Adult Chest Surgery

Chapter 3. Chest Imaging: Role of CT, PET/CT, and MRI

Chest imaging is a critical diagnostic tool for evaluating thoracic anomalies in anatomic structure and disease. The variety of imaging technologies available for diagnostic evaluation in the chest includes plain-film radiography, computed tomography (CT), positron-emission tomography (PET), concurrent PET/CT, and magnetic resonance imaging (MRI). These radiologic procedures are further enhanced by oral or intravenously administered contrast materials used alone or in combination. The role of radiologic imaging in thoracic surgery is likely to gain even more importance as imaging technologies provide ever more accurate means of visualization.

Patients are often referred to thoracic surgeons for evaluation and treatment of incidental findings on chest CT or radiography. These incidental findings can be fortuitous for the patient, providing the opportunity for treatment before the development of symptoms heralding advanced disease. The thoracic surgeon may choose to further the evaluation with registered PET/CT or to follow indeterminate findings over time with serial CT scans.

CT is the backbone imaging modality for preoperative evaluation. The adrenal glands are always included in routine chest CT images because this is a common site of lung cancer metastases. CT can be supplemented with PET/CT or MRI for special purposes. These modalities are often useful for problem solving. PET/CT has dramatically increased the ability of imaging to contribute to accurate preoperative staging in lung cancer, thereby setting patients on the proper treatment course from the outset. The resulting change in lung cancer staging flows from the ability of PET/CT both to recognize unsuspected distant metastases and to identify coexisting benign disease. For example, before the availability of PET/CT, inflammation in contralateral lymph nodes often was attributed erroneously to a tumor of more advanced stage and patients were not offered potentially curative resection. Adjuvant PET/CT can provide preoperative staging information capable of upstaging (30%) or downstaging (15%) disease in an individual patient.In the setting of heterogeneous disease, PET/CT can be used to select the "best" biopsy site, in turn decreasing the number of biopsy specimens required to definitively classify difficult-to-identify cancers such as diffuse malignant pleural mesothelioma.

MRI is less useful than PET/CT, particularly with the advent of multidetector CT scanners, which permit data to be acquired with voxels of equal dimension in all three planes, thus providing sagittal and coronal images with CT that were previously only possible with MRI. This is not to say that MRI is without advantages. MRI can sensitively differentiate tissues, including blood, with differentiation of the various states of hemoglobin. In addition, fascial planes are more sharply delineated by MRI. However, MRI demonstrates calcification as a signal void and thus may be considered inferior to CT for detecting calcifications. The use of MRI for problem solving is more apt to reflect the problem under consideration than a standard approach, although standardized imaging generally is applicable for visualizing the complete thorax in patients with diffuse malignant pleural mesothelioma.

In other instances, the area of interest, such as the thoracic inlet, brachial plexus, or lung apex, will be imaged without imaging the rest of the thorax by MRI. MRI and CT are equivalent for imaging lymphadenopathy in the mediastinum. MRI is by nature less than contiguous and otherwise should be viewed as complementary to CT imaging.


CT provides detailed anatomic images of the chest in which a variety of soft tissues can be recognized, along with water, fat, and bones; the resulting basic transaxial images are the in vivo equivalent of transaxial anatomic pictures of a cadaver (Fig. 3-1). State-of-the-art multidetector CT scanners are capable of acquiring ever-increasing numbers of individual slices of data at one time. Four-detector scanners scan the chest in approximately 20 seconds, a practical time for patient breath-holding. Readily available clinical models with the capability of producing 16–64 slices at one time can scan the chest in 10 seconds or less. Alternatively, very small structures can be studied using ever-smaller slice thicknesses. Development currently focuses on providing up to 256 slices at one time. Along with cardiac gating, this technology permits unprecedented in vivo evaluation of ultrastructure in the lungs as well as the heart.

Figure 3-1.


Transaxial CT image above level of aortic arch, presented in lung window. Skin, muscles, bones, and mediastinum are all visible. Right and left brachiocephalic veins are seen anterior to the principal branches of the aorta. This patient has adenocarcinoma in the right upper lobe.

The technical parameters of CT scanning can be altered to improve the visualization of different tissues. CT of the chest is performed with a small focal spot using kilovolts between 100 and 120 and milliamperes between 20 and 200 or more. Reconstructed images include contiguous 5-mm images with soft tissue smoothing for visualization of the heart, great vessels, and mediastinal structures. These are generally referred to as mediastinalimages and are displayed with a window level of 25–40 Hounsfield units (HU) and a relatively narrow window width, such as 360 HU. In addition, separate images with edge enhancement for optimizing visualization of lungultrastructure are displayed with a window level of –600 HU and a wide window width of 1500–2000 HU.

Hounsfield units derive their name from the developer of the CT scanner, Nobel laureate Sir Godfrey N. Hounsfield. The scale arbitrarily assigns water the attenuation value of 0, air –1000, and bone up to +1000. These numerical values of normalized x-ray attenuation define the gray scale of all CT images. The display windows highlight various structures based on the relationships between the underlying fundamental gray scale and the composition of various tissues in the body.

Intravenous iodinated contrast material is used commonly to provide optimal delineation of vascular structures, particularly when they lie in close proximity to the pathologic entity. Thus lung cancer staging is performed most often with intravenous contrast. NPO conditions should be instituted 4 hours before the examination to minimize nausea and vomiting. In many instances, patients with a history of contrast material allergy may be imaged with MRI instead of CT. When it is mandatory to use CT in a patient who has had a prior contrast material reaction, pretreatment with oral steroids and Benadryl can be considered. Oral contrast material is rarely used because air and fat often provide adequate contrast for identifying gastrointestinal tract structures.

Increasing concern for nephrogenic systemic fibrosis has led to the institution of reduced-dose regimens for patients with impaired renal function. Half the standard dose of contrast material is used for patients with an estimated glomerular filtration rate of between 30 and 60 mL/min/1.73 m2; intravenous contrast material should not be given to a patient with an estimated glomerular filtration rate of less than 30 mL/min/1.73 m2. In the setting of impaired renal function, as with contrast material allergy, consulting the radiologist before the study will ensure that the best possible study is selected for the given patient. Even with normal renal function, it is advisable to separate examinations that require administration of intravenous contrast material by at least 24 hours.

