MRI of the lung (3/3)—current applications and future perspectives

Background MRI of the lung is recommended in a number of clinical indications. Having a non-radiation alternative is particularly attractive in children and young subjects, or pregnant women. Methods Provided there is sufficient expertise, magnetic resonance imaging (MRI) may be considered as the preferential modality in specific clinical conditions such as cystic fibrosis and acute pulmonary embolism, since additional functional information on respiratory mechanics and regional lung perfusion is provided. In other cases, such as tumours and pneumonia in children, lung MRI may be considered an alternative or adjunct to other modalities with at least similar diagnostic value. Results In interstitial lung disease, the clinical utility of MRI remains to be proven, but it could provide additional information that will be beneficial in research, or at some stage in clinical practice. Customised protocols for chest imaging combine fast breath-hold acquisitions from a “buffet” of sequences. Having introduced details of imaging protocols in previous articles, the aim of this manuscript is to discuss the advantages and limitations of lung MRI in current clinical practice. Conclusion New developments and future perspectives such as motion-compensated imaging with self-navigated sequences or fast Fourier decomposition MRI for non-contrast enhanced ventilation- and perfusion-weighted imaging of the lung are discussed. Main Messages • MRI evolves as a third lung imaging modality, combining morphological and functional information. • It may be considered first choice in cystic fibrosis and pulmonary embolism of young and pregnant patients. • In other cases (tumours, pneumonia in children), it is an alternative or adjunct to X-ray and CT. • In interstitial lung disease, it serves for research, but the clinical value remains to be proven. • New users are advised to make themselves familiar with the particular advantages and limitations.


Results
In interstitial lung disease, the clinical utility of MRI remains to be proven, but it could provide additional information that will be beneficial in research, or at some stage in clinical practice. Customised protocols for chest imaging combine fast breath-hold acquisitions from a "buffet" of sequences. Having introduced details of imaging protocols in previous articles, the aim of this manuscript is to discuss the advantages and limitations of lung MRI in current clinical practice. Conclusion New developments and future perspectives such as motion-compensated imaging with self-navigated sequences or fast Fourier decomposition MRI for noncontrast enhanced ventilation-and perfusion-weighted imaging of the lung are discussed.

Introduction
Magnetic resonance imaging (MRI) of the lung has been a challenge due to limitations such as low proton density in the lung and the fast signal decay due to susceptibility artefacts at air-tissue interfaces. Thanks to recent technical advances such as parallel imaging, shared echo-technique and rotating phase encoding, lung MRI can be recommended in a number of clinical indications [1].
The introduction into clinical routine is facilitated by customising comprehensive MR protocols that apply fast breath-hold acquisition techniques from a "buffet" of sequences that are optimised for chest imaging [2]. The basic imaging protocol comprises a non-contrast-enhanced protocol based on fast breath hold T1-and T2-weighted sequences to detect lung infiltrates, nodules or masses. Additional steady-state free precession sequence (SSFP) imaging can be performed with free breathing and is highly sensitive for detection of central pulmonary embolism, and provides information on respiratory mechanics [2,3]. Respiration-triggered T2-weighted sequences are available for uncooperative patients and those with breath-holding difficulties [4]. The sensitivity of this basic protocol for infiltrates and lung nodules is reported to be similar to CT [5][6][7]. Additional contrast-enhanced fat-saturated threedimensional gradient echo (3D-GRE) sequences are warranted for unclear masses, consolidations or pleural effusion detected in the basic protocol [8]. Three components are available for the assement of pulmonary vasculature and lung perfusion: An initial free-breathing unenhanced examination followed by dynamic contrast-enhanced perfusion imaging and a high-resolution angiogram [9].
With these customised protocols, lung MRI offers alternative solutions to routine diagnostic challenges, in particular for the imaging of the mediastinum. It also provides an alternative radiation-free diagnostic option that is especially relevant to young and pregnant patients, as well as subjects who need to undergo multiple investigations, e.g. for research purposes.
The details of the MR physics background as well as the protocol tree and its branches have been addressed in the two preceeding articles (citations 1/3 and 3/3). The aim of this paper is to discuss the advantages and limitations of lung MRI for a number of selected clinical applications and to outline current developments and future perspectives.
