Imaging of Bronchial Carcinoma 42

(1)

S. O. Schoenberg, MD; C. Fink, MD; B. Fischer

Department of Clinical Radiology, University Hospitals – Grosshadern, Ludwig Maximilian University of Munich, Marchioninistr. 15, 81377 Munich, Germany

C O N T E N T S

42.1 Epidemiology 471 42.2 Radiological Staging 471 42.2.1 T Stage 471

42.2.2 N Stage 472 42.2.3 M Stage 473

42.3 Role of Parallel-Imaging Techniques 474 42.3.1 Increase of Temporal and

Spatial Resolution 474

42.3.2 Reduction of Signal Decay and Blurring in Single-Shot Acquisitions 476 42.3.3 Reduction of Acquisition Time 476 42.3.4 Increase of Anatomic Coverage 479 42.4 Conclusion 479

References 479

Imaging of Bronchial Carcinoma 42

Stefan O. Schoenberg, Christian Fink and Bastian Fischer

42.1

Epidemiology

Non-small-cell lung cancer (NSCLC) is still the lead- ing cause of death from a malignancy in men, with a rising incidence in women. In the past decade, the therapeutic options for advanced stages of NSCLC have been substantially improved, including new chemotherapeutic and anti-angiogenic agents for ﬁ rst- and second-line medical treatment, new pro- tocols for neo-adjuvant and adjuvant chemotherapy as well as improved techniques for surgical resection of centrally growing tumors. These developments

require improved non-invasive imaging modalities for staging of resectability and operability in order to select the adequate treatment options for the individual patient.

42.2

Radiological Staging

The staging of non-small-cell lung cancer has to be performed in an interdisciplinary approach consid- ering all clinical, radiological, nuclear-medicine, and histological results. The radiological staging is done according to the TNM classiﬁ cation with T describing the extent of the primary tumor, N the presence and location of metastatic lymph nodes and M the presence or absence of distant metastases. It is important to remember that the individual stages of the TNM classiﬁ cation have undergone numerous revisions and thus need to be considered in their most recent version (Mountain 1997; Mountain and Dresler 1997). For the radiologist, it is also important to know which therapy the patient is possibly undergoing in order to optimize the imaging strategy. Spiral CT is currently viewed as the backbone of radiological staging. With the new 16- to 64-channel generation of multi-slice CT systems, the entire thorax can be scanned in less the 10 s with 1-mm slice thickness.

42.2.1 T Stage

For the assessment of the tumor extension, special attention has to be drawn to the assessment of chest- wall and mediastinal invasion, since these ﬁ ndings directly affect the tumor stage. The only reliable CT sign is a mass in the chest wall or the presence of rib destruction (Libshitz 1990; Pearlberg 1987).

However, these signs have a poor sensitivity of only

(2)

20% with moderate specifi city. The accuracy can be moderately increased if other fi ndings such as an obtuse angle with the thoracic wall, thickening of the underlying pleura, and presence of chest pain are also considered (Scott et al. 1998). Highest accuracies are found in dynamic expiratory scans that demonstrate a decreased or absent mobility of the ingrown tumor (Murata et al. 1994); however, this technique has not gained wide-spread use in the clinical routine. Due to the oblique course of the left and right main stem bronchus, delineation of an infi ltration of the carina is diffi cult to assess on standard spiral CT with 5-mm- thick slices. Multi-slice CT has advantages compared to standard spiral CT since the course of the entire bronchial tree can be delineated on three-dimensional reconstructions of thin-slice images (Grandy 2001). Multiplanar reconstructions also reduce errors from partial volume effects, particularly in the apex of the lung where the convex surface of the lungs might erroneously cause an impression of superior sulcus infi ltration.

Accuracy for the assessment of mediastinal invasion by CT, however, is limited. Defi nite proof of tracheal invasion is only present in the case of intralu- minal or circumferential tumor growth, while the sole contact between the tumor and the trachea does not suffi ce to make the diagnosis of tracheal invasion. In these cases, mediastinoscopy or endotracheal sonog- raphy is necessary to make the diagnosis (Herth and Becker 2000). While tumors with partial encasement can usually be considered resectable, the assessment of the mediastinal fat planes as a criterion for operability is less reliable (Herman et al. 1994) with both false-positive and false-negative fi ndings.

