radial spokes. Because of its low sensitivity to motion, radial stack-of-stars acquisition allows a free-breathing T1W sequence (T1WI-starVIBE) that can depict pulmonary lesions, especially for patients who are unable to hold their breath (18,19). The T1WI-starVIBE sequence can also provide high-resolution images (16). Several previous studies showed that T1WI-VIBE and T1WI-starVIBE can detect pulmonary nodules/mass and depict lesions morphological features (18).
To our knowledge, the study concerning differentiating IAC from tuberculoma using the modified and conventional anatomical sequences, such as T1WI-starVIBE, T2WI-TSE-fBLADE and T1WI-VIBE, to depict morphological features of nodules is scarce. In this study, we aimed to explore the differential diagnosis capability of these MRI sequences, as a complementary tool to CT. We present this article in accordance with the STARD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1677/rc).
Methods
Patients
The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This retrospective study was approved by the ethics committee of Shanghai Public Health Clinical Center (No. 2019-S021-02). The necessity for written informed consent was waived, as data were analyzed retrospectively and anonymously. The medical and imaging records of patients who underwent thoracic MRI examinations between June 2019 and June 2020 were reviewed. Inclusion criteria included: (I) the pulmonary solid nodules were clearly depicted by CT scan and MRI, including T1WI-starVIBE, T1WI-VIBE and T2WI-TSE-fBLADE; (II) the diameter of pulmonary nodules was smaller than 3 cm; and (III) the pulmonary nodules were pathologically confirmed as IAC or tuberculoma. Other pulmonary nodules resulting from squamous cells carcinoma, metastatic tumors, hamartoma, etc., and nodules with ground glass opacity component or calcification were excluded (Figure 1).
Figure 1.
The workflow of the study. GGO, ground-glass opacity; MRI, magnetic resonance imaging.
Sixty-seven patients including 38 males and 29 females were finally enrolled in this study, their median age was 54 years (18–82 years). Eighty-two pulmonary nodules were detected from those 67 patients, including 38 IACs and 44 tuberculomas. Operative or biopsy samples from 42 or 25 patients were collected for pathologic diagnosis. The interval period between thoracic MRI and CT scan (median 2 days) and between thoracic MRI and operations/biopsies (median 3 days) were 2–7 days.
CT imaging protocol
The patients were laid on a 320-detector CT (Aquilion Vision, Canon Medical Systems, Kawasaki, Japan) in supine position with full inspiration. The CT parameters included: pitch, 0.813; tube voltage, 120 kV; automatic tube current with SD10 (Sure Exp 3D set, maximum: 440 mA, minimum: 60 mA); The radiation dose of the thoracic CT was calculated in terms of the dose-length product (DLP). An effective radiation dose was calculated by a formula DLP× conversion factor (0.017 mSv‧mGy−1cm−1 for the chest). Average radiation dose was 5.22 mSv. All CT images with the slice thickness/interval of 1 mm/1 mm were reconstructed by means of adaptive iterative dose-reduction with three-dimensional processing (AIDR 3D, standard) and a high frequency reconstruction algorithm (FC56) for the lung window setting. The lung window width and level were adjusted appropriately by the reference standards of 1,200 and −600 Hounsfield unit (HU).
MRI protocol
All patients underwent 3T MRI (MAGNETOM Skyra, Siemens Healthcare, Erlangen, Germany) with a 32-channel body-phased array coil. The patients assumed in supine position with arms raising above their body. The lesions were positioned in the positioning map of horizontal, sagittal and coronal position. The MRI sequences included: (I) breath-hold T1WI-VIBE [spectral attenuated inversion recovery (SPAIR)], repetition time (TR): 3.67 ms, echo time (TE): 1.78 ms, matrix: 320×320, field of view (FOV): 380×380, slice thickness: 3.0 mm; (II) free-breathing T1WI-star VIBE (SPAIR), TR: 2.79 ms, TE: 1.39 ms, matrix: 320×320, FOV: 380×380, slice thickness: 3.0 mm; (III) respiratory triggered T2WI-TSE-fBLADE (STIR), TR: 3740.0 ms, TE: 69.0 ms, inversion time (TI): 240 ms, matrix: 320×320, FOV: 380×380, slice thickness: 3.0 mm. The total scan time was nearly 5 minutes due to the patients’ respiratory state.
Image analysis
CT images were evaluated in clinical routine by a radiologist with twenty years of experience, who was not informed about the MRI findings and pathological diagnoses. The location and size of each nodule was recorded. Nodule size was defined as the mean value of the longest and orthogonal shortest axial diameters. The nodules morphological features included margin (clear/unclear), morphology (regular/irregular), lobulation, spiculation, cavity, air-bronchogram and pleural indentation. The CT was considered as gold standard for recording the location of detected nodules.
The MRI data analysis was individually performed by two radiologists with 8 years (reader 1) and 22 years (reader 2) of expertise in chest imaging diagnosis, respectively, who were unaware of the CT findings and pathological diagnoses. Each MRI sequences were evaluated independently by both readers in a blinded fashion in the interval time of 4 weeks. All images were reviewed and analyzed using RadiAnt DICOM Viewer 2021.2.2 (Medixant, Poznan, Poland, https://www.radiantviewer.com/) after removing the name of sequences. The recorded contents for each nodule included the location, size, and morphological features by the 3 MRI sequences.
For assessment of nodules morphological features depicted by MRI, a 4-point visual scoring system was based on the following parameters: the clarity of margin, morphology, lobulation, spiculation, cavity, air bronchogram and pleural indentation (0, definitely unclear/irregular/absent; 1, probably unclear/irregular/absent; 2, probably clear/regular/present; 3, definitely clear/regular/present). For each nodule, the capability of nodules morphological features depiction by those 3 MRI sequences were compared with those of CT. Finally, the confirmation rate defined as the ratio of confirmation score (“0” and “3”) was used to evaluate the capability of nodules morphological features depiction. Score “0” and “1” mean the negative result of the related feature, score “2” and “3” the positive result.
Statistical analysis
Statistical analysis was conducted using IBM SPSS 26 (IBM Corporation, Armonk, New York, USA, https://www.ibm.com/analytics/spss-statistics-software) and MedCalc 18.11.3 (MedCalc Software Bvba, Ostend, Belgium, https://www.medcalc.org/). All continuous variables were presented as “mean ± standard deviation”, categorical variables were shown as percentage. The intraclass correlation coefficient (ICC) was used to assess the consistence of nodules diameter measurements between CT and MRI sequences (T1WI-VIBE, T1WI-starVIBE and T2WI-TSE-fBLADE), as well as the inter-reader agreement rates. As the ICC approached 1, the diameter measurement consistency between the tested methods was better (excellent >0.75, mild 0.40< ICC ≤0.74, poor ≤0.40). Comparison of categorical variables was performed using Chi-squared test. The inter-method agreement of morphologic features assessment by CT and MRI sequences, as well as inter-reader agreement rates, were compared using Kappa test with the following levels of agreement (20): poor 0–0.20, fair 0.21–0.40, moderate 0.41–0.60, substantial 0.61–0.80, and excellent 0.81–1.00. Multivariate logistic regression was also performed to compare the IAC group with the tuberculoma group. Receiver operating characteristic (ROC) analyses were conducted for the logistic regression model. The area under the curve (AUC) values were used to evaluate the diagnostic efficacy by CT and MRI when comparing the IAC group with the tuberculoma group. Delong test was used to compare AUC values. P values of less than 0.05 were considered statistically significant.