Clinical scenarios in which MRI might be considered a first choice modality

Cystic fibrosis
Cystic fibrosis (CF) lung disease is caused by mutations in the CFTR gene and remains one of the most frequent lethal inherited diseases in the Caucasian population. Due to the progress in therapy and management of CF lung disease in the past decades, the life expectancy of CF patients has increased substantially, with a current median survival of approximately 40 years and is expected to increase even further [10,11]. It is known that clinical parameters including spirometric pulmonary function testing (PFT) suffer from limited sensitivity and provide no regional information. With the advances in imaging in general and the ability to characterise and quantify CF in greater detail, imaging will likely play an increasing role in the improved understanding of the disease process and the progression of disease. Furthermore imaging will serve as a biomarker for the development of new treatments.. However, this also means that it becomes vital to reduce the overall (cumulative) radiation burden in this population, as this could lead to iatrogenic carcinogenesis [12].
Magnetic resonance imaging (MRI) is reported to be comparable to CT with regard to the detection of morphological changes in the CF lung [13][14][15]. At the same time MRI is superior to CT when it comes to the assessment of functional changes such as altered pulmonary perfusion [16]. Moreover, using the described MR protocols, it is possible to visualise bronchiectasis, bronchial wall thickening, mucus plugging, air fluid levels, consolidation and segmental consolidation and destruction [13], (Fig. 1a, b).
The accuracy of MRI in detecting bronchiectasis is dependent on a number of factors, including bronchial level and diameter, wall thickness, and the signal from within the bronchial wall and lumen. Central bronchi and bronchiectasis (central, peripheral) are well visualised on MRI, whereas normal peripheral bronchi starting at the 3rd to 4th generation are poorly visualised. The depiction of bronchial wall thickening depends on bronchial size and signal [13]. A high signal of the bronchial wall on T2-weighted (T2w) images represents increased fluid, i.e. oedema, possibly caused by active inflammation. Enhancement of the thickened bronchial wall on post-contrast, fat-suppressed T1weighted images is thought to be related to inflammatory activity. It is important to note that compared to MRI, CT can only detect wall thickening and is not able to comment on the cause [13].
Mucus plugging is well visualised on MRI even down to the small airways due to the high T2 signal of its fluid content. It is recognised as a high T2 signal filling of the bronchus along its course with branching in the periphery giving a grape-like or tree-in-bud appearance, respectively. As mucus plugs do not enhance, they are easily differentiated from bronchial wall thickening [13].
Bronchial air fluid levels are indicative of active infection, occurring in saccular or varicose bronchiectasis, and can be visualised by their high T2 signal. However, discriminating a bronchus with an air fluid level from one with partial mucus plugging or a severely thickened wall can be difficult. When evaluating the signal characteristics on T2-and T1-weighted images with and without contrast enhancement, air fluid levels can usually be differentiated.
Pulmonary consolidation in CF is mainly caused by alveolar filling with inflammatory material leading to a high signal on T2w images. Comparable to CT, MRI is able to visualise air bronchograms as low signal areas following the course of the bronchi within the consolidation [17,18]. With progression of the disease, complete destruction of lung segments or lobes can occur with similar appearances on MRI and CT.
Compared to CT, the strength of MRI is the additional assessment of "function", i.e. perfusion, pulmonary hemodynamics and ventilation. In CF, regional ventilatory defects cause changes in regional lung perfusion due to the hypoxic vasoconstriction response or tissue destruction. Using MRI, lung perfusion can be assessed by contrast-enhanced lung perfusion imaging [19]. Using contrast-enhanced 3D MRI, perfusion defects in 11 children with CF were reported to correlate well with the degree of tissue destruction [16]. Furthermore it was shown that at the age of 0-6 years lung perfusion changes were more prominent than morphological changes. However, establishing quantitative assessment tools for lung morphology is challenging for several reasons. First, signal intensities as derived from MRI are not calibrated as compared to CT. Second, the signal-to-noise ratio (SNR) in the lung is low and heterogeneous due to several physical circumstances [20]. Moreover, due to the lack of linearity between the MR signal and the concentration of applied contrast media, quantification of pulmonary perfusion using MRI is challenging [21]. Nevertheless, Risse et al. [22] described the importance of the qualitative assessment of the contrast time course component when analysing contrast-enhanced 3D MRI to categorise perfusion changes as normal, delayed, reduced, reduced and delayed as well as perfusion loss [22]. Using dedicated postprocessing tools, these data can be displayed in 3D [23], (Fig. 1c, d, e).