Due to the superior tissue contrast of magnetic resonance imaging, vessels can be delineated from surrounding mediastinal structures without any contrast agents as a result of the black-blood effect.

In case of slow fl ow, however, infl owing blood might still contribute to the signal within the plane, giving the false impression of a thrombosed vessel or solid structure such as a lymph node. With the introduction of fast time-resolved three-dimensional contrast-enhanced MR angiography (3D-CE-MRA), these limitations can be overcome. Several studies have demonstrated the superiority of MRI for delineation of the extrapleural fat plane and detection of tumorous chest wall invasion, particularly on coronal scans (Manfredi et al. 1996). Sensitivity and specifi - city reach 90% (Padovani et al. 1993). MR imaging should be always performed for tumors with superior sulcus invasion. Invasion of pulmonary arteries

and veins is well visualized on the individual 3D-CE- MRA scans (Schoenberg et al. 1998). Sometimes, atelectasis can be differentiated from the primary tumor by the different signal intensity on MR images (Kauczor and Kreitner 1999). These advantages are of importance since nowadays tumors with limited invasion of the pericardium, the mediastinal pleura and fat as well as inﬁ ltration of the vagal and phrenic nerve routinely undergo resection. Even advanced tumors with limited invasion of the left atrium or even the carina can be resected in selective cases (Mitchell et al. 1999; Shirakusa et al. 1998).

However, the so-called desmoplastic reaction caused by tumor-induced proliferation of benign connective tissue adjacent to the tumor can result in an overes- timation of the stage of the tumor (Webb et al. 1991).

MR is advantageous for the delineation of the desmoplastic reaction due to the larger differences in signal intensity between the tumor and adjacent connective tissue. On the other hand, MR is inferior to CT for the assessment of endo-bronchial tumor growth due to the poorer spatial resolution and the signal loss in air-containing structures.

In conclusion, the sole use of CT alone is not suffi cient for the accurate staging of borderline fi ndings such as tracheal or mediastinal infi ltration. In any case, multi-slice CT should be used to reformat the data set in multiple planes for the most accurate determination of the single longest tumor diameter according to the RECIST criteria. Compared to CT, MRI has various advantages for assessment of the T stage. Nevertheless, so far detection of mediastinal invasion has not proved to be superior compared to CT despite the better soft tissue contrast (Manfredi et al. 1996; Webb et al. 1985).

42.2.2 N Stage

Staging of lymph nodes still remains a major chal- lenge to cross-sectional imaging since only 10% to 15% of all patients are found to be stage I, and thus do not reveal any metastatic lymph nodes. Complete resection of these nodes improves the prognosis of the patient (Keller et al. 2000). The key to the accurate staging and resection of lymph nodes is the correct localization of theses nodes along anatomic structures (Mountain and Dresler 1997). Stand- ardized lymph nodes maps should be used to cor- relate the location of an enlarged lymph node in CT, mediastinoscopy and bronchoscopy; however, not all

(3)

enlarged lymph nodes seen on CT can be reached by mediastinoscopy or bronchoscopy. Accuracy of lymph node staging with CT is only moderate with substantial differences in the reported accuracies ranging from 40% to 90% and high interobserver variability (Kiyono et al. 1988). The upper limit of normal lymph node size varies with their position between 10 mm and 12 mm (Guyatt et al. 1995). In addition, metastatic spread can skip regional lymph nodes, thus giving the wrong impression of an N0 stage. The main problem arises from the fact that the criterion of lymph node size has only limited value for the assessment of lymph node metastases. How- ever, even nodes smaller than 10 mm may be affected by metastatic spread. On the other hand, lymph nodes larger than 10 mm may be infl ammatory. Thus, the cut-off value of 10 mm for differentiation between normal and metastatic nodes has only an accuracy of 70% to 80% (Dales et al. 1990; Kernstine et al. 1999). With an increasing stage of the primary tumor, the sensitivity for detection of enlarged lymph nodes increases, while the specifi city decreases due to benign infl ammatory nodes. The negative predictive value of CT for the presence of enlarged lymph nodes is high, thus the predominant role of CT is to exclude metastatic nodes and to confi rm an N0 stage. A negative CT scan avoids mediastinoscopy with similar results at surgery (Daly et al. 1993). Another estab- lished role is the confi rmation of massively enlarged lymph nodes in the N3 position, which usually is considered inoperable.

