Use of diffusion-weighted magnetic resonance imaging to distinguish between lung cancer and focal inflammatory lesions: a comparison of intravoxel incoherent motion derived parameters and apparent diffusion coefficient

Abstract

Background

Using imaging techniques to diagnose malignant and inflammatory lesions in the lung can be challenging.

Purpose

To compare intravoxel incoherent motion (IVIM) and apparent diffusion coefficient (ADC) magnetic resonance imaging (MRI) analysis in their ability to discriminate lung cancer from focal inflammatory lung lesions.

Material and Methods

Thirty-eight patients with lung masses were included: 30 lung cancers and eight inflammatory lesions. Patients were imaged with 3.0T MRI diffusion weighted imaging (DWI) using 10 b values (range, 0–1000 s/mm²). Tissue diffusivity (D), pseudo-diffusion coefficient (D*), and perfusion fraction (f) were calculated using segmented biexponential analysis. ADC (total) was calculated with monoexponential fitting of the DWI data. D, D*, f, and ADC were compared between lung cancer and inflammatory lung lesions. Receiver operating characteristic analysis was performed for all DWI parameters.

Results

The ADC was significantly higher for inflammatory lesions than for lung cancer ([1.21 ± 0.20] × 10⁻³ mm²/s vs. [0.97 ± 0.15] × 10⁻³ mm²/s; P = 0.004). By IVIM, f was found to be significantly higher in inflammatory lesions than lung cancer ([46.10 ± 12.92] % vs. [29.29 ± 10.89] %; P = 0.005). There was no difference in D and D* between lung cancer and inflammatory lesions (P = 0.747 and 0.124, respectively). f showed comparable diagnostic performance with ADC in differentiating lung cancer from inflammatory lung lesions, with areas under the curve of 0.833 and 0.826, sensitivity 80.0% and 73.3%, and specificity 75.0% and 87.5%, respectively.

Conclusion

The IVIM parameter f value provides comparable diagnostic performance with ADC and could be used as a surrogate marker for differentiating lung cancer from inflammatory lesions.

Keywords

Intravoxel incoherent motion (IVIM)diffusion magnetic resonance imaging lung neoplasms inflammation

Introduction

Focal lung lesions are common entities encountered in clinical chest imaging. Computed tomography (CT) is often the modality of choice for diagnosis and management of focal lung lesions because of its excellent spatial resolution and superb contrast between lesions and normal lung parenchyma. However, it remains challenging for CT to differentiate a malignancy from an inflammatory lesion because of considerable overlap in their morphological features (1,2). Fludeoxyglucose (18F) (18F-FDG) positron emission tomography (PET) is more sensitive and specific in the diagnosis of malignancy than CT, but many benign conditions such as infection and granuloma can also show intense 18F-FDG uptake and produce false-positive results (2 –4). Therefore, a non-invasive tool that can provide more details about the internal characteristics of the lesions beyond the morphology and metabolic changes is needed.

Diffusion-weighted imaging (DWI) can detect in vivo molecular random translational motion (i.e. Brownian motion) using magnetic resonance imaging (MRI). The derivation of the apparent diffusion coefficient (ADC) has been widely applied during the last two decades. ADC has proved to be a valuable technique in the assessment of lung cancer, especially in the detection and characterization of tumors and mediastinal lymph node metastases, with fewer false positives than PET (5 –9). The intravoxel incoherent motion (IVIM) method of MRI analysis (10) further separates images of diffusion into pure diffusion and perfusion (11). IVIM can cover the overall molecular displacements resulting not only from Brownian motion, but also from movement of blood in the microvasculature, which mimics true diffusion and allows reflection of tissue diffusivity and tissue microcapillary perfusion to be separated without contrast agent injection (12). IVIM was initially limited to neuroimaging because of low signal and motion artifacts (13). However, MRI scanners have advanced, with improved gradient performance and parallel imaging techniques, increased signal-to-noise ratio (SNR) and decreased scan time, allowing IVIM to be applied to imaging of different parts of the body (14). IVIM parameters, including pure diffusion coefficient (D), pseudo-diffusion coefficient (D*) and perfusion fraction (f), have been used to distinguish various cancers from healthy tissue or other lesions (15).