Varying the section thickness sometimes can improve visualization. The thinnest section that can be obtained is directly related to the size of the focal spot with which the scan was performed, currently providing images as thin as 0.6 mm. Images of 1–2 mm reconstructed with an edge-enhancing algorithm are still the most commonly provided thin-section images of lungs. These high-resolution CT (HRCT) images can be derived from the same data acquisition on multidetector scanners. Interspersed HRCT images permit visualization of the lung parenchyma and pleura and are most helpful for evaluating diffuse diseases such as emphysema and bronchiectasis. Contiguous or overlapping thin-section images are used for studying small nodules. These are appropriate for evaluating the features of a given nodule when reconstructed using a lung algorithm and identifying fat and calcification when reconstructed using a soft tissue algorithm.

Edge enhancement produces artifacts that can be mistaken for calcification. Thus, evaluating for the presence of calcifications should be performed using mediastinal soft tissue reconstruction. A second caveat must be offered when looking for calcified pulmonary nodules. Since contrast material may give small vessels a dense appearance very similar to calcification, non-contrast-enhanced CT may be preferable in the setting of prior granulomatous infection. Nodule surveillance guidelines published by the Fleischner Society in November 2005 can help to minimize the number of CT examinations performed, particularly for very small nodules in patients at low risk for lung cancer(Table 3-1).

Table 3-1. Fleischner Society 2005 Recommendations for CT Follow-up of Small Nodules

Nodule Size

Low-Risk Patient

High-Risk Patient

4 mm

No follow-up required.

If no change at 12 months, no further follow-up required.

4–6 mm

If no change at 12 months, no further follow-up required.

Follow-up CT at 6–12 months. If no change, CT at 18–24 months.

6–8 mm

Follow-up CT at 6–12 months. If no change, then CT at 18–24 months.

Follow-up CT at 3–6 months. If no change, then CT at 18–24 months.

>8 mm

CT at 3, 9, and 24 months or contrast-enhanced CT or PET/CT or biopsy.

CT at 3, 9, and 24 months or contrast-enhanced CT or PET/CT or biopsy.


Specialized CT examinations are performed according to disease-specific algorithms. Interstitial lung disease is evaluated on 1- to 2-mm-thick images obtained without contrast material. Increasingly, these thin-section HRCT images are obtained from volumetric data acquired from a single breath-hold and reconstructed retrospectively at specified intervals. Radiologists generally will evaluate HRCT images in conjunction with standard renderings of the volumetric CT data set. It is very important to obtain and view HRCT images with the proper field of view. Reducing the size of the image reduces the information available from the images.

CT pulmonary angiography is performed with a higher concentration of iodine, such as 370 g/100 mL, compared with 300 g/100 mL for ordinary chest CT contrast. With the more crucial timing requirement for imaging contrast material in pulmonary arteries, a test bolus or automated bolus tracking software is often used to refine the timing rather than relying on an approximation of the circulation time from the antecubital fossa to the main pulmonary artery at 20 seconds. Rendering of CT pulmonary angiographic images requires thin-section imaging and often uses a variety of special reconstructions and multiplanar images in sagittal and coronal planes. A plane for reconstruction also may be chosen to follow the axis of the pulmonary arteries at the bifurcation. It is important to remember, especially when working with patients who have hypercoagulable states owing to processes such as cancer, that pulmonary emboli may be visualized on standard CT images, such as those performed for staging. Secondary criteria, including atelectasis and pleural effusions, may be absent. In patients who have had a lung resection, particularly pneumonectomy, a common location for the accumulation of thrombus is at the site of lung resection in the terminus of the pulmonary artery stump. Thrombus in such a location may persist for long periods of time.

Three-dimensional reconstruction is performed increasingly for understanding the anatomy in relation to the function and pathology. This strategy is being used to evaluate airways for bronchomalacia and in the fitting of bronchial stents, as well as in patients in whom virtual bronchoscopy can provide visualization of a point beyond the proximal obstruction. Functional information regarding obstruction may be added to such examinations by acquiring a second CT data set at reduced dose, such as 80 mA, during expiration. Acquiring images in frank expiration to eliminate respiratory motion is a useful strategy when the suspected level of obstruction is distal, at a level such as the terminal bronchioles. Pathology in the midmediastinum with a complex relationship to the heart and great vessels may be mapped using three-dimensional reconstruction with color rendering of the images for surgical planning. Cardiac gating can eliminate confusion that may arise from cardiac motion for this purpose. Since the cardiac gating increases the dose required, it should be used for imaging only when it will provide crucial information.

Some centers evaluate nodules with dynamic imaging during and following administration of intravenous contrast material. A nodule is likely benign if enhancement is less than 10 HU. The nodule is likely malignant if enhancement is greater than 20 HU. A nodule that demonstrates enhancement of 15 HU or more is not usually a granuloma. This technique has proved to be user-dependent and therefore may yield disappointing results in centers where it is not performed commonly. It can be used when PET/CT is not available, however, and may be advantageous in regions where there is endemic granulomatous disease such as histoplasmosis.

Perioperative CT scans are performed for a variety of reasons and may or may not include the administration of intravenous contrast material. Infected pleura may enhance with contrast material and better delineate lung from pleural effusion. Evaluations related to pneumothoraces generally do not benefit from intravenous contrast material administration. Oral contrast material may be used for the evaluation of potential leaks after surgery such as esophagectomy. The oral contrast material should be water soluble and administered by a surgeon with clear purpose in choosing both the route of administration and volume of contrast material. Preliminary images performed before the administration of oral contrast material are often the most important images obtained. These images permit the detection of subtle changes such as those caused by the introduction of oral contrast material that leaks into spaces such as the pleural space. The preliminary images eliminate confusion related to a variety of sources of radiopaque materials such as surgical clips and previously administered contrast material. Barium can remain in the lung and in the pleural space permanently.