Results
Patients and nodules features
Patient demographic data and nodule features were illustrated in Table 1. All three MRI sequences (T1WI-starVIBE, T1WI-VIBE and T2WI-TSE-fBLADE) depicted nodules location accurately by both readers. The mean diameter of all nodules measured by CT was 19.56±6.85 mm (range, 7–30 mm) and the mean diameter measured by T1WI-starVIBE was 18.69±7.03 mm (range, 6.2–30 mm)/18.57±6.88 mm (range, 6–30 mm) (reader 1/2, ICC: 0.9698/0.9676, 95% CI: 0.9536–0.9804/0.9502–0.9790), indicating excellent inter-reader agreement (ICC: 0.9961, 95% CI: 0.9939–0.9975). T1WI-starVIBE underestimated the mean diameter by 0.86±1.71 mm/0.99±1.75 mm for reader 1/2. The mean diameter measured by T1WI-VIBE was 18.37±7 mm (range, 5–30 mm)/18.29±7.08 mm (range, 5–30 mm) (reader 1/2, ICC: 0.9560/0.9571, 95% CI: 0.9326–0.9714/0.9342–0.9721), indicating excellent inter-reader agreement (ICC: 0.997, 95% CI: 0.9954–0.9981). T1WI-VIBE underestimated the mean diameter by 1.19±2.06 mm/1.27±2.04 mm for reader 1/2. The mean diameter measured by T2WI-TSE-fBLADE was 19.4±6.95 mm (range, 6–30 mm)/19.37±6.9 mm (range, 6–30 mm) (reader 1/2, ICC: 0.9598/0.9614, 95% CI: 0.9384–0.9739/0.9407–0.9749), indicating excellent inter-reader agreement (ICC: 0.9942, 95% CI: 0.9911–0.9963). T2WI-TSE-fBLADE underestimated the mean diameter by 0.15±1.96 mm/0.19±1.91 mm for reader 1/2.
Table 1. Patients and nodules features.
| Variable | Value |
|---|---|
| Patient (n=67) | |
| Age (years), median [range] | 54 [13–82] |
| Gender (male/female) | 38/29 |
| Interval time of CT and MRI (days), median [range] | 2 [0–12] |
| Number of nodules per patient | |
| 1 | 54 |
| 2 | 11 |
| 3 | 2 |
| Nodules (n=82) | |
| Diameter (mm), mean ± SD [range] | 19.56±6.85 [7–30] |
| Number of nodules per size category | |
| 6 to <8 mm | 3 |
| ≥8 to <10 mm | 2 |
| ≥10 to <20 mm | 37 |
| ≥20 mm | 40 |
| Number of nodules per location, n (%) | |
| Right upper lobe | 33 (40.24) |
| Right middle lobe | 4 (4.88) |
| Right lower lobe | 16 (19.51) |
| Left upper lobe | 15 (18.29) |
| Left lower lobe | 14 (17.07) |
| Number of nodules per surgical pathology, n (%) | |
| Tuberculoma | 44 (53.66) |
| IAC | 38 (46.34) |
CT, computed tomography; IAC, invasive adenocarcinoma; MRI, magnetic resonance imaging; SD, standard deviation.
Morphological features
The nodules morphological features depicted by CT and MRI sequences were showed in Table 2. The confirmation rate of morphological features depicted by CT was 95.12–100%, 62.2–96.34%/59.76–96.34% (reader 1/2) by T1WI-starVIBE, 48.78–96.34%/43.9–96.34% (reader 1/2) by T1WI-VIBE, 53.66–89.02%/52.44–90.24% (reader 1/2) by T2WI-TSE-fBLADE. All inter-reader agreements achieved “excellent”.
Table 2. The confirmation rate and inter-reader agreement for assessment of morphological features obtained with MRI sequences (n=82).
| Imaging features | Methods | Observers | Visual score | Confirmation rate (%) | Kappa value | 95% CI | |||
|---|---|---|---|---|---|---|---|---|---|
| 0 | 1 | 2 | 3 | ||||||
| Margin | CT | 24 | 0 | 1 | 57 | 98.78 | |||
| T1WI-starVIBE | Reader 1 | 22 | 5 | 11 | 44 | 80.49 | 0.96986 | 0.93672–1 | |
| Reader 2 | 21 | 4 | 16 | 41 | 75.61 | ||||
| T1WI-VIBE | Reader 1 | 25 | 10 | 10 | 37 | 75.61 | 0.94862 | 0.88297–1 | |
| Reader 2 | 21 | 13 | 10 | 38 | 71.95 | ||||
| T2WI-TSE-fBLADE | Reader 1 | 21 | 4 | 10 | 47 | 82.93 | 0.99625 | 0.98897–1 | |
| Reader 2 | 20 | 5 | 10 | 47 | 81.71 | ||||
| Morphology | CT | 36 | 0 | 0 | 46 | 100 | |||
| T1WI-starVIBE | Reader 1 | 32 | 3 | 6 | 41 | 89.02 | 0.99068 | 0.98014–1 | |
| Reader 2 | 31 | 4 | 8 | 39 | 85.37 | ||||
| T1WI-VIBE | Reader 1 | 41 | 3 | 7 | 31 | 87.8 | 0.99062 | 0.97998–1 | |
| Reader 2 | 38 | 6 | 7 | 31 | 84.15 | ||||
| T2WI-TSE-fBLADE | Reader 1 | 38 | 5 | 4 | 35 | 89.02 | 0.97903 | 0.95343–1 | |