To our knowledge, only one other study has evaluated IVIM parameters for lung cancer, in the assessment of solitary pulmonary nodules (SPNs) (16). The use of IVIM to differentiate lung cancer from focal inflammatory lesions has not been described. The aim of this study was to examine the diagnostic performance of IVIM-DWI derived parameters, including D, D*, and f, against that of ADC.

Material and Methods

Study population

This prospective study was approved by the institutional review board of our hospital, and written informed consent was obtained from all study subjects. Between November 2013 and May 2014, consecutive patients who met the following inclusion criteria were included in this study: (i) CT scan revealed one focal pulmonary nodule or mass larger than 1.5 cm; (ii) cavity, calcification, or ground glass portion inside the lesion was no more than one-third; (c) no previous treatment was given. The exclusion criteria were: (i) contraindications to MRI; (ii) patients with poor image quality resulting from motion or a severe magnetic susceptibility artifact were excluded from data analysis.

Diagnostic criteria

The diagnoses of all primary lung cancers and five of eight benign lesions were made histologically by either surgery or biopsy within 10 days from MRI examination. The diagnoses of three inflammatory nodules were established with radiologic follow-up studies over a period of 1 month that revealed disappearance or significant regression of the nodules after initiation of antibacterial or steroid therapy.

Conventional MRI

MRI was performed with a 3.0 T scanner (Achieva 3.0T, Philips Healthcare, Best, The Netherlands) with gradient amplitude up to 80 mT/m and a 16-element phased-array coil (SENSE XL Torso). Routine MR sequences included the following: axial gradient echo T1-weighted (T1W) imaging, and axial and coronal turbo spin-echo T2-weighted (T2W) imaging. MR images were obtained during end-inspiration breath-holding. Detailed imaging parameters are described in Table 1.

Table 1.

Conventional MRI sequence parameters.

Sequence	Plane	TR/TE (ms)	Flip angle (degrees)	Matrix	Field of view	Section thickness/gap (mm)	Acquisition time (s)
T1W GRE	Axial	10/2.3	15	256 × 188	375	5.0/0.5	60
T2W TSE	Axial	900/80	90	312 × 266	375	7.0/0.7	72
T2W TSE	Coronal	970/80	90	256 × 220	350	5.0/0.5	23

GRE, gradient echo; MRI, magnetic resonance imaging; TE, echo time; TR, repetition time; TSE, turbo spin echo.

Diffusion-weighted imaging

Multi-b diffusion-weighted MRI scans (b = 0, 25, 50, 75, 100, 200, 400, 600, 800, 1000 s/mm²) were acquired using a single-shot echo-planar imaging pulse sequence with free breathing. The parameters were as follows: repetition time, 899 ms; echo time, 56 ms; field of view, 375 mm; matrix, 256 × 256; section thickness, 3.0 mm with gap 0.3 mm. Parallel imaging was used, with an acceleration factor of 3 and a half scan factor of 69.8%. The receiver bandwidth was 3475.5 Hz/pixel, and fat was suppressed using spectral presaturation inversion recovery (SPIR). Total acquisition time was 3 min 36 s.

MRI analysis

ADC of the lung lesions was calculated using a monoexponential fit of signal intensity for b = 0 and 1000 s/mm², with the following equation:

S (b) = S_{0} exp (- bADC)

(1)

A biexponential approach was used for calculation of the pure diffusion coefficient D, the perfusion-related diffusion coefficient D*, and the perfusion fraction f as follows:

S (b) / S_{0} = (1 - f) exp (- bD) + f exp [- b (D + D^{*})]

(2)

Where S(b) is the signal intensity for a given b value and S₀ is the signal intensity at b = 0 s/mm².

The IVIM parameters were generated from an in-house program fitted on a voxel-by-voxel basis using the Levenberg–Marquardt algorithm. Because the contribution of D* on signal decay can be neglected at high b values (b ≥ 200 s/mm²), D was obtained using high b values and a mono-exponential fit. Subsequently, with the resulting D as a fixed parameter, f and D* were calculated using a non-linear regression algorithm for all b values.