In addition to quantitative and qualitative scoring methods, clinical practice relies on visual assessment. It should be feasible to introduce an MR scoring system that is comparable to CT [24][25][26]. Up to now published studies either used a modified Brody [15] or an adapted Bhalla/Helbich score [14]. None of theses scores was tested for reproducibility. Besides morphology, functional parameters are important for a comprehensive diagnosis and have to be integrated into a dedicated MR score, also to generate an additional benefit over CT. A recently presented morpho-functional MRI score is easily applicable and reproducible for the semi-quantitative morphological and functional evaluation of a large severity spectrum of CF lung disease [27].
Based on the current state of affairs, perfusion MRI can be applied to monitor therapy and may be capable of differentiating between regions with reversible and irreversible disease. In contrast to CT [24-26, 28, 29] a dedicated scoring system as well as quantitative readouts for pulmonary MRI is lacking and will require development.

Acute pulmonary embolism in young or pregnant patients
The current imaging reference technique in evaluation of acute pulmonary embolism is helical computed tomography [30]. Its major advantage over ventilation and perfusion scintigraphy and SPECT are the availability and the comparably short acquisition time of the study with almost immediate delivery of the necessary information for patient care [30]. However, radiation exposure by CT is significant; therefore an alternative method for young patients and pregnant women would be appreciated. To be competitive with CT, an abbreviated MR protocol focusing on lung vessel imaging and lung perfusion may be accomplished within 15 min in-room time.
Although MR angiography has been demonstrated as an excellent tool in dedicated centres [31], more recent data from a large multicentre study suggest that the technique in isolation produced unsatisfactory results [32]. Therefore, combinations of different available MRI techniques for the detection of pulmonary embolism may be of better value [33]. This protocol was further modified and extended into a two-step algorithm [9]. As a first step, a steady-state GRE sequence acquired in two or three planes during free breathing would serve for an early detection of large central emboli within the first 5 min of the examination-according to the literature with a sensitivity of 90% and a specificity of close to 100% [33][34][35]. Any patient with a massive, central embolism detected at this point could be directly referred to intensive care and treatment; the time to diagnosis would be at least as short as with contrast-enhanced helical CT. If this first step of the examination produces a negative or unclear result, the protocol would be continued with the contrast-enhanced steps including first pass perfusion imaging, high spatial resolution contrast-enhanced (CE) MRA Coronal steady-state free precession images acquired during free breathing (a) and contrast-enhanced coronal 3d flash angiogram acquired in breathhold (b; embolus inside the right lower lobe artery circled); c series of subtracted images from the first pass perfusion study, perfusion deficits marked with open arrows at the image obtained at peak lung enhancement; 1.5-T MRI scanner and a final acquisition with a volumetric interpolated 3D FLASH sequence in transverse orientation (Fig. 2). Despite its composition of multiple sequences, the two-step examination could be completed within 15 min in-room time, which makes it feasible as a quick test for daily clinical routine. In many cases, such as in pregnant woman, when administration of contrast material or radiation exposure is contra-indicated, the examination can be limited to the first step, the free breathing or breathhold acquisition of steadystate GRE sequences alone. Furthermore, since these steps are partially redundant, at least one acquisition would be expected to be diagnostic even in non-compliant patients.
Lung MRI as an alternative or adjunct to other modalities

Central mass with atelectasis and pleural effusion
Undoubtedly, contrast-enhanced multiple detector row computed tomography is the method of first choice in imaging thoracic malignancies. MRI is considered as an alternative method, e.g. when the application of iodinated contrast media is contraindicated. For this purpose, MRI with dedicated standard protocols can provide a comprehensive morphologic TNM evaluation [36]. A contrast-enhanced examination can be achieved within 25 min in-room time. Intra-pulmonary masses larger than the clinically relevant size of 4-5 mm in diameter can be easily detected. The extent of mediastinal, hilar and supraclavicular lymph node enlargment can be assessed with excellent soft tissue contrast. Metastatic disease involving the liver, the adrenal glands and the skeleton of the thorax are fully covered. The feasibility of extending the examination to whole-body staging with comparable results as achieved by PET/CT has been demonstrated [37][38][39][40]. The only limitation compared to CT is the detection of nodules smaller than the clinically relevant size of 4-5 mm. Beyond being just a surrogate for a CT scan in some cases, MRI can offer additional advantages. In large pulmonary masses, the excellent soft tissue contrast of MRI allows for the distinction of tumour from atelectasis and pleural effusion, e.g. for image-guided radiotherapy planning. Administration of T1-shortening contrast material specifically contributes to detecting tumour necrosis, chest wall or mediastinal invasion, and pleural reaction/carcinomatosis [8] (Fig. 3). Furthermore, MRI contributes comprehensive functional information on respiratory mechanics, tumour mobility [41] and lung perfusion [42,43]. The clinical value of complementing the purely morphologic staging by imaging of perfusion and tumour motion in specific clinical settings and situations has been demonstrated and is subject to further investigation.