So far, MR is considered equal to CT for the staging of lymph nodes despite the fact, that different criteria other than size were assessed for differentiation of benign from metastatic nodes. Differentiation by means of different T1 relaxation times or gadolinium-chelate enhancement had only moderate success (Laissy et al. 1994). MR-compatible superparamagnetic iron-oxide contrast agents (USPIO = ultrasmall superparamagnetic iron oxide) are cleared from the blood by normal lymph nodes with subsequent signal loss in the node. Metastatic nodes show a much smaller signal loss due to fewer uptakes. Although remarkable results with accuracy of 85% have been reported in one study, the same authors reported only a sensitivity of less than 70% for nodes smaller than 15 mm due to the limited spatial resolution of the sequences used (Nguyen et al. 1999). Other authors found lower speciﬁ cities due to overlap of the signal characteristics between benign and metastatic nodes (Pannu et al. 2000). Therefore, at the moment, particularly marginally enlarged lymph nodes in the N3-

position need to be conﬁ rmed by mediastinoscopy or positron-emission tomography (PET), respectively PET-CT. Several studies have reported sensitivities and speciﬁ cities for PET exceeding 85% (Schmidt et al. 2004; Wahl et al. 1994). If the functional informa- tion of PET is combined with the morphologic infor- mation from CT, the overall accuracy can be further increased to over 90% (Vansteenkiste et al. 1998).

This trend is further enhanced by modern PET-CT systems, combining even multi-ring PET and multi- slice CT within one single scanner. Due to the high negative predictive value of over 95%, some authors advocate abandoning the need for mediastinoscopy in cases of a normal PET.

42.2.3 M Stage

Staging of distant metastases is of pivotal importance for adequate patient management since 40% of all patients with newly diagnosed bronchial carcinoma already have distant metastases (Boring et al. 1992), 25% of which are outside the thorax, predominately hepatic, adrenal, osseous and intracranial (Sider and Horejs 1988). Patients with stage-T3 tumors, patients with metastatic nodes in N2-position as well as patients with adenocarcinoma and poorly differentiated carcinomas have a higher risk of distant metastases even in the face of negative lymph nodes (Sider and Horejs 1988; Salvatierra et al. 1990).

Another indication for extra-thoracic tumor staging holds true for patients with a high morbidity from other underlying diseases for whom the presence of distant metastases would avoid the risk from peri- operative mortality. Due to the speed of multi-slice CT, local staging of the primary tumor is often completed with imaging of the entire abdomen, pelvis, and cranium using a slice thickness of less than of equal to 3 mm.

For the individual organs such as the adrenal glands, CNS, and liver, MRI is already considered superior to CT. Special fat-suppressing techniques, so-called in- phase/out-of-phase imaging, allow the subtraction of fatty components within the organ tissue. This has been successfully used to differentiate benign, fat-containing adrenal adenomas from carcinomas (Reinig et al. 1994). MR is the modality of choice to exclude intracranial metastases and is more sensitive than CT.

This is particularly important for stage-IIIA patients where even the smallest brain metastases need to be excluded before a risky operation is performed. Opti-

(4)

mized imaging strategies with sensitive MR sequences such as fat-suppressed short-tau inversion-recovery (STIR) techniques allow a whole-body staging with equal results compared to standard, clinically used staging procedures that involve multiple modalities (Schmidt et al. 2004). For staging of metastasis, MRI is in competition to FDG-PET, for which the results are extremely good. Prospectively designed studies could show a correct change of the M-stage in 10% of all cases compared to a conventional metastasis search, thus directly inﬂ uencing therapeutic management (Bury et al. 1996). Unexpected metastases were found in 10%

of cases, whereas 10% to 20% of false-positive CT ﬁ ndings were down-staged to M0 (Dietlein et al. 2000;

Valk et al. 1995). This change of therapeutic management was found to be cost-effective (Dietlein et al.