First, regions of interest (ROI) measurements of IVIM parametric maps were obtained with Image J software (National institute of Health, Bethesda, MD, USA). ROIs were drawn to encompass as much of the homogeneous area of the lesion as possible and avoid necrosis or artifacts by reference to conventional MR images. Second, ROI measurement of ADC was obtained using a workstation (Extended Workspace, Philips Healthcare, Best, The Netherlands). The ROIs were kept as close as possible to those on IVIM parametric maps. All ROIs were placed in consensus by two radiologists (with 8 and 15 years of experience in MRI diagnosis), and the mean value of three measurements was obtained.

Statistical analysis

Statistical tests were carried out using statistical software (SPSS, version 17.0.0; SPSS/IBM, Chicago, IL, USA). Data are expressed as means and standard deviations, and independent samples t-tests were used to detect significant differences in ADC, D, f, and D* between malignant and benign lesions. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the diagnostic performance of the IVIM parameters and ADC to predict malignancy and to determine suitable threshold values. The area under the ROC curves (AUCs) for the IVIM parameters was compared to that for ADC, and the optimal cutoff value was defined as the point that yielded the best sensitivity and specificity for the differentiation. Statistical significance was considered for P < 0.05.

Results

Study population diagnosis

Forty-five patients were included. Lung MRI examinations were successfully performed in all patients. Seven patients were excluded because of poor image quality resulting from motion artifacts or severe magnetic susceptibility artifact. Finally, 38 cases were included in this study (mean age, 58.80 ± 10.93 years; range, 31–79 years; 22 men, 16 women). The final diagnosis was primary lung cancer in 30 patients (18 adenocarcinomas, 8 squamous carcinomas, 1 small cell lung cancer, 2 lymphoepithelioma-like carcinomas, and 1 compound large cell neuroendocrine carcinoma) and inflammatory lesions in eight patients (3 inflammatory nodules, 2 cases of organizing pneumonia, 2 granulomas, and 1 cryptococcal lesion).

MRI characteristics

The longest diameter of the lesion was in the range of 1.5–6.8 cm, with a mean of 3.21 ± 1.62 cm. Both lung cancer and inflammatory lesions showed high signal intensity in DWI. The inflammatory lesions showed a rapid signal decay curve in the range of lower b values (b < 200 s/mm²) (Fig. 1). The signal decay curve was more gradual in the lung cancer group (Fig. 2). T1W imaging, ADC, D, D*, and f maps of representative cases of inflammatory lesions and lung cancer are shown in Figs. 1 and 2.

Fig. 1.

Representative example of MRI of an inflammatory lesion. The patient was a 49-year-old man with organizing pneumonia in the anterior segment of the right lower lobe. (a) T1W imaging demonstrates a mass with intermediate signal intensity. (b–e) Parameter map of ADC, D, D*, and f calculated from DWI data. The measured D, f, D*, and ADC values were 0.80 × 10^–3 mm²/s, 33.61%, 63.41 × 10^–3 mm²/s, and 1.13 × 10^–3 mm²/s, respectively. (f) The signal-decay curve, plotted as a function of the b values, is steep in low b values.

Fig. 2.

Representative example of MRI of lung cancer. The patient was a 39-year-old man with adenocarcinoma in the apical segment of the anterior segment of right upper lobe. (a) T1W imaging demonstrates an irregular mass with intermediate signal intensity. (b–e) Parameter map of ADC, D, D*, and f calculated from DWI data. The measured D, f, D*, and ADC values were 0.61 × 10^–3 mm²/s, 20.87%, 20.14 × 10^–3 mm²/s, and 0.94 × 10^–3 mm²/s, respectively. (f) The signal-decay curve, plotted as a function of the b values, is more gradual in low b values.

The ADC values and IVIM parameters for the two groups are shown as means ± standard deviations in Table 2. ADC was significantly higher for inflammatory lung lesions than for lung cancer ([1.21 ± 0.20] × 10^–3 mm²/s vs. [0.97 ± 0.15] × 10^–3 mm²/s; P = 0.004). In terms of IVIM parameters, perfusion fraction f was found to be significantly higher in inflammatory lesions than in lung cancer ([46.10 ± 12.92] % vs. [29.29 ± 10.89] %; P = 0.005). There was no difference in pure diffusion coefficient D and perfusion-related coefficient D* between lung cancer and inflammatory lesions (P = 0.747 and 0.124, respectively).