| Reader 2 | 40 | 5 | 3 | 34 | 90.24 | ||||
| Lobulation | CT | 31 | 0 | 1 | 50 | 98.78 | |||
| T1WI-starVIBE | Reader 1 | 39 | 3 | 19 | 21 | 73.17 | 0.94871 | 0.8785–1 | |
| Reader 2 | 37 | 2 | 21 | 22 | 71.95 | ||||
| T1WI-VIBE | Reader 1 | 33 | 6 | 29 | 14 | 57.32 | 0.9631 | 0.92308–1 | |
| Reader 2 | 31 | 7 | 31 | 13 | 53.66 | ||||
| T2WI-TSE-fBLADE | Reader 1 | 33 | 2 | 16 | 31 | 78.05 | 0.96578 | 0.92685–1 | |
| Reader 2 | 31 | 2 | 18 | 31 | 75.61 | ||||
| Spiculation | CT | 26 | 0 | 4 | 52 | 95.12 | |||
| T1WI-starVIBE | Reader 1 | 59 | 10 | 10 | 3 | 75.61 | 0.91739 | 0.838–0.99679 | |
| Reader 2 | 56 | 13 | 12 | 1 | 69.51 | ||||
| T1WI-VIBE | Reader 1 | 49 | 15 | 16 | 2 | 62.2 | 0.9838 | 0.9611–1 | |
| Reader 2 | 48 | 17 | 15 | 2 | 60.98 | ||||
| T2WI-TSE-fBLADE | Reader 1 | 40 | 6 | 32 | 4 | 53.66 | 0.92045 | 0.84339–0.99752 | |
| Reader 2 | 39 | 6 | 33 | 4 | 52.44 | ||||
| Cavity | CT | 71 | 0 | 0 | 11 | 100 | |||
| T1WI-starVIBE | Reader 1 | 69 | 1 | 2 | 10 | 96.34 | 0.94434 | 0.83634–1 | |
| Reader 2 | 70 | 1 | 2 | 9 | 96.34 | ||||
| T1WI-VIBE | Reader 1 | 71 | 0 | 3 | 8 | 96.34 | 1 | 1–1 | |
| Reader 2 | 71 | 0 | 3 | 8 | 96.34 | ||||
| T2WI-TSE-fBLADE | Reader 1 | 66 | 0 | 10 | 6 | 87.8 | 1 | 1–1 | |
| Reader 2 | 66 | 0 | 10 | 6 | 87.8 | ||||
| Air bronchogram | CT | 55 | 0 | 0 | 27 | 100 | |||
| T1WI-starVIBE | Reader 1 | 55 | 2 | 20 | 5 | 73.17 | 0.96511 | 0.91703–1 | |
| Reader 2 | 53 | 3 | 22 | 4 | 69.51 | ||||
| T1WI-VIBE | Reader 1 | 59 | 1 | 16 | 6 | 79.27 | 0.93672 | 0.87147–1 | |
| Reader 2 | 57 | 1 | 19 | 5 | 75.61 | ||||
| T2WI-TSE-fBLADE | Reader 1 | 43 | 1 | 22 | 16 | 71.95 | 0.98428 | 0.95347–1 | |
| Reader 2 | 42 | 1 | 23 | 16 | 70.73 | ||||
| Pleural indentation | CT | 22 | 1 | 2 | 57 | 96.34 | |||
| T1WI-starVIBE | Reader 1 | 31 | 0 | 31 | 20 | 62.2 | 0.95437 | 0.90646–1 | |
| Reader 2 | 29 | 1 | 32 | 20 | 59.76 | ||||
| T1WI-VIBE | Reader 1 | 27 | 3 | 39 | 13 | 48.78 | 0.93264 | 0.87152–0.99376 | |
| Reader 2 | 24 | 5 | 41 | 12 | 43.9 | ||||
| T2WI-TSE-fBLADE | Reader 1 | 23 | 2 | 16 | 41 | 78.05 | 0.92005 | 0.84087–0.99923 | |
| Reader 2 | 21 | 2 | 17 | 42 | 76.83 | ||||
CI, confidence interval; CT, computed tomography; MRI, magnetic resonance imaging; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging; TSE, turbo-spin echo; VIBE, volume interpolated breath-hold examination.
Table 3 shows the results of inter-method and inter-reader agreements for the assessment of morphological features depicted by CT and MRI sequences. Inter-method agreements for the CT and MRI were considered as “moderate” to “excellent” for evaluating the following parameters: margin (T1WI-starVIBE, reader 1/2: kappa =0.46/0.447; T2WI-TSE-fBLADE, reader 1&2: kappa =0.447), morphology (T1WI-starVIBE, reader 1&2: kappa =0.776; T1WI-VIBE, reader 1&2: kappa =0.662; T2WI-TSE-fBLADE, reader 1/2: kappa =0.636/0.638), lobulation (T1WI-starVIBE, reader 1/2: kappa =0.539/0.605; T1WI-VIBE, reader 1/2: kappa =0.654/0.677; T2WI-TSE-fBLADE, reader 1/2: kappa =0.646/0.59), spiculation (T2WI-TSE-fBLADE, reader 2: kappa =0.411), cavity (T1WI-starVIBE, reader 1/2: kappa =0.848/0.895; T1WI-VIBE, reader 1&2: kappa =0.79; T2WI-TSE-fBLADE, reader 1&2: kappa =0.604), air bronchogram (T1WI-starVIBE, reader 1/2: kappa =0.437/0.415; T1WI-VIBE, reader 1/2: kappa =0.449/0.517) and pleural indentation (T1WI-starVIBE, reader 1/2: kappa =0.727/0.696; T1WI-VIBE, reader 1/2: kappa =0.751/0.72; T2WI-TSE-fBLADE, reader 1/2: kappa =0.765/0.758), while considered as “poor” for evaluating spiculation (T1WI-starVIBE, reader 1&2: kappa =0.161; T1WI-VIBE, reader 1/2: kappa =0.191/0.176). All inter-reader agreements achieved “excellent”.
Table 3. Nodules morphologic features assessed by CT and MRI sequences (n=82).