2000). In the future, whole-body MRI maybe a com- petitive technique for systemic metastases screening with fewer upfront costs (Schmidt et al. 2004, 2004).

Recent data with dedicated whole-body MRI scanners report almost equal results for metastases screening compared to PET-CT (Schmidt et al. 2004; Antoch et al. 2003).

42.3

Role of Parallel-Imaging Techniques

Image quality of MRI used to be degraded by poorer spatial resolution and long scan times resulting in artefacts from respiratory and cardiac motion. In addition, the scan set-up and re-positioning of the patient in dedicated coils was time-intensive in the past. The fl exibility of the MRI systems has changed dramatically within the last 5 years. Techniques of cardiac and respiratory gating can be now routinely applied. Multi-channel MRI systems featuring numerous fl exible surface coil systems combined with auto- mated table movement allow covering large areas of the body with maximum signal-to-noise ratio. Faster gradient systems reduce three-dimensional acquisitions to a single breath-hold while maintaining high spatial resolution. All these advances are further enhanced with the introduction of parallel-imaging techniques. In particular, this refers to increases in temporal and spatial resolution, reduction of signal decay from air-fi lled structures, decrease of blurring in single-shot acquisitions, and reduction of overall acquisition time. In addition, anatomic coverage can be substantially expanded.

42.3.1

Increase of Temporal and Spatial Resolution

During bolus-injection of gadolinium chelates, the pulmonary arteries and veins can be assessed along with the pulmonary perfusion and the systemic vas- cular supply of the lung, a technique also known as multiphasic time-resolved 3D contrast-enhanced MR angiography (TR-3D-CE-MRA). Compression of vessels or tumor invasion is also reliably seen (Schoenberg et al. 1998). Analysis of the time- resolved 3D MRA data sets allows detecting perfusion defects. In addition, tumor feeding vessels arising from systemic vessels such as bronchial arteries are also seen in later phases of the contrast media transit. The use of parallel imaging allows increasing temporal resolution while preserving spatial resolution. The scan time for an individual time frame of TR-3D-CE-MRA can be decreased to 1.2 s despite a spatial resolution 1.6×3.0×4.0 mm³ if a parallel- imaging acceleration factor of R=2 in combination with the GRAPPA algorithm is applied (Nikolaou et al. 2004). Even higher temporal and spatial resolution can be achieved when parallel imaging is combined with view-sharing techniques, a technique known as TREAT (time-resolved echo-shared angiographic technique) (Fink et al. 2005).

Besides separate visualization of arteries and veins, qualitative as well as quantitative assessment of pulmonary perfusion is feasible. For qualitative assessment, the TR-3D-CE-MRA data sets can be evaluated for segmental and lobar perfusion defects. In addition, the ratio of the relative perfusion between the lung affected by the tumor and the contralateral side can be visually assessed. Data from one study shows that perfusion defi cits resulting from tumor infi ltration of the lobar or segmental pulmonary arteries can be identifi ed with good correlation to perfusion scintigraphy (Fink et al. 2004). This was confi rmed by our own results from a pilot study on patients with central NSCLC (Schoenberg et al. 2005). In addition, tumor infi ltration of the lung can be differentiated from atelectasis as the latter reveals a delayed hyper-enhancement on the time-resolved images (Fig. 42.1).