The PET/CT scanner combines a gamma camera and multidetector CT scanner in the same instrument, allowing the images from both examinations to be displayed together with registration to increase the identification of subtle signs of pathology. The PET scanner records the positron emissions from a radioactive tracer, that is, [18F]fluorodeoxyglucose (18F-FDG), hence the name positron-emission tomography. Both scans generally image from the top of the skull through the pubic symphysis. In the case of melanoma, the lower extremity imaging is extended. The long range of PET/CT scans provides a total body examination but also permits physiologic motion over time, which creates the potential for discordance between PET and CT images. This can occur as a result of peristalsis in the gastrointestinal tract and, especially, breathing. While a patient could not be expected to breath-hold through the entire examination, lasting minutes, a separate chest CT can be obtained during a breath-hold to improve the resolution of lung imaging. In some centers, PET/CT is a routine procedure for indications such as solitary pulmonary nodule, lung cancer, and mesothelioma.

Concurrent PET/CT imaging combines the ability to detect subtle metabolic changes through the preferential uptake of 18F-FDG by metabolically active cells responsible for the growth of abnormal cells (PET) with precise anatomic location of disease (CT), tumor, or affected tissue (Fig. 3-2). It is currently a reimbursable procedure when used for diagnosing solitary pulmonary nodules and tumor staging. The patient usually must fast for a minimum of 6 hours before the injection of 18F-FDG. In some centers, the patient may be instructed to have very specific meals at specified times before the examination to reduce cardiac uptake of 18F-FDG through saturation of receptors. The patient rests quietly for approximately 1 hour after intravenous injection of the radioisotope to permit the tracer to disperse throughout the body before imaging. Some institutions also give dilute oral contrast material to improve visualization and identification of abdominal structures. The PET and CT scans are both performed on the same scanner without moving the patient. Patients having the examination for the investigation of pathology within the chest also should have a CT of the chest during a breath-hold while in the same position. Although this scan is generally of lower quality than a diagnostic CT, such a scan will improve evaluation of the lungs significantly compared with a standard CT obtained to correct for attenuation. Intravenous contrast material is not yet a standard feature of this examination because conventional iodine contrast agents interfere with the PET scan portion of the examination.

Figure 3-2.


Transaxial CT and transaxial PET images presented over a fused transaxial image and projection image, which also can rotate during interpretation. Small nodule in right upper lobe demonstrates no 18F-FDG avidity. This determination is reliable only for nodules larger than 7 mm in diameter.

Pulmonary nodules that measure 7–8 mm in diameter or greater can be evaluated reliably with PET/CT. While it is possible for smaller pulmonary nodules and lymph nodes to demonstrate avidity for 18F-FDG, the scan requires very high metabolic activity, leaving uncertainty when a nodule is not apparently avid for 18F-FDG.

PET scans also require calibration for accuracy. The process of quantifying 18F-FDG avidity, or uptake, is complex, with extensive quality control measures. Quantification procedures vary somewhat between institutions, particularly in regard to correction and reporting of the standard uptake value (SUV). The SUV relates the activity concentration in a volume of tissue to the amount of injected dose and the patient's body weight. The maximum SUV indicates the affinity of a pathologic process, such as a tumor, for glucose. This correlates with the aggressiveness of a tumor histologically. There is no correlation between SUV and CT attenuation measured in Hounsfield units.

The pitfalls of PET/CT scanning remain numerous at this time despite its extreme utility for thoracic surgical evaluations. Investigations of intravenous contrast material and more widespread addition of breath-hold images for the lungs, it is hoped, will lead to a diagnostic-level CT scan within a PET/CT investigation and enable radiologists to keep the radiation of patients as low as reasonably achievable. PET/CT provides the best preoperative staging currently possible. The results of the two types of scans combined may be thought of as concordant or discordant based on whether the findings can be correlated. The consistency with pathologic truth is a separate and equally important consideration. PET/CT has not replaced conventional imaging in its ability to exclude brain metastases from lung cancer. The accuracy and therefore utility of PET/CT for the brain are limited. While brain metastasis may be found on PET/CT, the lack of a finding does not exclude metastasis. The pitfalls of PET/CT correlative imaging in bone also have proved to be a significant limitation. Metastases, depending on phase and rate of bone destruction, can be concordant or discordant, leading to a number of confusing situations, some of which would be better resolved with conventional radionuclide bone scanning. In a number of potentially confusing situations, evaluation of SUV in various locations may reconcile PET and CT findings with the pathologic processes present. Inflammation can have a very high SUV but generally will have low to moderate values, whereas an avid tumor will have a much higher SUV (e.g., tumor SUV of 10 and inflammatory process SUV of 3). It is also possible for tumors to have little or no avidity for 18F-FDG. Two tumors in the lung are particularly important in this regard, bronchioalveolar cell carcinoma (BAC) and carcinoid. The 18F-FDG uptake of a tuberculoma can be extremely high, with an SUV as high as 20. The decision to operate must consider factors from CT and the clinical evaluation of the patient. Clarity is sometimes increased by serial PET/CT scans to watch lesions over time.


MRI uses a variety of pulse sequences to identify unique characteristics of soft tissues and fluids that cannot be detected with CT scanning. MRI can sensitively identify blood and determine the length of time it has been present based on the state of hemoglobin as it changes to deoxyhemoglobin and further degradation products. MRI can readily determine the direction of blood flow in a vessel, useful information not provided at the same level by CT and not addressed by PET/CT. The use of MRI in the chest has increased as data acquisition times have diminished, allowing breath-hold imaging of the lungs (Fig. 3-3). MRI examinations are customized to the problem being evaluated. Coils used to perform the examination not only provide improved imaging but also control technical parameters such as field of view. The bore of the available MRI scanner itself may limit the sizes of patients who can have chest MRI. Larger-bore and open scanners have decreased this limitation, but a patient may have to go to a special location to have such an examination. It is helpful to explore patient claustrophobia before ordering the examination. Patients who are concerned about having the examination often benefit from oral premedication that permits normal outpatient scanning. Of course, the standard exclusions for any MRI, such as aneurysm clip, recent surgery, and pacing devices, apply to chest MRI.

Figure 3-3.


Coronal and sagittal MRI images of the chest have less spatial resolution than CT images, whereas contrast resolution is greater. Contrast enhancement on coronal image reveals small pleural soft tissue nodules above the black-appearing fascial plane at the right hemidiaphragm. Mediastinal structures are well seen with enhancement of vascular structures. Sagittal T2-weighted image reveals intact diaphragmatic fascia. Cortical bone is visualized as a black signal void.