| Imaging features (%) | CT | T1WI-starVIBE | T1WI-VIBE | T2WI-TSE-fBLADE | |||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| TB | IAC | All | P | TB | IAC | All | P | TB | IAC | All | P | TB | IAC | All | P | ||||
| Margin (unclear) | |||||||||||||||||||
| Reader 1 | 29.55 | 28.95 | 29.27 | 0.95 | 22.73 | 44.74 | 32.93 | 0.03 | 31.82 | 55.26 | 42.68 | 0.03 | 29.55 | 31.58 | 30.49 | 0.84 | |||
| Kappa reader 1 (CT & MRI) | 0.59 | 0.339 | 0.460 | 0.519 | 0.194 | 0.351 | 0.454 | 0.439 | 0.447 | ||||||||||
| Reader 2 | 22.73 | 39.47 | 30.49 | 0.1 | 29.55 | 55.26 | 41.46 | 0.02 | 29.55 | 31.58 | 30.49 | 0.84 | |||||||
| Kappa reader 2 (CT & MRI) | 0.59 | 0.307 | 0.447 | 0.454 | 0.194 | 0.318 | 0.454 | 0.439 | 0.447 | ||||||||||
| Kappa reader 1 & 2 | 1 | 0.892 | 0.944 | 0.947 | 1 | 0.975 | 1 | 1 | 1 | ||||||||||
| Morphology (irregular) | |||||||||||||||||||
| Reader 1 | 27.27 | 63.16 | 43.9 | 0.001 | 29.55 | 57.89 | 42.68 | 0.01 | 40.91 | 68.42 | 53.66 | 0.01 | 38.64 | 68.42 | 52.44 | 0.007 | |||
| Kappa reader 1 (CT & MRI) | 0.721 | 0.78 | 0.776 | 0.703 | 0.534 | 0.662 | 0.65 | 0.534 | 0.636 | ||||||||||
| Reader 2 | 29.55 | 57.89 | 42.68 | 0.01 | 40.91 | 68.42 | 53.66 | 0.01 | 38.64 | 73.68 | 54.88 | <0.001 | |||||||
| Kappa reader 2 (CT & MRI) | 0.721 | 0.78 | 0.776 | 0.703 | 0.534 | 0.662 | 0.645 | 0.519 | 0.638 | ||||||||||
| Kappa reader 1 & 2 | 1 | 1 | 1 | 1 | 1 | 1 | 1 | 0.872 | 0.951 | ||||||||||
| Lobulation | |||||||||||||||||||
| Reader 1 | 38.64 | 89.47 | 62.2 | <0.001 | 29.55 | 71.05 | 48.78 | <0.001 | 31.82 | 76.32 | 52.44 | <0.001 | 36.36 | 81.58 | 57.32 | <0.001 | |||
| Kappa reader 1 (CT & MRI) | 0.499 | 0.29 | 0.539 | 0.653 | 0.37 | 0.654 | 0.661 | 0.265 | 0.646 | ||||||||||
| Reader 2 | 34.09 | 73.68 | 52.44 | <0.001 | 31.82 | 78.95 | 53.66 | <0.001 | 38.64 | 84.21 | 59.76 | <0.001 | |||||||
| Kappa reader 2 (CT & MRI) | 0.608 | 0.327 | 0.605 | 0.653 | 0.418 | 0.677 | 0.617 | 0.084 | 0.59 | ||||||||||
| Kappa reader 1 & 2 | 0.895 | 0.934 | 0.927 | 1 | 0.924 | 0.976 | 0.952 | 0.907 | 0.95 | ||||||||||
| Spiculation | |||||||||||||||||||
| Reader 1 | 59.09 | 78.95 | 68.29 | 0.054 | 6.82 | 26.32 | 15.85 | 0.04 | 18.18 | 26.32 | 21.95 | 0.38 | 29.55 | 60.53 | 43.9 | 0.005 | |||
| Kappa reader 1 (CT & MRI) | 0.096 | 0.174 | 0.161 | 0.185 | 0.174 | 0.191 | 0.365 | 0.341 | 0.393 | ||||||||||
| Reader 2 | 4.55 | 28.95 | 15.85 | 0.003 | 15.91 | 26.32 | 20.73 | 0.25 | 27.27 | 65.79 | 45.12 | <0.001 | |||||||
| Kappa reader 2 (CT & MRI) | 0.064 | 0.196 | 0.161 | 0.151 | 0.174 | 0.176 | 0.328 | 0.42 | 0.411 | ||||||||||
| Kappa reader 1 & 2 | 0.788 | 0.934 | 0.909 | 0.92 | 1 | 0.964 | 0.944 | 0.887 | 0.926 | ||||||||||
| Cavity | |||||||||||||||||||
| Reader 1 | 13.64 | 13.16 | 13.41 | 0.95 | 18.18 | 10.53 | 14.63 | 0.51 | 15.91 | 10.53 | 13.41 | 0.70 | 22.73 | 15.79 | 19.51 | 0.43 | |||
| Kappa reader 1 (CT & MRI) | 0.831 | 0.874 | 0.848 | 0.73 | 0.874 | 0.79 | 0.548 | 0.682 | 0.604 | ||||||||||
| Reader 2 | 15.91 | 10.53 | 13.41 | 0.70 | 15.91 | 10.53 | 13.41 | 0.70 | 22.73 | 15.79 | 19.51 | 0.43 | |||||||
| Kappa reader 2 (CT & MRI) | 0.91 | 0.874 | 0.895 | 0.73 | 0.874 | 0.79 | 0.548 | 0.682 | 0.604 | ||||||||||
| Kappa reader 1 & 2 | 0.92 | 1 | 0.949 | 1 | 1 | 1 | 1 | 1 | 1 | ||||||||||
| Air bronchogram | |||||||||||||||||||
| Reader 1 | 20.45 | 47.37 | 32.93 | 0.01 | 18.18 | 44.74 | 30.49 | 0.009 | 11.36 | 44.74 | 26.83 | 0.001 | 38.64 | 55.26 | 46.34 | 0.13 | |||
| Kappa reader 1 (CT & MRI) | 0.49 | 0.312 | 0.437 | 0.498 | 0.312 | 0.449 | 0.475 | 0.11 | 0.325 | ||||||||||
| Reader 2 | 20.45 | 44.74 | 31.71 | 0.02 | 11.36 | 50 | 29.27 | <0.001 | 40.91 | 55.26 | 47.56 | 0.19 | |||||||
| Kappa reader 2 (CT & MRI) | 0.441 | 0.312 | 0.415 | 0.498 | 0.421 | 0.517 | 0.44 | 0.11 | 0.306 | ||||||||||
| Kappa reader 1 & 2 | 0.927 | 1 | 0.972 | 1 | 0.895 | 0.94 | 0.953 | 1 | 0.976 | ||||||||||
| Pleural indentation | |||||||||||||||||||
| Reader 1 | 72.73 | 71.05 | 71.95 | 0.87 | 65.91 | 57.89 | 62.2 | 0.46 | 61.36 | 65.79 | 63.41 | 0.68 | 68.18 | 71.05 | 69.51 | 0.78 | |||
| Kappa reader 1 (CT & MRI) | 0.841 | 0.605 | 0.727 | 0.747 | 0.757 | 0.751 | 0.782 | 0.744 | 0.765 | ||||||||||
| Reader 2 | 63.64 | 63.16 | 63.41 | 0.96 | 61.36 | 68.42 | 64.63 | 0.51 | 72.73 | 71.05 | 71.95 | 0.87 | |||||||
| Kappa reader 2 (CT & MRI) | 0.792 | 0.586 | 0.696 | 0.645 | 0.813 | 0.72 | 0.771 | 0.744 | 0.758 | ||||||||||
| Kappa reader 1 & 2 | 0.95 | 0.89 | 0.922 | 0.904 | 0.94 | 0.921 | 0.782 | 1 | 0.882 | ||||||||||
CT, computed tomography; IAC, invasive adenocarcinoma; MRI, magnetic resonance imaging; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging; TB, tuberculoma; TSE, turbo-spin echo; VIBE, volume interpolated breath-hold examination.