Quantitative perfusion can be calculated based on a single-compartment model. One study of healthy volunteers yielded mean values for the upper lung zones of 7.1±2.3 ml/100 ml for regional blood volume (RBV) and 197±97 ml/min/100 ml for regional blood ﬂ ow (RBF) and for lower lung zones, 12.5±3.9 ml/100 ml for RBV and 382±111 ml/min/

(5)

Fig. 42.1. A 43-year-old male patient with NSCLC of the right lower lobe. High-resolution 3D-CE-MRA (upper row) with par- allel imaging (GRAPPA, acceleration factor R=2) and a spatial resolution of 1.2×0.9×1.2 mm³ shows infi ltration of the lower lobe pulmonary arteries and veins. Time-resolved 3D-CE- MRA (middle row) reveals absence of pulmonary perfusion in the corresponding location with delayed hyper-enhancement of the atelectasis. Cardiac-gated cine TrueFISP sequences iden- tify limited focal invasion of the right atrium (arrows) without signs of infi ltration of the inferior vena cava. The diagnosis of a resectable stage-IIIB tumor was made (T4N2M0), which was confi rmed by thoracic surgery

(6)

100 ml for RBF (Nikolaou et al. 2004). These results are in good agreement with reference data from the literature. However, so far, these data have only been veriﬁ ed in volunteer studies.

In regard to high-resolution 3D-CE-MRA, a further gain in spatial resolution is possible with the use of parallel imaging. By application of GRAPPA with an acceleration factor of R=3, the entire pulmonary vasculature can be imaged with a spatial resolution close to 1 mm³ within a single breath-hold of approx- imately 20 s (Fig. 42.2).

42.3.2

Reduction of Signal Decay and Blurring in Single-Shot Acquisitions

In theory, single-shot turbo-spin-echo sequences such as HASTE are well suited for imaging the lung since the acquisition of a single slice is completed within a scan time of less than 1 s. This substantially reduces artefacts from cardiac motion or incom- plete breath-holds. A disadvantage of this type of sequence is the presence of blurring as a result of long echo trains and off-resonances, which reduces the anatomic delineation of mediastinal structures.

In addition, the signal decay due to short T2* relaxation times still substantially reduces the visualization of pulmonary parenchyma. Shortening of the echo trains by parallel imaging improves these limitations to a large degree (Fig. 42.2); cf. also Chap. 10. Small satellite metastases with a size of more than 8 mm can be reliably detected (Fig. 42.3). It is important to locate these metastases on axial, sagittal, and coronal images in order to differentiate lesions within the same lobe (tumor stage T4) from those in abutting lobes (tumor stage M1). Small atelectases resulting from proximal tumor occlusion of bronchi can also be reliably detected.

42.3.3

Reduction of Acquisition Time

Besides assessment of perfusion, MRI is capable of determining pulmonary ventilation. However, this initially required the technically demanding hyper- polarization of helium-3 with the drawback of limited availability (Bock 1997). A much easier promising approach is the use of oxygen-enhanced MR imaging as a surrogate for assessment of ventilation and diffusion (Lofﬂ er et al. 2000). So far, the use of this technique was time-intensive, technically demanding and not robust enough for clinical application since the complex sequence scheme required low heart rates and long breathing cycles for adequate cardiac and respiratory gating. With the implementation of parallel imaging, the time for read-out of the MR signal can be substantially decreased, allowing the application of this technique also in patients with fast heart rates and short end-expiratory plateaus (Dietrich et al. 2005);

cf. Chap. 38. Although oxygen-enhanced MRI represents a mixture of ventilation, diffusion, and perfusion by detecting the signal of molecular oxygen dissolv- ing in blood, the technique shows good agreement to ventilation scintigraphy (Schoenberg et al. 2005). In cases of complete tumor occlusion of the bronchus in a dependent lung segment or lobe, an absence of signal is noted, while in those cases with a patent bronchus but occluded pulmonary arterial supply, the oxygen signal is preserved (Fig. 42.2). In animal studies of oxygen-enhanced MRI, it was shown that the signal component from ventilation overweighs those of diffusion and perfusion (Keilholz et al. 2002).