Pneumonia in young patients
The potential of MRI to replace chest radiography, particularly in very young children, was already investigated several years ago [18,[44][45][46]. Much of this work was conducted on low-field MRI (mainly 0.2-T scanners) using steady-state free precession sequences [47]. On average, three thick slices are acquired in coronal orientation with a mean breathhold time of 4-5 s. However, nowadays, only a few institutions regularly use low-field lung MRI in paediatric radiology. Nevertheless, the experience from this work may be considered valid for the suggested protocols for 1.5-T scanners since image quality has significantly improved. Therefore, T2-weighted fat-suppressed as well as dynamic contrast-enhanced T1-GRE sequences are applied with a slice thickness between 5 and 6 mm [48]. Disease entities encompassing community-acquired pneumonia, empyema, fungal infections and chronic bronchitis are detectable [49] (Fig. 4).
Recent studies demonstrated the feasibility of chest examinations on 3-T high-field MRI. ECG triggering seems to be mandatory to reduce pulsation artefacts. One study used the navigator techniques to reduce breathing artefacts. High-field chest MRI may allow differentiation between inflammation-and fibrosis-predominant lesions in UIP and NSIP in adult patients [50]. Moreover, a recent comparison between 3-T lung MRI and HRCT as gold standard showed an excellent correlation with non-cystic fibrosis chronic lung disease in children [51]. Breathhold T2-as well as T1-weighted sequences with ECG triggering were acquired. In summary, lung MRI may prove to become a valuable tool for detection as well as characterisation of inflammatory lung disease in children.

Clinical scenarios in which CT remains the first choice modality
Emphysema/COPD Chronic obstructive pulmonary disease (COPD) is one of the leading causes of morbidity and mortality worldwide. At present it is the fourth most common cause of death among adults, but its prevalence is increasing [52]. COPD is characterised by incompletely reversible airflow obstruction due to a mixture of airway obstruction (obstructive bronchiolitis) and parenchymal destruction (emphysema) [52]. Severity of COPD is clinically assessed by lung function tests and diffusion capacity for carbon monoxide. CT is considered the reference standard for imaging COPD-related morphological changes. Imaging COPD with proton MRI is a major challenge due to the loss of lung tissue and reduced blood volume due to hypoxic vasoconstriction combined with hyperinflation, all resulting in a marked reduction of lung parenchymal signal [53]. The strength of MRI for imaging COPD lies with the assessment of functional parameters like perfusion and respiratory dynamics [54].
In COPD emphysematous destruction is difficult to diagnose because of a loss of MR signal. Hyperinflation severely affects diaphragmatic geometry with subsequent reduction of the mechanical properties, while the accessory neck and rib muscles become less effective [55]. The common clinical measurements of COPD do not provide insights into how structural alterations in the lung lead to dysfunction in the breathing mechanics, although treatments such as lung volume reduction surgery (LVRS) are thought to improve lung function by facilitating breathing mechanics and increasing elastic recoil [56].
In contrast to normal subjects with regular, synchronous diaphragm and chest wall motion, patients with emphysema frequently have reduced, irregular or asynchronous motion, with a significant decrease in the maximum amplitude and  [57]. In some patients the diaphragm movement is not coordinated (e.g. the ventral portion of the hemidiaphragm moves inferiorly while the dorsal part moves cranially [58]), while paradoxical diaphragmatic motion correlated with mild and moderate hyperinflation [59].