Diagnostic performance of morphological features
The morphological features, including margin (T1WI-starVIBE, T1WI-VIBE), morphology (CT, MRI), lobulation (CT, MRI), spiculation (T1WI-starVIBE, T2WI-TSE-fBLADE) and air bronchogram (CT, T1WI-starVIBE and T1WI-VIBE), were found to have statistically significant difference between IAC and tuberculoma group (P<0.05) (Table 3; Figures 2,3, Figure S1).
Figure 2.
A 64-year-old female with IAC. (A) CT showed a SN (measured 15.4 mm) located in the right upper lobe, with morphological features including clear margin, irregular morphology, lobulation, spiculation (typically depicted in the other slices), air bronchogram and pleural indentation; (B) T1WI-starVIBE showed the nodule (14.3 mm) located in the right upper lobe, with morphological features including clear margin, irregular morphology, lobulation, air bronchogram (probably present) and pleural indentation (probably present in other slices); (C) T1WI-VIBE showed the nodule (13.4 mm) located in the right upper lobe, with morphological features including clear margin, irregular morphology, lobulation, air bronchogram (probably present) and pleural indentation (in other slices); (D) T2WI-TSE-fBLADE showed the nodule (14.6 mm) located in the right upper lobe, with morphological features including clear margin, irregular morphology, lobulation, and pleural indentation. CT, computed tomography; IAC, invasive adenocarcinoma; SN, solid nodule; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging; TSE, turbo-spin echo; VIBE, volume interpolated breath-hold examination.
Figure 3.
A 50-year-old female with tuberculoma. (A) CT showed a SN (measured 18.2 mm) located in the left lower lobe, with morphological features including clear margin, regular morphology, and pleural indentation; (B) T1WI-starVIBE showed the nodule (16.6 mm) located in the left lower lobe, with morphological features including clear margin, regular morphology, and pleural indentation probably present; (C) T1WI-VIBE showed the nodule (15.7 mm) located in the left lower lobe, with morphological features including clear margin, regular morphology, and pleural indentation probably present; (D) T2WI-TSE-fBLADE showed the nodule (16.7 mm) located in the left lower lobe, with morphological features including clear margin, regular morphology, and pleural indentation. CT, computed tomography; SN, solid nodule; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging; TSE, turbo-spin echo; VIBE, volume interpolated breath-hold examination.
Multivariate logistic regression analysis was used to establish a precision model containing the most informative features associated with the risk of IAC (Table 4). For the combined 3 MRI sequences, the features that were found to be most significantly related to IAC were morphology [irregular, T2WI-TSE-fBLADE: reader 1/2, odds ratio (OR) 0.314/0.241, P=0.041/0.02], lobulation (present, T1WI-VIBE: reader 1/2, OR 8.12/9.036, P<0.001), and air bronchogram (present, T1WI-VIBE: reader 1/2, OR 5.367/4.927, P=0.01/0.01). For the combination of CT and MRI sequences, the features that were found to be most significantly related to IAC were morphology (irregular, T2WI-TSE-fBLADE: reader 1/2, OR 0.252/0.221, P=0.02/0.02), lobulation (present, CT: reader 1/2, OR 17.91/17.151, P<0.001), and air bronchogram (present, T1WI-VIBE: reader 1/2, OR 5.055/5.367, P=0.02/0.01).
Table 4. Ability of logistic regression model to differentiate IAC from tuberculoma (n=82).
| Variable | Characteristics | Reader | OR | 95% CI | P value |
|---|---|---|---|---|---|
| CT | Lobulation | Reader | 14.44 | 4.072–51.21 | <0.001 |
| Air bronchogram | Reader | 3.943 | 1.214–12.801 | 0.02 | |
| T1WI-starVIBE | Lobulation | Reader 1 | 5.853 | 2.254–15.202 | <0.001 |
| Reader 2 | 4.317 | 1.606–11.602 | 0.004 | ||
| Spiculation | Reader 2 | 5.719 | 1.103–29.656 | 0.04 | |
| T1WI-VIBE | Lobulation | Reader 1 | 7.445 | 2.53–21.907 | <0.001 |
| Reader 2 | 7.572 | 2.529–22.669 | <0.001 | ||
| Air bronchogram | Reader 1 | 6.97 | 1.957–24.815 | 0.003 | |
| Reader 2 | 7.249 | 2.073–25.35 | 0.002 | ||
| T2WI-TSE-fBLADE | Morphology | Reader 1 | 0.35 | 0.128–0.959 | 0.041 |
| Reader 2 | 0.318 | 0.108–0.931 | 0.04 | ||
| Lobulation | Reader 1 | 6.942 | 2.427–19.855 | <0.001 | |
| Reader 2 | 6.037 | 1.93–18.883 | 0.002 | ||
| Spiculation | Reader 2 | 3.44 | 1.191–9.939 | 0.02 | |
| MRI | T1WI-VIBE: lobulation | Reader 1 | 8.12 | 2.62–25.167 | <0.001 |
| Reader 2 | 9.036 | 2.755–29.636 | <0.001 | ||
| T1WI-VIBE: air bronchogram | Reader 1 | 5.367 | 1.495–19.262 | 0.01 | |
| Reader 2 | 4.927 | 1.391–17.458 | 0.01 | ||
| T2WI-fBLADE: morphology | Reader 1 | 0.314 | 0.103–0.955 | 0.041 | |
| Reader 2 | 0.241 | 0.074–0.791 | 0.02 | ||
| CT + MRI | T1WI-VIBE: air bronchogram | Reader 1 | 5.055 | 1.319–19.368 | 0.02 |
| Reader 2 | 5.367 | 1.399–20.594 | 0.01 | ||
| T2WI-fBLADE: morphology | Reader 1 | 0.252 | 0.077–0.827 | 0.02 | |
| Reader 2 | 0.221 | 0.065–0.753 | 0.02 | ||
| CT: lobulation | Reader 1 | 17.91 | 4.342–73.879 | <0.001 | |
| Reader 2 | 17.151 | 4.107–71.616 | <0.001 |
CI, confidence interval; CT, computed tomography; IAC, invasive adenocarcinoma; MRI, magnetic resonance imaging; OR, odds ratio; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging; VIBE, volume interpolated breath-hold examination.