Another important fi eld for which reduction of the acquisition time is of key importance is cardiac imaging. Due to improvements of surgical techniques, tumors with focal invasion of the left or right atrium can still undergo resection, whereas those with mas- sive infi ltration of the orifi ces of several pulmonary

Fig. 42.2. A 54-year-old male with a non-resectable stage-IIIB (T4N2M0) NSCLC. On the coronal HASTE sequence with short- ened echo train by parallel imaging (external reference scan, GRAPPA, R=2), a large mediastinal invasion of the tumor to the contralateral side is noted (arrow). This is confi rmed on the sagittal HASTE images (second row) on which the tumor growth along the right upper lobe bronchus and left main stem bronchus into the mediastinum in between the trachea and the large vessels can be clearly identifi ed due to the excellent tissue contrast between the tumor, the fat and the black-blood signal void of the vessels. Also note the sharp delineation of the anatomic structures due to reduced blurring in the images as a result of parallel imaging. No invasion of the bronchi is visible; even the upper lobe bronchus can be clearly delineated surrounded by tumor growth. The color-coded parameter maps of the oxygen-enhanced inversion-recovery HASTE sequence (third row) show preserved ventilation in all lung lobes including the right upper lobe. However, the arterial phase of the time-resolved 3D-CE-MRA scan (lower row, left) demonstrates a lack of perfusion in the right upper lobe due to complete occlusion of the right upper pulmonary artery (arrow) by the tumor confi rmed on the high-resolution 3D-CE-MRA image (lower row, right).

The combination of oxygen-enhanced MRI and time-resolved MRA correctly identiﬁ ed the ventilation/perfusion mismatch

컄컄

(7)

(8)

Fig. 42.3. A 56-year-old male with a resectable stage-IIIB NSCLC (T4N2M0). Besides the centrally growing large tumor of the right upper lobe, a small 8-mm metastasis is clearly identiﬁ ed on the HASTE images with parallel imaging (external reference scan, GRAPPA, R=2). In the two perpendicular orientations the lesion can be located in the same lobe, consistent with a T4 stage

the heart, and the large vessels of the mediastinum in multiple angulations. The introduction of TSENSE into clinical routine for cardiac MRI allows acquiring four to ﬁ ve slices within a single breath-hold without a noticeable loss of image quality if specially designed multi-channel coils are used (Wintersperger et al.

2006). The total scan time thus can be reduced to less than 5 min for time-resolved evaluation of tumor inﬁ ltration in two perpendicular imaging planes.

veins or inﬁ ltration of the ventricle are not amenable to surgery. The relationship between the tumor and the adjacent myocardium can be displayed in high detail with time-resolved ECG-gated cine TrueFISP imaging (Fig. 42.1). Without the use of parallel imaging, this approach is time-intensive since only one slice is acquired per breath-hold. It therefore takes total scan times in the order of 10 to 15 min in order to display the complex relationship between the tumor,

(9)

42.3.4

Increase of Anatomic Coverage

The introduction of multi-channel whole-body MRI systems in combination with highly accelerated parallel imaging allows accurate screening of metastasis with very good agreement to PET-CT. For metastatic involvement of the liver and adrenal glands, bones and central nervous systems, equal or even superior results have been found with sensitivities and spe- ciﬁ cities exceeding 95% (cf. Chap. 41) (Schmidt et al. 2005). Since metastatic disease in NSCLC most frequently involves these target organs, whole-body MRI with parallel imaging is expected to gain an increasing role for staging of this tumor entity. While at the moment whole-body screening for metastasis by itself requires total scan times of 50 min, further improvements in coil design might enable higher scan acceleration with parallel-imaging factors beyond 3 and thus allow integrating this approach into the MRI scan for local tumor staging.

42.4 Conclusion

The combination of fast T1-weighted and T2- weighted imaging, cardiac MRI, MR angiography, oxygen-enhanced MRI, and perfusion MRI offers a comprehensive, non-invasive, morphologic, and functional assessment of lung cancer (Kauczor and Kreitner 2000). Parallel-imaging techniques improve image quality, robustness, and acquisition speed of this approach, which therefore represents an attractive alternative to a step-wise multi-modality algorithm of multi-slice CT and nuclear-medicine studies for assessment of the operability and resectability of centrally growing advanced tumors.

References

Antoch G et al (2003) Whole-body dual-modality PET/CT and whole-body MRI for tumor staging in oncology. JAMA 290:3199–3206

Bock M (1997) Simultaneous T2* and diffusion measurements with 3He. Magn Reson Med 38:890–895

Boring CC, Squires TS, Tong T (1992) Cancer statistics, 1992.