Table 2.

Comparison of ADC and IVIM parameters between lung cancer and inflammatory lesions.

Parameters	Lung cancer (n = 30)	Inflammatory lesions (n = 8)	P value
ADC (×10^–3 mm²/s)	0.97 ± 0.15	1.21 ± 0.20	0.004
D (×10^–3 mm²/s)	0.65 ± 0.14	0.68 ± 0.18	0.747
D* (×10^–3 mm²/s)	32.73 ± 15.72	52.72 ± 35.19	0.124
f (%)	29.29 ± 10.89	46.10 ± 12.92	0.005

ADC, apparent diffusion coefficient; D, pure diffusion coefficient, D*, pseudo-diffusion coefficient, f, perfusion fraction; IVIM, intravoxel incoherent motion.

Diagnostic performance of the MRI parameters

ROC curves for ADC, D, D*, and f values are illustrated in Fig. 3. Both ADC and f provide malignancy tests that are different from the non-discrimination line (AUC _ADC = 0.833, P = 0.0006; AUC _f = 0.829, P = 0.0001, respectively). Both D and D* were poor malignancy markers with an AUC of 0.538 (P = 0.7834) and 0.679 (P = 0.1681), respectively. No significant difference was found between ADC and f as malignancy markers (P = 0.9481). The optimal cutoff values of ADC and f for discrimination were 1.0224 × 10^–3 mm²/s and 37.4309%, respectively, which yielded sensitivity 73.3% and specificity 87.5%, and sensitivity 80.0% and specificity 75.0%, respectively.

Fig. 3.

Receiver operating characteristic (ROC) curves for ADC and IVIM parameters. There was no significant difference between ADC and f in the area under the curve (AUC) (0.833 vs. 0.829, P = 0.9481).

Discussion

This study evaluated the use of IVIM parameters in the diagnosis of focal lung lesions, and compared them to ADC. Our study showed that IVIM-derived parameters can provide comparable diagnostic performance to ADC, and can facilitate the understanding of the tissue characteristics, including diffusion and perfusion, of benign and malignant lung diseases.

The development of these MRI techniques has the potential to improve the diagnosis of lung cancer by non-invasive methods. CT relies quite heavily on morphological distinctions, so in many cases of lung cancer other methods are also needed for diagnosis, including growth monitoring, material-enhanced CT, and PET (16 –18). DWI has already been shown to have benefit in lung cancer diagnosis with potentially fewer false positive results than PET (5 –9). Meta-analysis suggests DWI has an accuracy of 91% in differentiating malignant from benign lung lesions, slightly lower than the accuracies achieved by 18 F-FDG PET at 94%, dynamic contrast-enhanced CT at 93%, and dynamic contrast-enhanced MRI at 94% (19). However, DWI is non-invasive and does not involve any additional injection of contrast medium. IVIM has shown higher accuracy than ADC, resulting from DWI, in diagnosing many different types of cancer (15,20 –22). Both ADC and IVIM parameters can distinguish between benign and malignant SPNs (16), although that particular study found that the signal intensity of the lesion-to-spinal cord ratio (LSR) was more accurate and specific than ADC, D, or f.

ADC calculated from DWI is equal to the true diffusion coefficient D only when there is no other incoherent motion in the biologic tissues. The IVIM model with a biexponential approach extracts additional microcirculation information by applying multiple b values (11). However, the number of b values used for the IVIM model varies from 4 to more than 10 in published literature. There is still no consensus on the number and magnitudes of b values that should be used (23). Since measurements at lower b values have been shown to be less stable and reproducible than measurements at higher b values, it has been suggested that measurements at a larger number of lower b values should be obtained in order to reduce measurement errors and signal-to noise variations (23,24). Our study chose 10 b values with 6 b values ≤200 s/mm², which is sufficiently robust to guarantee the accuracy and reproducibility of the measurement.