The AUC values were shown in Table 5 and Figure S2. The AUC value of the morphological features by CT for differential diagnosis was 0.808 (sensitivity 89.47%, specificity 61.36%, P<0.001). The AUC values of combined MRI sequences and the combination of CT with MRI were 0.839/0.857 (sensitivity 89.47%/76.32%, specificity 61.36%/79.55%, P<0.001; reader 1/2) and 0.867/0.877 (sensitivity 73.68%/76.32%, specificity 86.36%/86.36%, P<0.001; reader 1/2). The combination of CT with MRI had significantly better prediction performance than T1WI-starVIBE (reader1/2: P=0.002/0.003), T2WI-TSE-fBLADE (reader 1: P=0.03) and CT (reader 2: P=0.045), while also better than combined MRI sequences (reader 1&2), CT (reader 1), T1WI-VIBE (reader 1&2) and T2WI-TSE-fBLADE (reader 2) (P>0.05). The combined MRI sequences had significantly better prediction performance than T1WI-starVIBE (reader 1/2: P=0.007/0.006) and T1WI-VIBE (reader 2: P=0.04), while also better than CT (reader 1&2), T1WI-VIBE (reader 1) and T2WI-TSE-fBLADE (reader 1&2) (P>0.05).
Table 5. Comparison of ROC curves for logistic regression model by CT, MRI and CT-MRI in differential diagnosis of IAC and tuberculoma (n=82).
| Methods | Reader | AUC | 95% CI | Sensitivity (%) | Specificity (%) | P value |
|---|---|---|---|---|---|---|
| CT | – | 0.808 | 0.706–0.887 | 89.47 | 61.36 | <0.001 |
| T1WI–starVIBE | Reader 1 | 0.708 | 0.597–0.803 | 71.05 | 70.45 | <0.001 |
| Reader 2 | 0.721 | 0.611–0.814 | 73.68 | 61.36 | <0.001 | |
| T1WI–VIBE | Reader 1 | 0.792 | 0.688–0.874 | 89.47 | 61.36 | <0.001 |
| Reader 2 | 0.809 | 0.707–0.888 | 89.47 | 61.36 | <0.001 | |
| T2WI–TSE–fBLADE | Reader 1 | 0.775 | 0.670–0.860 | 81.58 | 63.64 | <0.001 |
| Reader 2 | 0.822 | 0.722–0.898 | 84.21 | 70.45 | <0.001 | |
| MRI | Reader 1 | 0.839 | 0.741–0.911 | 89.47 | 61.36 | <0.001 |
| Reader 2 | 0.857 | 0.762–0.925 | 76.32 | 79.55 | <0.001 | |
| CT + MRI | Reader 1 | 0.867 | 0.774–0.932 | 73.68 | 86.36 | <0.001 |
| Reader 2 | 0.877 | 0.786–0.939 | 76.32 | 86.36 | <0.001 |
AUC, area under the curve; CI, confidence interval; CT, computed tomography; IAC, invasive adenocarcinoma; MRI, magnetic resonance imaging; ROC, receiver operating characteristic; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging; TSE, turbo-spin echo; VIBE, volume interpolated breath-hold examination.
Discussion
This retrospective study demonstrated that morphological features depicted by MRI have diagnostic potential for IAC and tuberculoma. Thus, the non-ionizing modality MRI can be used as a diagnostic tool complementary to conventional CT diagnosis of pulmonary nodules, especially for patients who are sensitive to ionizing radiation, such as children, pregnant women or the patients who need repeated CT scans due to nodules’ follow-up.
Previous reports have shown the accurate detection of pulmonary nodules by MRI sequences including T1WI-VIBE, stack-of star VIBE/T1WI-starVIBE, T2WI-TSE-fBLADE (14,18,21,22). To the detected pulmonary nodules, the accurate location and size measurement are essential (23). All the nodules were located accurately by those 3 MRI sequences in our study. T1WI-starVIBE, T1WI-VIBE and T2WI-TSE-fBLADE averagely underestimated the nodules diameter by approximately 0.15–1.27 mm, which is consistent with the previous studies (21,24). The difference might be resulted from the different breathing state during the images acquisition. The CT scanning requires the person to hold breath after inhaling enough air, while the T1WI-starVIBE and T2WI-TSE-fBLADE do not, they are more appropriate to those who cannot hold breath. The different slice thick and interval, as well as the window location and width of CT and MRI may also affect the nodules size measurement. A previous study showed that ultrashort echo (UTE) underestimated the nodules diameter, which may be potentially caused by the tendency of MRI in general to smooth the margins of structures and the images acquired in free breathing, which may also further blur nodule margins (24). The intra-reader and inter-reader variability of lesions size measurement can also be observed in previous CT studies (25). Therefore, the differences of nodules size measurement between MRI sequences and CT in our study were acceptable in routine clinical practice.
The previous studies were mainly focused on the pulmonary nodules/masses detection, while the assessment of nodules’ morphological features by MRI is scarce (18,24,26,27). UTE refers to a TE shorter than 200 µs, which has delivered image quality close to that of CT and has been used for diagnosis of pulmonary diseases (11,22,24,27-30). The UTE morphological features agreed well with that of CT scanning (24,27). However, UTE is not a conventional sequence widely available in all MRI scanners. Can conventional MRI sequences achieve comparable morphological features as CT? This question attracts clinicians’ attention. Previous studies recommended two major MR sequences that could be used in this setting: the TSE-based technique and the GRE technique to reduce cardiac- and respiratory-motion artifacts (14). T1WI-VIBE, one of the major 3D GRE sequences, showed an acceptable capability of pulmonary nodules detection (range from 54.1% to 87.3%), which is lower than that of conventional TSE sequences (14). In addition, T1WI-VIBE sequence can be used for nodule detection only if the person is able to hold breath. T1WI-starVIBE is based on acquiring k-space data along radial spokes, which is insensitivity to motion, thus allows a free-breathing T1W sequence. T1WI-starVIBE achieves higher resolution images and higher matrix than T1WI-VIBE. Using higher matrix, T1WI-starVIBE maximizes the definition of lesion structures (18). A respiratory-navigated sequence by radial k-space acquisition, commonly known as ‘fBLADE’, provides excellent T2 contrast because of its insensitive to cardiac motion (14,31).
Similar to previous report, we found that the confirmation rate of morphological features assessed by CT was highest (95.12–100%) in contrast to the 43.9–96.34% by MRI (32). Overall, T1WI-starVIBE achieved the best performance (59.76–96.34%). However, T2WI-TSE-fBALDE achieved the best performance in assessing nodules margin (81.71–82.93%), morphology (89.02–90.24%), lobulation (75.61–78.05%) and pleural indentation (76.83–78.05%), which were vital in the nodules’ differential diagnosis (32). Inter-method agreements for the CT and MRI sequences were “moderate/fair” to “excellent” for evaluating nodules morphological features except for “spiculation”, which showed “poor” by T1WI-starVIBE and T1WI-VIBE. We considered that uneven magnetic susceptibility was likely to occur at the edge of the nodule (tumor-lung interface), and the edge of the nodule was easily blurred during imaging, resulting in poor display of the spiculation sign (24). Overall, inter-reader agreement using those MRI sequences for nodule morphological features was excellent, which further documented the reliability of the method.