One study demonstrated a correlation between the change of parenchymal signal intensity measured by MRI at inspiration and expiration and FEV 1 (r00.508) as a predictor of airflow obstruction [60].
Several studies have shown that airway obstruction in patients with COPD tends to be located in airways smaller than 2 mm internal diameter [61]. These airways are located between the 4th and the 14th generation of the tracheobronchial tree. Severe peripheral airflow obstruction also affects the proximal airways from subsegmental bronchi to trachea. For the assessment of tracheal instability, such as seen in tracheobronchial malacia (which may mimick the clinical appearance of small airways disease), MR cine acquisitions during continuous respiration or forced expiration can be recommended [62]. The depiction of airway dimensions and size of the airway walls by MRI in physiological condition is limited to the central bronchial tree. For the depiction of bronchiectasis, sequences with a high spatial resolution are essential (see section on CF). The previously described 3D volume interpolated gradient echo sequence (VIBE) offers a sufficient spatial resolution with a sensitivity of 79% and a specificity of 98% regarding visual depiction of bronchiectasis compared to CT [3].
Gas exchange in the lungs is optimally maintained by matching of ventilation and perfusion. In patients with COPD, ventilation is impaired due to airway obstruction and parenchymal destruction. In regions with reduced ventilation, hypoxic vasoconstriction occurs [63,64] causing a reduction of local pulmonary blood flow with redistribution to better ventilated lung regions [65,66]. The reduction of the pulmonary vascular bed is related to the severity of parenchymal destruction; however the distribution of perfusion does not necessarily match parenchymal destruction [67,68]. Conventional radionuclide perfusion scintigraphy has been used to assess these abnormalities, but it has substantial limitations with respect to spatial and temporal resolution.
MR perfusion allows for a high diagnostic accuracy in detecting perfusion abnormalities of the lung [69,70]. Additionally, MR perfusion ratios correlate well with radionuclide perfusion scintigraphy ratios [71,72]. Lobar and segmental analysis of the perfusion defects can be performed [68]. Perfusion abnormalities in COPD clearly differ from those caused by vascular obstruction. While wedgeshaped perfusion defects occur in embolic obstruction, a generally low degree of inhomogeneous contrast enhancement is found in COPD with emphysema [73]. Furthermore, the peak signal intensity is reduced. These features allow for easy visual differentiation and compare well with work done using CT perfusion experiments [74]. In patients with COPD the quantitative evaluation of 3D perfusion showed that the mean pulmonary blood flow (PBF), pulmonary blood volume (PBV) and mean transit time (MTT) are diffusely decreased and the changes are heterogeneous [75]. Calculated mean PBF and PBV are significantly decreased, and MTT is significantly shortened [76].

Interstitial lung disease
Interstitial lung disease (ILD) encompasses numerous pathologic disorders of different etiologies, generally manifesting with an inflammatory reaction known as "alveolitis", which may progress towards fibrosis. Because the nature of these disorders is highly heterogeneous, imaging findings alone are often insufficient for making the final diagnosis, and integration of morphologic aspects with clinical and functional data is required. In the last 3 decades, computed tomography (CT) has clarified the elementary alterations and morphologic patterns characterising the infiltrative changes of ILD. In contrast, MRI has only recently overcome many of the technical issues related to lung imaging, providing a standardised image quality, which in many instances is now comparable to CT. This partly explains the relatively limited number of MRI studies that have been clinically performed in ILD patients. Nonetheless, published data suggest at least three possible applications for lung MRI in ILD: (1) visualisation and recognition of morphological changes and their patterns, (2) assessment of the inflammatory activity of the disease and (3) effects of lung morphologic changes on functional parameters such as contrast enhancement and perfusion.
The essential morphologic findings in ILD include airspace disease, interstitial abnormalities or a combination of the two. Because MR signal increases proportionally to proton density, air-space infiltrates appear on the T2-weighted images as hyperintense areas against the dark background of the normal lung parenchyma. When pulmonary vascular markings are not obscured, these areas can be assimilated to the ground-glass opacities detected by CT [17,77]. More dense opacities appear as consolidations, which can be easily assessed by MRI [78]. Similar to consolidations, interstitial abnormalities increase signal intensity presenting with curvilinear bands, nodules and reticulations, which can be associated to a variable degree of parenchymal distortion [50,79,80]. Fibrotic changes that extensively involve both peripheral and perihilar portions of the lung are generally well demonstrated on T2-weighted images, albeit that one needs to consider extracellular interstitial water as a potential differential diagnosis in patients with suspected congestive heart failure. Subtle changes in the subpleural regions may become more difficult to visualise, notably when parenchymal distortion is not present, demonstrating the superiority of CT in this respect. T1-weighted VIBE images offer higher spatial resolution, and post-contrast acquisition with fat-suppression is recommended to increase the signal of altered subpleural lung tissue against a background represented by chest wall muscles, ribs and normal lung parenchyma. Honeycombing, which manifests with reticular changes and irregular cystic transformation of the lung, can also be assessed using this technique [50] (Figs. 5, 6, 7 and 8).