The need to visualize blood vessels, nerves, and the variety of substances, including blood, that can be found in the pleural or pericardial space serves as a guide in choosing MRI for a particular patient. As a problem-solving tool, the examination will be customized by the radiologist. In unusual situations, it is best to discuss the problem with the radiologist before ordering the examination. This will ensure adequate scanner time and the best chance that important clinical questions will be answered. Until recently, 20 mL of intravenous gadolinium contrast material was administered routinely both to identify the tissue-enhancement characteristics of many tumors and to perform magnetic resonance angiography. As with iodinated contrast material for CT scans, however, documented cases of nephrogenic systemic fibrosis have led to the restriction of contrast material administration, particularly in the setting of impaired renal function.

MRI is the primary imaging modality for thoracic outlet syndrome. For this type of evaluation, the patient is imaged with the arms up and with the arms down. The vessels and nerves of the thoracic inlet and brachial plexus region are studied using limited field of view and special blood flow techniques.

MRI is also performed routinely for diffuse malignant pleural mesothelioma. This is the most standardized chest MRI examination, imaging the complete chest and upper abdomen. Intravenous gadolinium contrast material is administered to detect the enhancement of tumor masses. Of note, recent incorporation of PET/CT into mesothelioma protocols has resulted in more specific identification of sites that will yield a productive tumor biopsy than can be achieved with MRI alone. The MRI itself is more helpful for clarifying the integrity of fascial planes at the diaphragmatic and mediastinal boundaries of the tumor. Operability is determined through this combination of tests to determine unforeseen distant disease and local extension beyond the scope of extrapleural pneumonectomy (see Chap. 103).

Pancoast tumors also are imaged with MRI to best evaluate relationships between the brachial plexus and apex (see Chaps. 67 and 68). Depending on lesion size and clinical considerations, the examination may be planned as a brachial plexus examination or a full field-of-view examination, as performed for mesotheliomas.

MRI is a primary tool for the noninvasive evaluation of adrenal masses. When an adrenal lesion does not exhibit the low attenuation associated with adrenal adenomas, the MRI may confirm the presence of an adenoma and obviate the need for biopsy.

Cardiac MRI, performed with cardiac gating to eliminate the motion of the heart, is also used increasingly for surgical planning in the removal of large central mediastinal masses and evaluation of structures adjacent to the heart that are not well seen on CT scans.


The orientation of the radiologist to anatomy is derived from the anatomic position. Projection radiographs therefore are viewed as you would look at the patient, placing the patient's right side at your own left side. Coronal cross-sectional images use the same orientation, generally presented from anterior to posterior. Axial images are viewed from the perspective of standing at the supine patient's feet, again placing the patient's right side at your own left side. Sagittal images generally are presented from the patient's left side to the right side, although this convention is not applicable to all studies, particularly in MRI. It is helpful to reconcile the position of the heart in determining whether sagittal images are on the right or left side of the patient's body. The Visible Human Project and the proliferation of Web-based medical education materials now allow easy access to comparison images for anatomic identification in all three of these planes. Sophisticated image processing of volumetric CT data sets increasingly enables radiologists to provide data for surgery in the perspective of the surgeon. Surgeons also have learned to use conventional axial CT images to determine operability and plan specific surgeries.


Lung anatomy most frequently presents three lobes in the right lung and two lobes in the left lung. The right upper lobe consists of three segments: anterior, posterior, and apical. The apical bronchus may be thought of as the chimney of the lung with its vertical orientation. The right middle lobe consists of two segments, medial and lateral. The lateral segment extends more posteriorly, causing it to project behind the right lung on a lateral view of the chest. The right lower lobe also has an apical segment, as well as four basilar segments: medial, anterior, lateral, and posterior. The left upper lobe includes a combined apical-posterior segment, an anterior segment, and superior and inferior lingular divisions. The left lower lobe is similar to the right lower lobe, although the basilar segmental anatomy is somewhat more variable, commonly having only three segments, such as anterior, lateral, and posterior. Segment identification is best confirmed by reconciling both airway and fissural anatomy (Figs. 3-4, 3-5, 3-6, and 3-7).

Figure 3-4.


Transaxial CT image of the chest rendered for lungs at the level of the carina. Segmental airways to the right upper lobe are well seen. An avascular plane localizing the major fissure is better seen on the left than on the right.


Figure 3-5.


Transaxial CT image of the chest rendered for lungs at the level of the right minor fissure. Bronchus intermedius is well seen on the right. Complete left major fissure extends to the left hilum.


Figure 3-6.


Transaxial CT image of the chest rendered for lungs at the level of the lower lobe airway bifurcations. Right middle lobe bronchus gives rise to medial and lateral branches. The medial branch is also seen dividing on this image. Pulmonary veins are entering the left atrium. Lung medial to the vein (V3B) on the right is in the right upper lobe. Left lower lobe segmental bronchi are demonstrated behind the left major fissure.


Figure 3-7.


Transaxial CT image of the chest rendered for lungs at the level of the pulmonary venous confluence. Major fissures are seen anterior to the pulmonary venous confluence in each lung. Minor thickening of the right major fissure produces a subtle line on this 5-mm image, whereas the left major fissure is visualized only as an avascular plane. HRCT images at this level, along with images at the level of the aortic arch and carina, respectively, are useful for assessing the distribution of diffuse lung diseases in lung parenchyma.


The periphery of the lung is covered in visceral pleura, along the chest wall and each fissure. Fissures are frequently incomplete. Common anatomic variants include the superior and inferior accessory fissures, the azygos fissure (when azygos vein migration is incomplete), and the left minor fissure. Unlike the cobblestone structure of the lung, the pleura is a primary source of linear structure in the lungs. On CT images, fissures are seen as avascular planes on 5-mm images and as fine white lines on images up to approximately 2 mm in thickness. Mediastinal pleural surfaces are least well seen by CT and benefit from MRI, whereas fissures are not visualized on many MRI sequences unless they are abnormally thickened (Fig. 3-8).

Figure 3-8.


Transaxial HRCT image of the chest rendered for pleura reveals detail regarding airways seen here, with minimal dilatation in the right lower lobe lung parenchyma, and precisely defines the location of pleural fissures visualized as thin white lines. The right middle lobe only exists within the circle of pleura defining the minor fissure. Lung parenchyma surrounding the minor fissure is in the right upper lobe. Subpleural nodules are noted in the right lower lobe. HRCT images for lung parenchyma are not appropriate for the detection of calcium, as suggested by this image.