We also found that unclear margin (assessed by T1WI-starVIBE, T1WI-VIBE), irregular morphology, lobulation, spiculation (T1WI-starVIBE, T2WI-TSE-fBLADE) and air bronchogram (CT, T1WI-starVIBE, T1WI-VIBE) could be predictors of IAC. Those predictors assessed by CT are limited, although those are vital for radiologists in distinguishing benign and malignant solid nodules in daily clinical practice (32). There are considerable overlaps between tuberculoma and IAC regarding CT morphological features, such as spiculation, lobulation and pleural indentation (7). In our study, we found that unclear margin (T1WI-starVIBE: P=0.03, T1WI-VIBE: P=0.03, 0.018) and spiculation (T1WI-starVIBE: P=0.04, 0.003, T2WI-TSE-fBLADE: P=0.005, <0.001) depicted by MRI could differentiate IAC from tuberculoma, while those by CT cannot (P>0.05). CT achieves high spatial resolution images and can depict nodules’ margin clearly. IAC and tuberculoma can both presented as non-calcified nodules with clear margin and spiculation by CT, which confused the clinicians. However, IAC is more often with irregular morphology and lobulation, so that uneven magnetic signal may occur at the tumor-lung interface. Those may result in the nodules unclear margin depiction by MRI. Spiculation signs of IAC and tuberculoma have different pathological mechanisms (32). Spiculation of IAC represents the tumor invades the surrounding structure with or without fibrosis and edema, while that of tuberculoma represents inflammatory infiltration and fibrosis. MRI with high tissue resolution may display those spiculation signs in varying degree, so that enable the differentiation between IAC and tuberculoma. We also can’t differentiate IAC from tuberculoma by assessing pleural indentation not only in CT, but also in MRI. Pleural indentation is a key morphological feature that suggests the possibility of malignant, and even visceral pleural invasion (33). However, tuberculoma can also affect the pleural. The relationship between the nodule and the pleura can be divided into several different conditions, such as pleural attachment, indentation, or both, and different relationships may correlate with different pathological results (33,34). In our study, both pleural indentation and accompany with attachment were identified as pleural indentation. The combination of CT with MRI achieved best performance in the differential diagnosis, with AUC value of 0.867–0.877 (P≤0.001), although the difference was modest but with statistically significant. The combined MRI sequences also achieved better performance than CT (AUC, 0.839/0.857 versus 0.808), although, the difference had no statistically significance. Those support our supposition that the conventional and modified MRI sequences can be complementary roles to CT, concerning nodules morphological assessment to differentiate IAC from tuberculoma.
Diffusion-weighted imaging (DWI) assessed the diffusivity of water molecules within tissue and widely investigated to differentiate malignant from benign lesion (14). The reported accuracy ranged from 50% to 85.7% (14), which was not better than our results. The study concerning using DWI to differentiate IAC from tuberculoma is also scarce. However, it may take longer time for thoracic DWI scanning to achieve high quality images, which can meet the diagnosis qualification in the routine clinical practice.
There are some limitations in this study. Firstly, due to the strict enrollment protocol, the sample size of our study was small. Secondly, UTE was not studied, which is valuable to nodule detection and morphological features assessment. However, UTE is not a common technique widely used in clinical practice and our study aimed to investigate the common MRI sequences capability in nodules morphological features assessment. Thirdly, some functional sequences, such as DWI, was not included in this study, which may provide more biological information.
Conclusions
Combined conventional and modified anatomical MRI sequences play potential roles in assessing pulmonary nodules morphological features. T1WI-starVIBE and T2WI-TSE-fBLADE achieved better performance. The MRI sequences can also provide more anatomical information which is helpful to elevate the diagnostic capability of CT only for differentiating IAC and tuberculoma.
Supplementary
The article’s supplementary files as
Acknowledgments
None.
Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Our study received ethical approval from the ethics committee of Shanghai Public Health Clinical Center (No. 2019-S021-02). The necessity for written informed consent was waived, as data were analyzed retrospectively and anonymously.
Footnotes
Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1677/rc
Funding: This work was supported by the Shanghai Municipal Science and Technology Commission (No. 22YF1443500), National Natural Science Foundation of China (No. 82172030).
Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1677/coif). The authors have no conflicts of interest to declare.
Data Sharing Statement
Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1677/dss
References
- 1.WHO. Cancer. Available online: https://www.who.int/news-room/fact-sheets/detail/cancer
- 2.Bach PB, Mirkin JN, Oliver TK, et al. Benefits and harms of CT screening for lung cancer: a systematic review. JAMA 2012;307:2418-29. 10.1001/jama.2012.5521 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Niyonkuru A, Chen X, Bakari KH, et al. Evaluation of the diagnostic efficacy of (18) F-Fluorine-2-Deoxy-D-Glucose PET/CT for lung cancer and pulmonary tuberculosis in a Tuberculosis-endemic Country. Cancer Med 2020;9:931-42. 10.1002/cam4.2770 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Fick CN, Dunne EG, Vanstraelen S, et al. High-risk features associated with recurrence in stage I lung adenocarcinoma. J Thorac Cardiovasc Surg 2025;169:436-444.e6. 10.1016/j.jtcvs.2024.05.009 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Ruan Y, Cao W, Han J, et al. Prognostic impact of the newly revised IASLC proposed grading system for invasive lung adenocarcinoma: a systematic review and meta-analysis. World J Surg Oncol 2024;22:302. 10.1186/s12957-024-03584-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.WHO. Tuberculosis. Available online: https://www.who.int/health-topics/tuberculosis#tab=tab_1