Although breath-holding or respiratory-triggered acquisitions can reduce the adverse effect of motion on DWI to some degree, both of these have disadvantages. Breath-holding can sometimes be too challenging for patients to perform successively. It also shows poor SNR, great sensitivity to distortion and low spatial resolution. Respiratory-triggered acquisition is dependent on the consistency of the patient’s breathing. Irregular breathing of the patient will decrease the time-efficiency or even make the trigger regime unusable (23,25,26). On the contrary, free breathing is more versatile, and several studies have proved the robustness of free breathing in body DWI (6,7,25 –27). All IVIM data obtained using free breathing in this study were clinically feasible with acceptable examination times and sufficient SNRs.

In our study, the ADC values of lung cancers were significantly lower than those of inflammatory lesions, comparable with previous studies (7,28,29). It has been widely accepted that the low ADC of malignant tumors results from high cellularity and consequent diffusion restriction. However, ADC represents all the incoherent motion and is affected by diffusion and perfusion, which reflect tissue microstructure and microcirculation, respectively. This was reinforced by our study, in which we observed that ADC was higher than true diffusion D in both groups, and all of the signal decay curves showed fast decay for lower b values suggestive of biexponential fitting.

It is interesting that our study did not show a significant difference in D between the two groups (0.65 ± 0.14 mm²/s vs. 0.68 ± 0.18 mm²/s), which means that the difference in ADC can be mainly ascribed to the perfusion related parameters (i.e. f or D*). This finding is in good agreement with one study differentiating between mass-forming chronic pancreatitis and pancreatic carcinoma (30). A possible explanation is that pure diffusion is also restricted in inflammation due to extravasation and recruitment of leukocytes to the target tissue, as well as cellular debris accumulation. Although the underlying histopathologic change during inflammation is different from that in lung cancer, the restricted diffusion was similar as demonstrated by DWI.

Another important finding of this study is that perfusion fraction f was significantly higher in inflammatory lesions than in lung cancers. This result is comparable with the only previous study using IVIM in the differentiation of lung cancer from obstructive pneumonia (29). The mean f value for lung cancer in our study was close to theirs (29.29% vs. 23.95%), but the mean f value for inflammatory lesions was even higher (46.10% vs. 34.66%). According to IVIM theory, f value always increases with increased tissue microcirculation. Malignant tumors are supposed to have higher f values than benign lesions because of neovascularization. However, these results are not unreasonable because inflammation is usually characterized by marked vascular changes, including vasodilation, increased permeability, and increased blood flow, which mostly occur at the capillary network. The finding was also in good agreement with a CT perfusion study using exogenous contrast media which found some SPNs with active infection had even higher perfusion, peak enhancement increment (PEI), and blood volume (BV) than malignant SPNs, and the time to peak (TTP) for active infection was steeper than that for malignancy (31). Several IVIM studies of other organs have also shown that the perfusion fraction f of benign lesions, especially when there is inflammation, is higher than that of malignancy (30,32). However, another possible reason for higher f values in inflammatory lesions is that the f value closely correlates with echo time (TE) and is influenced by relaxation effects and T2 contribution (33). The f value and ADC showed similar diagnostic performance in distinguishing lung cancer from inflammatory lesions. Despite the lack of improvement in diagnostic accuracy, the f value can facilitate the understanding of diffusion imaging, and further research into its role for detecting the microstructure or microcirculation changes of lung cancer is merited.

D* is considered to be proportional to the mean capillary segment length and average blood velocity (11). Our results showed D* tended to be higher in inflammatory lesions than lung cancer, but statistical significance was not attained. However, as many studies have stated, the calculation of D* may not be reliable due to its huge standard deviation, data instability and dependence on SNR levels (23,34). For the present, poor reproducibility of this perfusion parameter makes it unsuitable as a clinical parameter.

It should be noted that the technique described in the present study is more expensive and thus less readily available than other diagnostic approaches (such as chest CT and X-ray) in many parts of the world, including regions where granulomatous disease is common. Nonetheless, the purpose of our study was to obtain an initial experience of IVIM, and explore the feasibility of using this novel MRI technique for the diagnosis of a lung mass, and in particular the differentiation of lung cancer from a focal inflammatory lesion. Our findings suggest that this approach could be used as a complementary diagnostic technique rather than a routine investigation. Although its utility at present may be somewhat limited by its affordability, it is hoped that, in future, more widespread use of this technique will be possible.