The mediastinum is divided into three compartments: anterior mediastinum, middle mediastinum, and posterior mediastinum. The superior mediastinum is not a mediastinal compartment, but it is commonly understood to include lymph nodes and normal structures above the top of the aortic arch. There exist several classifications of mediastinal compartments, besides the anatomic divisions, that have been used by radiologists for many years to simplify the generation of a differential diagnosis (see Table 3-1). Felson's mediastinal compartments and the "modified" Felson mediastinal compartments are used most commonly.3,4 Ben Felson, a pioneering thoracic radiologist, defined projection radiography boundaries for the mediastinum, bringing the anterior mediastinum from the anterior chest wall through the retrosternal clear space to the anterior wall of the trachea. The later modification drew this posterior boundary at the anterior wall of the ascending aorta. In both, the posterior mediastinum begins 1 cm behind the anterior margin of the vertebral bodies, behind the anatomic mediastinum. In this way, differential diagnosis is simplified for the posterior compartment without undue complication of the middle mediastinum. Adding the esophagus to the middle mediastinal compartment can be useful for creating clear differential diagnoses that are directly applicable to clinical practice. The choice between the two versions of Felson's mediastinal compartments is best made based on the logic employed by the individual radiologist and mirrors variations within surgical practices as well. Those who want to work with short differential diagnosis lists and can readily remember to look at the potential pathology arising from an adjacent space will be well-served by the modified Felson's criteria applied to lateral chest radiographs. One further reminder is in order: The internal mammary vessels and any accompanying lymphadenopathy project with but are not part of the anterior mediastinum (Figs. 3-9, 3-10, 3-11, and 3-12).

Figure 3-9.


Transaxial CT image of the chest rendered for mediastinum at the level of the aortic arch. Brachiocephalic veins are joining to form the superior vena cava. Anterior mediastinal fat has a triangular shape in the thymic bed region of the anterior mediastinum.


Figure 3-10.


Transaxial CT image of the chest rendered for mediastinum at the level of the aorticopulmonic window. Small lymph node adjacent to the left side of the trachea and origin of left main stem bronchus is in station 4R, whereas the small lymph node lateral to the ligamentum arteriosum is in the aorticopulmonic window lymph node station. Anterior mediastinal fat again demonstrates normal thymic fat in a middle-aged adult. Internal mammary vessels are well seen flanking the sternum.


Figure 3-11.


Transaxial CT image of the chest rendered for mediastinum at the level of the bifurcation of the carina and pulmonary arteries.


Figure 3-12.


Transaxial CT image of the chest rendered for mediastinum at the level of the aortic root. The right atrium is right of aortic root, and the pulmonary outflow tract is left of the aortic root. The left atrium is posterior to the aortic root and anterior to the descending thoracic aorta. The azygos vessels flank the descending thoracic aorta. The esophagus is collapsed between the left atrium, the descending thoracic aorta, and the azygos vein. The posterior mediastinal compartment begins behind the descending thoracic aorta.


Solitary Pulmonary Nodule

A solitary pulmonary nodule found on a plain chest radiograph should be further analyzed with CT to characterize the nodule, determine whether additional nodules are present, and assess other associated findings, including lymphadenopathy and pleural effusions.2,5–8 The term mass is reserved for large nodules; over time, the definition has decreased progressively from 6 cm to 4 cm to, most recently, 3 cm. The size of a nodule itself correlates positively with the probability of malignancy, even in the absence of additional features to suggest malignancy. The margins of the nodule, pattern of calcification (if present), and presence or absence of fat help to distinguish benign from malignant lung nodules. Special features such as enlarged feeding and draining vessels also can help to indicate a very specific diagnosis. A nodule containing dense central calcification, whether solid or lamellated, is benign, requiring no further follow-up to determine the nature of a calcified granuloma. Thin-slice soft tissue reconstruction of CT data provides the most accurate assessment in this regard.9The common causes of solitary pulmonary nodules are listed in Table 3-2.

Table 3-2. Differential Diagnosis for Solitary Pulmonary Nodules

Nodule Etiology

Distinguishing Characteristics


Smooth margins


Solid or lamellated calcifications


Lobulated margins


Dystrophic, eccentric calcifications


Lobulated margins


Calcifications appear in rings or arcs




Lobulated margins


Infrequent calcification (vascular)


Feeding/draining vessels

Lung cancer

Spiculated, lobulated, or smooth margins


Dystrophic calcifications


Large lesions with necrosis


Cavitation in squamous cell carcinoma and adenocarcinoma


5 mm of atypical adenomatous hyperplasia


Ground-glass opacity


Well-demarcated margins


Part-solid nodule


Cystic spaces


Focal extensions to pleura


Very slow growth

Solid pulmonary metastasis

Nonspecific, although may have appearance characteristic of primary tumor


AVM = arteriovenous malformation; BAC = bronchioalveolar cell carcinoma.

Granulomas are the result of inflammatory processes and may vary in size as well as presence of calcification. Since the dense, solid, calcified nodule can vary in size, it is important to remember that such nodules are benign calcified granulomas. Granulomas are not necessarily calcified, though. A solitary noncalcified granuloma must be treated as an indeterminate pulmonary nodule. It is helpful to consider the prevalence of granulomatous disease in the patient population, which also will reflect the presence of endemic granulomatous diseases in the community. The size of the nodule is also a factor because tiny nodules, measuring less than 4 mm in diameter, rarely signify early malignancy.

Carcinoid tumors are low-grade neuroendocrine malignancies that commonly present as endobronchial lesions with a cherry-like appearance on bronchoscopy, often associated with mucoid impaction of the distal airways. Dystrophic calcification in a lobulated nodule seen obstructing a bronchus, perhaps with mucoid impaction also evident, is a classic description of a carcinoid tumor. Not all carcinoid tumors have every one of these features, however, and the presence or absence of individual imaging features does not correlate with typical versus atypical carcinoid. The most extreme form of neuroendocrine tumor, small cell lung cancer, may present with no obvious lung nodule but with striking lymph node enlargement.