- 7.Zhang J, Han T, Ren J, et al. Discriminating Small-Sized (2 cm or Less), Noncalcified, Solitary Pulmonary Tuberculoma and Solid Lung Adenocarcinoma in Tuberculosis-Endemic Areas. Diagnostics (Basel) 2021;11:930. 10.3390/diagnostics11060930 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Feng M, Yang X, Ma Q, et al. Retrospective analysis for the false positive diagnosis of PET-CT scan in lung cancer patients. Medicine (Baltimore) 2017;96:e7415. 10.1097/MD.0000000000007415 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Li CR, Li YZ, Li YM, et al. Dynamic and contrast enhanced CT imaging of lung carcinoma, pulmonary tuberculoma, and inflammatory pseudotumor. Eur Rev Med Pharmacol Sci 2017;21:1588-92. [PubMed] [Google Scholar]
- 10.Zhuo Y, Zhan Y, Zhang Z, et al. Clinical and CT Radiomics Nomogram for Preoperative Differentiation of Pulmonary Adenocarcinoma From Tuberculoma in Solitary Solid Nodule. Front Oncol 2021;11:701598. 10.3389/fonc.2021.701598 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Burris NS, Johnson KM, Larson PE, et al. Detection of Small Pulmonary Nodules with Ultrashort Echo Time Sequences in Oncology Patients by Using a PET/MR System. Radiology 2016;278:239-46. 10.1148/radiol.2015150489 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Azour L, Ohno Y, Biederer J, et al. Lung MRI: Indications, Capabilities, and Techniques-AJR Expert Panel Narrative Review. AJR Am J Roentgenol 2025;225:e2532637. 10.2214/AJR.25.32637 [DOI] [PubMed] [Google Scholar]
- 13.Carl M, Wang J, Lo J, et al. Optimized 3D UTE and ZTE MRI for high-resolution lung imaging: A comparative study. Magn Reson Med 2026;95:962-74. 10.1002/mrm.70100 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ohno Y, Kauczor HU, Hatabu H, et al. MRI for solitary pulmonary nodule and mass assessment: Current state of the art. J Magn Reson Imaging 2018;47:1437-58. 10.1002/jmri.26009 [DOI] [PubMed] [Google Scholar]
- 15.Allen BD, Schiebler ML, Sommer G, et al. Cost-effectiveness of lung MRI in lung cancer screening. Eur Radiol 2020;30:1738-46. 10.1007/s00330-019-06453-9 [DOI] [PubMed] [Google Scholar]
- 16.Kumar S, Rai R, Stemmer A, et al. Feasibility of free breathing Lung MRI for Radiotherapy using non-Cartesian k-space acquisition schemes. Br J Radiol 2017;90:20170037. 10.1259/bjr.20170037 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zhang L, Tian C, Wang P, et al. Comparative study of image quality between axial T2-weighted BLADE and turbo spin-echo MRI of the upper abdomen on 3.0 T. Jpn J Radiol 2015;33:585-90. 10.1007/s11604-015-0463-9 [DOI] [PubMed] [Google Scholar]
- 18.Yu N, Duan H, Yang C, et al. Free-breathing radial 3D fat-suppressed T1-weighted gradient echo (r-VIBE) sequence for assessment of pulmonary lesions: a prospective comparison of CT and MRI. Cancer Imaging 2021;21:68. 10.1186/s40644-021-00441-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Yan Q, Yang S, Shen J, et al. 3T magnetic resonance for evaluation of adult pulmonary tuberculosis. Int J Infect Dis 2020;93:287-94. 10.1016/j.ijid.2020.02.006 [DOI] [PubMed] [Google Scholar]
- 20.Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;33:159-74. [PubMed] [Google Scholar]
- 21.Yang S, Shan F, Yan Q, et al. A pilot study of native T1-mapping for focal pulmonary lesions in 3.0 T magnetic resonance imaging: size estimation and differential diagnosis. J Thorac Dis 2020;12:2517-28. 10.21037/jtd.2020.03.42 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ohno Y, Takenaka D, Yoshikawa T, et al. Efficacy of Ultrashort Echo Time Pulmonary MRI for Lung Nodule Detection and Lung-RADS Classification. Radiology 2022;302:697-706. 10.1148/radiol.211254 [DOI] [PubMed] [Google Scholar]
- 23.MacMahon H, Naidich DP, Goo JM, et al. Guidelines for Management of Incidental Pulmonary Nodules Detected on CT Images: From the Fleischner Society 2017. Radiology 2017;284:228-43. 10.1148/radiol.2017161659 [DOI] [PubMed] [Google Scholar]
- 24.Wielpütz MO, Lee HY, Koyama H, et al. Morphologic Characterization of Pulmonary Nodules With Ultrashort TE MRI at 3T. AJR Am J Roentgenol 2018;210:1216-25. 10.2214/AJR.17.18961 [DOI] [PubMed] [Google Scholar]
- 25.Revel MP, Bissery A, Bienvenu M, et al. Are two-dimensional CT measurements of small noncalcified pulmonary nodules reliable? Radiology 2004;231:453-8. 10.1148/radiol.2312030167 [DOI] [PubMed] [Google Scholar]
- 26.Meier-Schroers M, Homsi R, Schild HH, et al. Lung cancer screening with MRI: characterization of nodules with different non-enhanced MRI sequences. Acta Radiol 2019;60:168-76. 10.1177/0284185118778870 [DOI] [PubMed] [Google Scholar]
- 27.Feng H, Shi G, Liu H, et al. The Value of PETRA in Pulmonary Nodules of <3 cm Among Patients With Lung Cancer. Front Oncol 2021;11:649625. 10.3389/fonc.2021.649625 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bae K, Jeon KN, Hwang MJ, et al. Comparison of lung imaging using three-dimensional ultrashort echo time and zero echo time sequences: preliminary study. Eur Radiol 2019;29:2253-62. 10.1007/s00330-018-5889-x [DOI] [PubMed] [Google Scholar]
- 29.Feng L, Delacoste J, Smith D, et al. Simultaneous Evaluation of Lung Anatomy and Ventilation Using 4D Respiratory-Motion-Resolved Ultrashort Echo Time Sparse MRI. J Magn Reson Imaging 2019;49:411-22. 10.1002/jmri.26245 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Ohno Y, Koyama H, Yoshikawa T, et al. Pulmonary high-resolution ultrashort TE MR imaging: Comparison with thin-section standard- and low-dose computed tomography for the assessment of pulmonary parenchyma diseases. J Magn Reson Imaging 2016;43:512-32. 10.1002/jmri.25008 [DOI] [PubMed] [Google Scholar]
- 31.Raptis CA, Ludwig DR, Hammer MM, et al. Building blocks for thoracic MRI: Challenges, sequences, and protocol design. J Magn Reson Imaging 2019;50:682-701. 10.1002/jmri.26677 [DOI] [PubMed] [Google Scholar]
- 32.Snoeckx A, Reyntiens P, Desbuquoit D, et al. Evaluation of the solitary pulmonary nodule: size matters, but do not ignore the power of morphology. Insights Imaging 2018;9:73-86. 10.1007/s13244-017-0581-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Yang S, Yang L, Teng L, et al. Visceral pleural invasion by pulmonary adenocarcinoma ≤3 cm: the pathological correlation with pleural signs on computed tomography. J Thorac Dis 2018;10:3992-9. 10.21037/jtd.2018.06.125 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Sun Q, Li P, Zhang J, et al. CT Predictors of Visceral Pleural Invasion in Patients with Non-Small Cell Lung Cancers 30 mm or Smaller. Radiology 2024;310:e231611. 10.1148/radiol.231611 [DOI] [PubMed] [Google Scholar]