Our study has some limitations. The study population was limited in number, and there was an imbalance between the numbers of patients in each group (30 in the lung cancer group, 8 in the inflammatory lesion group). Furthermore, formal power calculations were not performed, and thus it cannot be excluded that our study was underpowered for some of the evaluated parameters. Therefore, it will be necessary to confirm our findings in a larger-scale study. Nonetheless, an important goal of this study was to explore the feasibility of IVIM in the diagnosis of local lung lesions, and we believe that feasibility was demonstrated despite the low number of patients enrolled. Also, in order to avoid the volume average effect and susceptibility to artifacts, only nodules or masses of more than 1.5 cm in diameter were included in this cohort. This recruitment biased the study population. In addition, this study did not compare the perfusion related parameters of IVIM with conventional perfusion imaging using exogenous contrast.

In conclusion, our study has shown the feasibility of evaluating focal lung lesions using IVIM. The lower ADC value of lung cancer may mainly be ascribed to a lower perfusion fraction f value than that in local inflammatory lesions. The f value provides comparable diagnostic performance with ADC. IVIM parameters can be used as a useful surrogate marker in differentiating lung cancer from inflammatory lesions.

Footnotes

Acknowledgments

We gratefully acknowledge the assistance of Mr. Zhipeng Zhang, from Philips Healthcare, China, for his help on in-house IVIM analysis program development.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Scientific Foundation of China (No. 81271654, 81271569 and U1301258).

References

Truong

Sabloff

. Multidetector CT of solitary pulmonary nodules. Radiol Clin North Am 2010; 48: 141–155.

Sim

Poon

. Imaging of solitary pulmonary nodule-a clinical review. Quant Imaging Med Surg 2013; 3: 316–326.

Bunyaviroch

Coleman

. PET evaluation of lung cancer. J Nucl Med 2006; 47: 451–469.

Chang

Lee

Goo

. False positive and false negative FDG-PET scans in various thoracic diseases. Korean J Radiol 2006; 7: 57–69.

Mori

Nomori

Ikeda

. Diffusion-weighted magnetic resonance imaging for diagnosing malignant pulmonary nodules/masses: comparison with positron emission tomography. J Thorac Oncol 2008; 3: 358–364.

Satoh

Kitazume

Ohdama

. Can malignant and benign pulmonary nodules be differentiated with diffusion-weighted MRI? Am J Roentgenol 2008; 191: 464–470.

Liu

. Usefulness of diffusion-weighted MR imaging in the evaluation of pulmonary lesions. Eur Radiol 2010; 20: 807–815.

Gumustas

Inan

Akansel

. Differentiation of malignant and benign lung lesions with diffusion-weighted MR imaging. Radiol Oncol 2012; 46: 106–113.

Kosucu

Tekinbas

Erol

. Mediastinal lymph nodes: assessment with diffusion-weighted MR imaging. J Magn Reson Imaging 2009; 30: 292–297.

10.

Le Bihan

Breton

Lallemand

. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 1986; 161: 401–407.

11.

Le Bihan

Breton

Lallemand

. Separation of diffusion and perfusion in intravoxel incoherent motion MR imaging. Radiology 1988; 168: 497–505.

12.

Takahara

Kwee

. Low b-value diffusion-weighted imaging: emerging applications in the body. J Magn Reson Imaging 2012; 35: 1266–1273.

13.

Le Bihan

. Intravoxel incoherent motion perfusion MR imaging: a wake-up call. Radiology 2008; 249: 748–752.

14.

Koh

Collins

. Diffusion-weighted MRI in the body: applications and challenges in oncology. Am J Roentgenol 2007; 188: 1622–1635.

15.

Chandarana

Lee

Hecht

. Comparison of biexponential and monoexponential model of diffusion weighted imaging in evaluation of renal lesions: preliminary experience. Invest Radiol 2011; 46: 285–291.

16.

Koyama

Ohno

Seki

. Value of diffusion-weighted MR imaging using various parameters for assessment and characterization of solitary pulmonary nodules. Eur J Radiol 2015; 84: 509–515.