Hamartomas can contain specific calcifications, described as rings and arcs, owing to cartilage that may be present within the hamartoma. Fat also may be seen in the same pulmonary nodule. As with carcinoid tumors, hamartomas also may have a lobulated contour. In the case of a hamartoma, the identification of fat on thin-section images is the most convincing evidence of benignity. Fat also can be seen in nodules or even consolidations if the lesion is caused by the aspiration of mineral oil, commonly referred to as lipoid pneumonia.

Solitary pulmonary nodules and small nodules are studied with serial CT examinations over 2 or more years to determine whether a nodule is benign or malignant. One outcome of screening studies such as the Early Lung Cancer Action Project is recognition of the extremely low incidence of cancer in tiny nodules.10 This observation contributed to the Fleischner Society Guidelines, which currently recommend only a single follow-up CT scan for solitary pulmonary nodules that measure less than 4 mm in diameter and then only in high-risk patients (Tables 3-1, 3-2, and 3-3). Clinical practitioners have yet to become comfortable with this "no follow-up" recommendation for patients at low risk of developing lung cancer, but this change has definitely increased patient and practitioner comfort with the 6- to 12-month interval surveillance CT.

Table 3-3. Differential Diagnosis of Mediastinal Masses


Differential Diagnosis



Anterior mediastinal mass

Thyroid mass

Contiguous with thyroid gland

Deviation of trachea


Thymoma or thymic cyst

Thymic bed

Smoothly marginated


Lymphoma and small cell lung cancer

All lymph node stations in superior mediastinum, thymic bed, prevascular space, aorticopulmonic window

Extramediastinal location hilar lymph nodes

May be lobulated

Mass may involve multiple lymph node groups

Hilar lymph node enlargement may be asymmetric

Hodgkin's disease spreads from thymic bed to middle mediastinum to hilar lymph nodes


Germ cell tumor

Variable, including prevascular space and thymic bed

Fat, hair, and teeth are diagnostic

May be homogeneous with smooth margins

Middle mediastinal mass

Duplication cyst (includes bronchogenic cyst)

Most often located at bifurcation of trachea and central airways

May be paraesophageal or intraparenchymal

Smoothly marginated

High-attenuation fluid



All lymph node stations, including subcarinal space

May appear as separate enlarged nodes or as a multiple lymph node mass

May be homogeneous or heterogeneous

Low attenuation indicates tuberculosis

Single site with enhancement indicates Castleman's disease


Pericardial cyst

Adjacent to heart, especially in cardiophrenic sulcus

Smoothly marginated

Water attenuation

Can also represent pericardial diverticulum if history of mediastinoscopy


Thyroid mass (intrathoracic goiter)

Thyroid, extending into thorax

15% of these masses extend behind the trachea

Appearance of thyroid tissue is heterogeneous, can include calcifications and focal fluid


Tracheal tumor

Within or surrounding trachea

Narrowing of trachea

Adenoid cystic carcinoma has more tumor outside the trachea than within it, so-called toothpaste lesion


Vascular variants and abnormalities

Posterior to trachea

Anterior or posterior to esophagus

Diverticulum of Kommerall with aberrant subclavian artery

Vascular rings and slings

Posterior mediastinal mass

Esophageal abnormalities and masses

Large esophageal mass can occupy middle and posterior mediastinal compartments

Esophageal cancer

Foregut duplication cyst


Neurogenic tumor

Connected to neural foramen

Smoothly marginated or lobulated

May have low attenuation

May contain fat


Extramedullary hematopoiesis

Paraspinal masses

Masses are often paired

Smoothly marginated


Note: Radiographic division of middle mediastinum may include esophagus, thyroid, and lymph nodes, leaving neurogenic tumors and extramedullary hematopoiesis as primary considerations in posterior mediastinum.

The recognition of malignant potential in small nodules has increased through more than a decade of lung cancer screening trials and is now well-supported by CT technology allowing volumetric imaging of nodules with slice thicknesses of less than 1 mm. The relationship to mortality risk from these lesions is less clear than the radiopathologic correlation consistently demonstrated. Through such radiopathologic correlations, we have learned that invasive adenocarcinoma is often present in the lesions described as BAC (Fig. 3-13). The presence of a number of features simulating inflammation is better understood within larger lung cancer lesions, as well as in small lung cancer lesions, and has resulted in earlier resection of many such lung cancers. The lepidic growth of tumor along alveolar walls is the hallmark of BAC. On CT, this can have the appearance of well-demarcated but subtle ground-glass opacity, which, by definition, allows visualization of vessels and airways within the nodule. The importance of these subtle opacities, particularly when present with findings suggestive of more advanced lung cancer, has caused them to be reclassified as nonsolid nodules. As the tumor increases, such nodules develop areas of more solid opacity that can result in a part-solid nodule. In this context, the development of bars corresponds to invasive adenocarcinoma. Small cysts, some of which may be indistinguishable from bronchi, and focal extensions to pleural surfaces, with and without deflection of the pleural reflection, are also seen. Since these lesions can be unifocal or multifocal and require different treatment strategies, extensive follow-up CT scanning is performed. Comparison of size measurements does not stand alone. Thin-section reconstruction in lung kernel provides the most consistent data for comparison. Although the Early Lung Cancer Action Project group projected improved survival for lesions having these features if surgically resected within 30 days, short-term follow-up still reveals resolution of many such opacities, particularly after a course of antibiotic therapy. In the case of an opacity owing to either inflammation or BAC, more than 2 years of follow-up may be required to prove the lack of growth over time. A nodule with the features of BAC discussed here that measures 5 mm or less in diameter is classified as atypical adenomatous hyperplasia despite identical histology. Small nodules that may include BAC features should be removed rather than biopsied for a more accurate histologic diagnosis.

Figure 3-13.


Transaxial 1-mm HRCT image of the chest rendered for lungs at the level of the aortic arch reveals adenocarcinoma with BAC features in the periphery of the left upper lobe. It contains a solid component anteriorly, ground-glass opacity medially and posteriorly, along with a small cystic feature centrally and focal radiations extending to the pleural surface anteriorly and laterally.

The presence of tumor elsewhere in the body also may increase concern for metastasis, although the lung nodule may be the initial presentation of an extrathoracic malignancy. Colon cancer, common in the age group of patients who develop lung cancer, is particularly associated with large "cannon ball" and potentially solitary pulmonary metastases.