CA Cancer J Clin 42:19–38

Bury T et al (1996) Staging of non-small-cell lung cancer by whole-body ﬂ uorine-18 deoxyglucose positron emission tomography. Eur J Nucl Med 23:204–206

Dales RE, Stark RM, Raman S (1990) Computed tomography to stage lung cancer. Approaching a controversy using meta- analysis. Am Rev Respir Dis 141 (5 Pt 1):1096–1101 Daly BD et al (1993) N2 lung cancer: outcome in patients with

false-negative computed tomographic scans of the chest. J Thorac Cardiovasc Surg 105:904–910; discussion 910–911 Dietlein M et al (2000) Cost-effectiveness of FDG-PET for the

management of potentially operable non-small cell lung cancer: priority for a PET-based strategy after nodal-negative CT results. Eur J Nucl Med 27:1598–1609

Dietrich O et al (2005) Fast oxygen-enhanced multislice imaging of the lung using parallel acquisition techniques. Magn Reson Med 53:1317–1325

Fink C et al (2004) Regional lung perfusion: assessment with partially parallel three-dimensional MR imaging. Radiol- ogy 231:175–184

Fink C et al (2005) Time-resolved contrast-enhanced three- dimensional magnetic resonance angiography of the chest: combination of parallel imaging with view sharing (TREAT). Invest Radiol 40:40–48

Grandy M (2001) Multislice-CT of patients with stage IIIa bronchial carcinoma for monitoring of neoadjuvant chemotherapy and presurgical evaluation. In: European Con- gress of Radiology, Vienna

Guyatt GH et al (1995) Interobserver variation in the computed tomographic evaluation of mediastinal lymph node size in patients with potentially resectable lung cancer. Ca- nadian Lung Oncology Group. Chest 107:116–119 Herman SJ et al (1994) Mediastinal invasion by bronchogenic

carcinoma: CT signs. Radiology 190:841–846

Herth F, Becker HD (2000) Endobronchial ultrasound of the airways and the mediastinum. Monaldi Arch Chest Dis 55:36–44

Kauczor HU, Kreitner KF (1999) MRI of the pulmonary parenchyma. Eur Radiol 9:1755–1764

Kauczor HU, Kreitner KF (2000) Contrast-enhanced MRI of the lung. Eur J Radiol 34:196–207

Keilholz SD, Knight-Scott J, Mata J, Fujiwara N, Altes TA, Berr SS, Hagspiel K (2002) The contributions of ventilation and perfusion in oxygen-enhanced pulmonary MR imaging: results from rabbit models of pulmonary embolism and bronchial obstruction. Proc Intl Soc Mag Reson Med 10:409

Keller SM et al (2000) Mediastinal lymph node dissection improves survival in patients with stages II and IIIa non-small cell lung cancer. Eastern Cooperative Oncology Group. Ann Thorac Surg 70:358-365; discussion 365–366

Kernstine KH et al (1999) PET, CT, and MRI with Combidex for mediastinal staging in non-small cell lung carcinoma.

Ann Thorac Surg 68:1022–1028

Kiyono K et al (1988) The number and size of normal mediastinal lymph nodes: a postmortem study. AJR Am J Roent- genol 150:771–776

Laissy JP et al (1994) Enlarged mediastinal lymph nodes in bronchogenic carcinoma: assessment with dynamic contrast-enhanced MR imaging. Work in progress. Radiology 191:263–267

(10)

Libshitz HI (1990) Computed tomography in bronchogenic carcinoma. Semin Roentgenol 25:64–72

Lofﬂ er R et al (2000) Optimization and evaluation of the signal intensity change in multisection oxygen-enhanced MR lung imaging. Magn Reson Med 43:860–866

Manfredi R et al (1996) Accuracy of computed tomography and magnetic resonance imaging in staging bronchogenic carcinoma. Magma 4:257–262

Mitchell JD et al (1999) Clinical experience with carinal resection. J Thorac Cardiovasc Surg 117:39-52; discussion 52–53