17.

Erasmus

Connolly

McAdams

. Solitary pulmonary nodules: Part I. Morphologic evaluation for differentiation of benign and malignant lesions. Radiographics 2000; 20: 43–58.

18.

Swensen

Viggiano

Midthun

. Lung nodule enhancement at CT: multicenter study. Radiology 2000; 214: 73–80.

19.

Chen

Zhang

Bao

. Meta-analysis of diffusion-weighted MRI in the differential diagnosis of lung lesions. J Magn Reson Imaging 2013; 37: 1351–1358.

20.

Jansen

Stambuk

. Comparing primary tumors and metastatic nodes in head and neck cancer using intravoxel incoherent motion imaging: a preliminary experience. J Comput Assist Tomogr 2013; 37: 346–352.

21.

Lemke

Laun

Klauss

. Differentiation of pancreas carcinoma from healthy pancreatic tissue using multiple b-values: comparison of apparent diffusion coefficient and intravoxel incoherent motion derived parameters. Invest Radiol 2009; 44: 769–775.

22.

Yoon

Lee

. Evaluation of hepatic focal lesions using diffusion-weighted MR imaging: comparison of apparent diffusion coefficient and intravoxel incoherent motion-derived parameters. J Magn Reson Imaging 2014; 39: 276–285.

23.

Koh

Collins

Orton

. Intravoxel incoherent motion in body diffusion-weighted MRI: reality and challenges. Am J Roentgenol 2011; 196: 1351–1361.

24.

Koh

Blackledge

Collins

. Reproducibility and changes in the apparent diffusion coefficients of solid tumours treated with combretastatin A4 phosphate and bevacizumab in a two-centre phase I clinical trial. Eur Radiol 2009; 19: 2728–2738.

25.

Watanabe

Kanematsu

Goshima

. Characterizing focal hepatic lesions by free-breathing intravoxel incoherent motion MRI at 3.0 T. Acta Radiol 2014; 55: 1166–1173.

26.

Jerome

Orton

d’Arcy

. Comparison of free-breathing with navigator-controlled acquisition regimes in abdominal diffusion-weighted magnetic resonance images: Effect on ADC and IVIM statistics. J Magn Reson Imaging 2014; 39: 235–240.

27.

Coolen

Vansteenkiste

De Keyzer

. Characterisation of solitary pulmonary lesions combining visual perfusion and quantitative diffusion MR imaging. Eur Radiol 2014; 24: 531–541.

28.

Usuda

Sagawa

Motono

. Diagnostic performance of diffusion weighted imaging of malignant and benign pulmonary nodules and masses: comparison with positron emission tomography. Asian Pac J Cancer Prev 2014; 15: 4629–4635.

29.

Wang

Lin

Liu

. Intravoxel incoherent motion diffusion-weighted MR imaging in differentiation of lung cancer from obstructive lung consolidation: comparison and correlation with pharmacokinetic analysis from dynamic contrast-enhanced MR imaging. Eur Radiol 2014; 24: 1914–1922.

30.

Klauss

Lemke

Grunberg

. Intravoxel incoherent motion MRI for the differentiation between mass forming chronic pancreatitis and pancreatic carcinoma. Invest Radiol 2011; 46: 57–63.

31.

Yang

Chen

. First-pass perfusion imaging of solitary pulmonary nodules with 64-detector row CT: comparison of perfusion parameters of malignant and benign lesions. Br J Radiol 2010; 83: 785–790.

32.

Zhang

Jia

Zhang

. Intravoxel incoherent motion MRI: emerging applications for nasopharyngeal carcinoma at the primary site. Eur Radiol 2014; 24: 1998–2004.

33.

Lemke

Laun

Simon

. An in vivo verification of the intravoxel incoherent motion effect in diffusion-weighted imaging of the abdomen. Magn Reson Med 2010; 64: 1580–1585.

34.

Andreou

Koh

Collins

. Measurement reproducibility of perfusion fraction and pseudodiffusion coefficient derived by intravoxel incoherent motion diffusion-weighted MR imaging in normal liver and metastases. Eur Radiol 2013; 23: 428–434.