Mediastinal Mass

Mediastinal masses can be tricky to image with CT. The most difficult decision is regarding the administration of intravenous contrast material. If a patient has a thyroid mass that can be treated with radioactive 131I, the administration of iodine contrast material is contraindicated. The administered iodine contrast material would saturate the iodine receptors to which radioactive 131I also binds, requiring a 3-month delay in treatment to allow the receptors to become available again to 131I. Not giving iodine contrast material, on the other hand, masks the diagnosis of Castleman's disease. In the case of Castleman's disease, or angiofollicular lymph node hyperplasia, the enhancement of the mass is most apparent by comparing scans before and after the administration of intravenous contrast material. Adopting the strategy of paired CT scans with and without contrast material for all nonthyroid mass examinations unnecessarily increases the radiation exposure for most patients. Thymoma, lymphoma, and teratoma may be imaged with or without contrast material. Contrast material sometimes adds clarity to the examination of mediastinal masses. The age of the patient is generally more helpful than contrast enhancement with these tumors (Fig. 3-14).

Figure 3-14.


Transaxial 5-mm non-contrast-enhanced images of the chest rendered for mediastinum reveal a homogeneous, smoothly marginated mass in the right side of the thymic bed. This appearance is classic for a thymoma, especially when seen in middle age.

Extensive Pleural Disease—Diffuse Malignant Pleural Mesothelioma

The pleural space is not well vascularized and therefore can provide an environment for infection that is not easily treated with antibiotics. Hence extensive pleural disease requires attention from the thoracic surgeon whether it is benign or malignant. Since 800 mL of pleural fluid or a change in pleural fluid volume of this magnitude can be undetectable on a bedside chest radiograph, pleural effusions detected by imaging are often significant. The characterization of small, medium, and large for the size of a pleural effusion is a gross approximation on chest radiographs, although such imaging may be more helpful for quantification than CT and MRI. Differences in patient position during the examination make it difficult to directly compare the sizes of pleural effusions over time using different modalities. The volume of a pleural effusion is often overstated on cross-sectional imaging reports compared with chest radiography. In the absence of quantification, the size of a pleural effusion on axial CT images is often determined by cranial-caudal extent, resulting in overestimation of the size of many significant pleural effusions. The same problem also applies to reporting the size of a pneumothorax. As image processing enters the clinical practice of radiology, more quantification may be provided on a routine basis. The more pressing issue in this regard is in the setting of primary pleural tumor with extensive pleural disease, such as in malignant pleural mesothelioma. Fluid and tumor masses may encase the lung with a thickness that warrants measurement despite the complexities involved. In some instances, the additional findings such as extrathoracic lymph nodes and invasion of vital structures, whether in the mediastinum or the abdomen or by extensive involvement of the chest wall, may be more important than quantification of tumor mass, fluid, or both within the pleural space (Fig. 3-15).

Figure 3-15.


Coronal MRI and transaxial CT images demonstrate lobulated pleural thickening with small collections of fluid encasing the right lung. MRI demonstrates the fascial plane between the pleural disease and the chest wall, diaphragm, and mediastinum. Note the extrapleural lymph nodes in the right anterior diaphragmatic region on the bottom CT image.

Since the pleural surface is very thin, pleurectomy may not be recognized on postoperative CT. Adjacent hemorrhage is often seen without conveying its postoperative significance. The performance of the extrapleural pneumonectomy (see Chap. 103), most often for malignant pleural mesothelioma but also on occasion appropriate for the more common adenocarcinomatosis of the pleural space and unusual tumor metastases, requires careful consideration of preoperative cross-sectional imaging, generally with CT, MRI, and PET/CT.11 The use of ultrasound is limited primarily to the localization of small collections of pleural fluid. Imaging modalities generally are complementary, but caution is warranted regarding the limitation of each modality in the assessment of extensive pleural disease.

Contrast-enhanced chest CT is the most basic of these imaging techniques, but it can be the best imaging modality for detection of small extrathoracic lymph nodes and bone destruction. CT is not sensitive to focal invasion of the abdomen and may overestimate invasion of mediastinal structures by contiguous tumor. Secondary signs, such as a pericardial effusion in the setting of pericardial invasion, may be helpful for correct assessment of disease extent by CT. Multiplanar reconstruction has increased the utility of the volumetric CT data conventionally acquired by multidetector CT.

MRI, performed with multiplanar T1- and T2-weighted sequences and intravenous injection of 20 mL of gadolinium contrast agent, provides the best demonstration of fascial planes. In particular, sagittal MRI provides the best preoperative evaluation for the integrity of the hemidiaphragm, and all three planes contribute in a similar manner to detection of mediastinal fascial planes. MRI demonstration of tissue characteristics also highlights the distinction between tumor masses and fluid in the pleural space. MRI provides less spatial resolution and may not image small structures, including tiny lung nodules, even when the structure is within the image. Furthermore, MRI sequences are not volumetric in the manner of CT scans and thus can fail to image small structures.

PET/CT is not always performed; however, it is being used increasingly to select the best possible biopsy target, thereby improving the initial diagnosis of malignant pleural mesothelioma. Multimodality therapy also may be offered based on PET/CT findings, particularly when intense 18F-FDG activity is seen in extrathoracic lymph nodes despite being smaller than can reliably be detected by the radioisotope and smaller than can be reliably identified by contrast-enhanced CT. Volumetric measurement of tumor burden also will be enhanced by functional information from PET scanning. Consequent improvements in the evaluation of treatment for malignant pleural mesothelioma also can lead to the use of PET/CT for evaluation of treatment adequacy in benign processes such as empyema.


The use of CT to improve survival of lung cancer patients does not depend on the outcome of the current lung cancer screening CT debate. Although the results of the National Lung Screening Trial have yet to be published, it is widely accepted that CT will find smaller, presumably earlier lesions, whether it is the patient's first lung cancer, a recurrent lung cancer, or independent development of a new lung cancer in a patient who has already had lung cancer. In the interim, thoracic surgeons are performing surveillance CT examinations in increasing numbers for patients at the highest risk, that is, those who have already developed lung cancer for which they have been treated successfully. Lung cancer recurrences tend to be seen relatively soon after initial resection, most often within the first 2 years of surgery. Determination of baseline and surveillance to establish stability are well-covered by semiannual CT for up to 5 years after a primary resection, such as lobectomy. This longer period of more frequent follow-up is in line with clinical practice and increases data points for very slowly growing, early lung cancers. Since the incidence of new lung cancer in a patient who has been diagnosed and treated previously for lung cancer is 3% per year, this population, at extreme risk, may deserve continued surveillance with annual CT indefinitely. Postoperative imaging in this manner will result in the development of the necessary expertise within an institution should screening eventually become recommended more widely.