Mountain CF (1997) Revisions in the International System for Staging Lung Cancer. Chest 111:1710–1717

Mountain CF, Dresler CM (1997) Regional lymph node clas- siﬁ cation for lung cancer staging. Chest 111:1718–1723 Murata K et al (1994) Chest wall and mediastinal invasion by

lung cancer: evaluation with multisection expiratory dynamic CT. Radiology 191:251–255

Nguyen BC et al (1999) Multicenter clinical trial of ultrasmall superparamagnetic iron oxide in the evaluation of mediastinal lymph nodes in patients with primary lung carcinoma. J Magn Reson Imaging 10:468–473

Nikolaou K et al (2004) Quantiﬁ cation of pulmonary blood ﬂ ow and volume in healthy volunteers by dynamic contrast-enhanced magnetic resonance imaging using a parallel imaging technique. Invest Radiol 39:537–545

Padovani B et al (1993) Chest wall invasion by bronchogenic carcinoma: evaluation with MR imaging. Radiology 187:33–38

Pannu HK et al (2000) MR imaging of mediastinal lymph nodes: evaluation using a superparamagnetic contrast agent. J Magn Reson Imaging 12:899–904

Pearlberg JL et al (1987) Limitations of CT in evaluation of neoplasms involving chest wall. J Comput Assist Tomogr 11:290–293

Reinig JW et al (1994) Differentiation of adrenal masses with MR imaging: comparison of techniques. Radiology 192:41–46 Salvatierra A et al (1990) Extrathoracic staging of broncho-

genic carcinoma. Chest 97:1052–1058

Schmidt GP et al (2004) [Comparison of high resolution whole-body MRI using parallel imaging and PET-CT. First experiences with a 32-channel MRI system]. Radiologe 44:889–898

Schmidt GP et al (2004) [Whole-body MRI and PET/CT in tumor diagnosis.]. Radiologe 44:1079–1087

Schmidt GP et al (2005) High-resolution whole-body magnetic resonance image tumor staging with the use of parallel imaging versus dual-modality positron emission tomography-computed tomography: experience on a 32-channel system. Invest Radiol 40:743–753

Schoenberg SO (2005) Comprehensive pulmonary MR imaging in patients with bronchial carcinoma using parallel acquisition techniques. In: European Congress of Radiol- ogy, Vienna

Schoenberg SO et al [Ultrafast MRI phlebography of the lungs]. Radiologe 38:597–605

Scott IR et al (1988) Resectable stage III lung cancer: CT, surgical, and pathologic correlation. Radiology 166(1 Pt 1):75–79

Shirakusa T et al(1998) Extended operation for T4 lung carcinoma. Ann Thorac Cardiovasc Surg 4:110–118

Sider L, Horejs D (1988) Frequency of extrathoracic metastases from bronchogenic carcinoma in patients with normal- sized hilar and mediastinal lymph nodes on CT. AJR Am J Roentgenol 151:893–895

Valk PE et al (1995) Staging non-small cell lung cancer by whole-body positron emission tomographic imaging. Ann Thorac Surg 60:1573–1581; discussion 1581–1582

Vansteenkiste JF et al (1998) FDG-PET scan in potentially operable non-small cell lung cancer: do anatometabolic PET-CT fusion images improve the localisation of regional lymph node metastases? The Leuven Lung Cancer Group.

Eur J Nucl Med 25:1495–1501

Wahl RL et al (1994) Staging of mediastinal non-small cell lung cancer with FDG PET, CT, and fusion images: prelimi- nary prospective evaluation. Radiology 191:371–377 Webb WR et al (1985) Bronchogenic carcinoma: staging with

MR compared with staging with CT and surgery. Radiol- ogy 156:117–124

Webb WR et al (1991) CT and MR imaging in staging non- small cell bronchogenic carcinoma: report of the Radio- logic Diagnostic Oncology Group. Radiology 178:705–713 Wintersperger BJ et al (2006) Cardiac CINE MR imaging with

a 32-channel cardiac coil and parallel imaging: impact of acceleration factors on image quality and volumetric accuracy. J Magn Reson Imaging 23:222–227