Differentiation of active inflammation from fibrosis is of significant clinical importance both for the prediction of therapy response and clinical outcome of ILD. Both MR signal and contrast-enhancement characteristics of inflammation and fibrosis have been investigated. Although initial studies performed on 1.5 T lacked sufficient image quality [81][82][83], the feasibility of the assessment of disease activity in ILD was demonstrated. Only recently, 3.0-T MRI has increased sensitivity to changes in proton density. In particular, Yi et al. [50] reported that MR signal of inflammatory and fibrotic lesions on T2-weighted images is hyperintense and isointense, respectively, compared to the signal from chest wall muscle, indicating an increased water content in the areas of inflammation.
Dynamic MRI using iv contrast administration also indicated that early enhancement and washout with discernible peak enhancement at 1 or 3 min after contrast injection was associated with positive and negative prediction values of 82 and 92%, respectively, in predicting disease activity [50]. The earlier enhancement and rapid washout would be in agreement with higher permeability of capillaries in the areas of inflammation compared to those of fibrosis.
A different approach to differentiating active inflammation from fibrosis was attempted in a recent study in a bleomycininduced lung injury model in rats [84]. Proton density and T2 relaxation were computed regionally in the injured lungs, and MR-derived parameters were compared to postmortem measures of water and collagen content. The authors concluded that proton density and T2 relaxation data acquired using MRI were sensitive to inflammation and fibrotic changes in the lung. Although they were able to distinguish diseased lungs as effectively as postmortem measurements, they were unable to differentiate between fibrosis and inflammation. In conclusion, these data are encouraging and support potential future applications of MRI in interstitial lung disease both in research and clinical settings.

Future perspectives
Significant efforts have been made to further improve robustness and reproducibility of lung MR image quality. Compared with x-ray and CT, the quality of lung MRI is more dependent on the ability of patients to follow breath hold instructions. Advanced acquisition schemes with inherent correction of respiratory motion and cardiac pulsation are therefore important for future developments. For example, a conjoint research group of medical physics and radiology departments supported by the German Research Foundation (Deutsche Forschungsgemeinschaft) is currently working on the development of self-navigated sequence designs and radial k-space methods for the assessment of lung morphology on free breathing. Similarly, a group of One way to improve the robustness of lung MRI against respiratory motion is to implement self-navigation. The prototype 3D-MRI sequence (a self-navigated T1-weighted 3D flash with quasi-random k-space ordering) acquires multiple full lung volumes during free breathing, which results in a set of images with unsharp delineation of structures that are subject to respiratory motion. With each acquisition nonspatially encoded DC signals are acquired at the center of kspace to be used as navigator. This signal contains sufficient information to detect motion and to select image information of the different acquisitions to either produce one motion corrected data set for morphologic imaging without patient compliance or to perform a detailed motion analysis [85]. Another approach is based on radial imaging with k-space weighted image contrast (KWIC). Again, redundant data are acquired over multiple respiratory cycles. Motion correction is achieved by radial data acquisition with extraction of the signal from k-space centre for the determination of the respiratory cycle. The views are then grouped and images are reconstructed for each respiratory phase. Further improvements of image quality are achieved with autofocusing, 3D image correlation, Kspace-weighted image contrast (KWIC) and principal component analysis [86][87][88]. Fibrosis associated with rounded consolidation. a) Subpleural reticular changes are visualised at the periphery of the lungs (thin arrows). b) After contrast administration the subtle linear enhancement at the pulmonary-chest wall interface indicates abnormal findings related to subpleural fibrosis (thin arrows). A rounded consolidation is present on the left in the lingula suspected for lung tumour in ILD (asterisk) Fig. 8 Bilateral hilar and mediastinal adenomegalies in sarcoidosis. Node enlargement (arrows) is demonstrated with gradient echo images before (a) and after administration of contrast material (b). Coronal perfusion images indicate vascular compression at the right hilum (arrow, c) and a wedge-shaped perfusion defect (asterisk, d) Fig. 9 Twenty-three year-old female with acute pulmonary embolism at the time point of diagnosis (a, b) and at follow-up study after 6 months (c, d). The initial dynamic contrast enhanced (DCE) study (a) as well as the perfusion-weighted Fourier-decomposition (FD) image (b) demonstrate multiple perfusion defects (open arrows). In the follow up study, both techniques (DCE; c and FD; d) demonstrate an almost homogeneous lung perfusion after effective anticoagulation Non-contrast-enhanced ventilation-perfusion scanning-poor man's "have it all"?