The performance of these CT scans is not without risk of radiation-induced lung cancer, but this risk can be minimized by reducing the radiation dose exposure to 80 mA without sacrificing high-resolution diagnostic-quality information regarding lung parenchyma. The effective radiation dose from one standard chest CT scan is approximately 7.95 milliSieverts (mSv), whereas the effective dose from a chest CT scan performed with 120 kV and 80 mA is 3.2 mSv. For 15 CT scans over 10 years, the cumulative doses would be 119 mSv versus 48 mSv. This is a 60% reduction in dose. Looked at from a different perspective, the same radiation exposure could provide 37 chest CT scans at 80 mA compared with 15 CT scans at 200 mA. The excess risk of developing lung cancer owing to this level of radiation exposure is 0.59% at 200 mA and 0.24% at 80 mA. Thus the excess risk attributable to chest CT scans is small compared with the risk of developing a second lung cancer.12 As a practical matter, 120 kV and 80 mA may be used for all postoperative chest CT scans, with iodionated contrast material used only for issues in the hila and, perhaps, for a baseline postoperative CT. Minimizing dose and exposure to iodine contrast material while performing continued surveillance CT scans in patients who have had lung cancer can maximize the benefit while minimizing the risk associated with surveillance CT scans.

From screening CT activities throughout the world, we have clearly learned more about the identification of early lung cancer, particularly using BAC features often associated with adenocarcinoma of lung.13 The contribution of radiology departments to the understanding of histologic features of adenocarcinoma, particularly bronchioalveolar features, has increased with screening programs and the introduction of ever-thinner CT images, represented in sagittal and coronal as well as axial planes. From a technical perspective, the ability to generate images has in some respects outpaced the ability of radiologists to analyze these images, and some form of computer-assisted analysis will be used in the future.

The use of imaging for screening of patients at high risk for developing lung cancer may change significantly over the next few years. In addition to guidelines regarding who should receive CT, at what age, and at what intervals in time, we may well see the introduction of a biomarker screening test for lung cancer. Image-guided therapy and the introduction of new drugs to treat tumors will further hone diagnostic evaluation with imaging and shape the future practice of thoracic surgery.

Radio frequency ablation, now in its infancy, may one day be used to treat early lung cancer, where today it is just used palliatively. If we can treat lung cancer earlier as well as less invasively, we also will avoid the complications of incomplete treatment and bronchopleural fistula formation currently seen with radio frequency ablation. This therapy ultimately may become part of a multimodality approach to lung cancer, requiring less radical surgery and permitting cure of patients who are presently unsuitable candidates for surgery. At that point, the existence of even more successful treatment options will mandate more widespread lung cancer screening. Lack of a definitive protocol for early lung cancer detection is the principal impediment to achieving a cure at this juncture.


Radiologic imaging plays a central role in the management of thoracic disease. Familiarity with the various imaging modalities can facilitate the relationship between the thoracic radiologist and surgeon and best address clinical problems. Lung cancer diagnosis, surgical planning, and long-term management of patients with thoracic problems benefit from close collaboration between thoracic surgeons and thoracic radiologists.


1. Seltzer MA, Yap CS, Silverman DH, et al: The impact of PET on the management of lung cancer: The referring physician's perspective. J Nucl Med 43:752–6, 2002. [PubMed: 12050318]

2. MacMahon H, Austin JH, Gamsu G, et al: Guidelines for management of small pulmonary nodules detected on CT scans: A statement from the Fleischner Society. Radiology 237:395–400, 2005. [PubMed: 16244247]

3. Felson B: Standard terminology in computerized tomography. Semin Roentgenol 13:185–6, 1978. [PubMed: 684453]

4. Whitten CR, Khan S, Munneke GJ, Grubnic S: A diagnostic approach to mediastinal abnormalities. Radiographics 27:657–71, 2007. [PubMed: 17495284]

5. Colson Y, Sanders J, Nason L, Jacobson F: Primary care physicians and the fight against lung cancer. Primary Care Rep 13:1–19, 2007. 

6. Carr S, Pechet T: Solitary pulmonary nodule: Assessment of a solitary pulmonary nodule. ACS Surg Online 1–8, 2006. 

7. Jeong YJ, Yi CA, Lee KS: Solitary pulmonary nodules: Detection, characterization, and guidance for further diagnostic workup and treatment. AJR 188:57–68, 2007. [PubMed: 17179346]

8. Cochrane Miller J: Evaluating pulmonary nodules. Radiol Rounds 2:1–4, 2004; available at 

9. Li F, Sone S, Abe H, et al: Malignant versus benign nodules at CT screening for lung cancer: Comparison of thin-section CT findings. Radiology 233:793–8, 2004. [PubMed: 15498895]

10. Henschke CI, Yankelevitz DF, Libby DM, et al: Survival of patients with stage I lung cancer detected on CT screening. N Engl J Med 355:1763–71, 2006. [PubMed: 17065637]

11. Gerbaudo VH, Sugarbaker DJ, Britz-Cunningham S, et al: Assessment of malignant pleural mesothelioma with 18F-FDG dual-head gamma-camera coincidence imaging: Comparison with histopathology. J Nucl Med43:1144–9, 2002. [PubMed: 12215551]

12. Brenner DJ: Radiation risks potentially associated with low-dose CT screening of adult smokers for lung cancer. Radiology 231:440–5, 2004. [PubMed: 15128988]

13. Travis WD, Garg K, Franklin WA, et al: Evolving concepts in the pathology and computed tomography imaging of lung adenocarcinoma and bronchioloalveolar carcinoma. J Clin Oncol 23:3279–87, 2005. [PubMed: 15886315]

If you find an error or have any questions, please email us at Thank you!