One of the latest developments in the field of proton-based lung MRI appears to be a very promising technology for non-contrast-enhanced ventilation and perfusion scanning. This novel approach, known as Fourier decomposition MRI [89], utilises a short echo dynamic SSFP acquisition of lung images with subsequent compensation for respiratory motion by using nonrigid image registration [90]. Spectral analysis of the image time series allows for identification of peaks at the respiratory and cardiac frequencies. Amplitude of these peaks is related to regional proton density change caused by deformation of lung parenchyma (highest signal with lowest pulmonary air content in expiration) and pulmonary blood flow (lowest signal with maximum blood flow in systole) [91]. Further image post-processing produces ventilation-and perfusion-weighted maps for regional assessment of lung function from a single acquisition series. However, the quantitative validation remains subject to further investigation ( Fig. 9) [92]. Nevertheless, there is a perspective that the method of choice for morphologic and functional assessment of acute pulmonary embolism in the very near future might be a non-contrast-enhanced free breathing MR scan of 10-15 min.

Hybrid PET/MRI
Hybrid PET/MRI has recently become available for clinical research. Compared to PET/CT, PET/MRI may be advantageous due to higher soft tissue contrast, while there is a slight reduction in ionising radiation dose [93]. Technically, integration of the two systems has been a significant challenge and required substantial modifications [94][95][96]. The size of the PET detectors had to be minimised; a PET detector that is insensitive to high magnetic fields had to be developed, and the adverse effect of PET detector parts on the homogeneity of the magnet's B0 field must be minimised. Moreover, interference between the radiofrequency signals, MRI gradients and PET electronic signals must be avoided [97]. Some of these problems were overcome by application of optical fibres and advances in gamma ray detector technology, which were initiated mainly by the advent of avalanche photodiodes; in addition the routine availability of fast scintillation materials resulted in the development of fully magnetic-field-insensitive high-performance PET detectors [98]. More recently, a truly integrated 3-T MRI/PET system was launched.
Although PET/MRI may not necessarily replace the role of PET/CT in thoracic oncological imaging [99], and specific clinical indications remain to be identified, MRI may be advantageous when compared to CT in the investigation of consolidating lung lesions, and malignant mediastinal and chest wall invasion. The additional diagnostic value of PET/MRI over PET/CT on nodal staging is questionable since both MRI and CT nodal staging is size based [100]. MRI, however, has been reported to be of higher accuracy than PET/CT when assessing the brain, liver and the bone for distant metastases [101]. Oncological research may be another potential area where multiple follow-up functional and anatomical imaging by PET/MRI may be advantageous over PET/CT. Lastly, with the increasing availability of radiotracers and novel compounds for molecular and physiological assessment, including labelling of target cells and compounds that target particular cell lines or processes, the combination of these powerful modalities may result in significant advances for research (and potentially clinical) purposes.

Conclusions
MRI is emerging as a valuable lung imaging modality, together with x-ray and CT. It offers a unique combination of morphological and functional information in a single examination without any radiation burden to the patient. However, although "push-button" protocols facilitate its clinical application, lung MRI can be still challenging, being the most comprehensive but also most expensive and least robust of the three modalities. New users are advised to make themselves familiar with the particular advantages and limitations of the technique and its diagnostic scope to appreciate its potential benefits. Given this, lung MRI will be increasingly used and even further improved by additional recent and future developments, in particular in the fields of motion compensation and functional imaging.