Chapter 4
Imaging
Sebastian Ley1 and Claus Peter Heussel2
1Diagnostische und Interventionelle Radiologie, Chirurgisches Klinikum München Süd, München, Germany. 2Diagnostische und Interventionelle Radiologie mit Nuklearmedizin, Thoraxklinik Heidelberg gGmbH, Heidelberg, Germany.
Correspondence: Sebastian Ley, Diagnostische und Interventionelle Radiologie, Chirurgisches Klinikum München Süd, Am Isarkanal 30, 81379 München, Germany. E-mail: ley@radiologie-ley.de
Many tracheobronchial, parenchymal and mediastinal diseases can be assessed by interventional pulmonology procedures. Imaging of the thoracic structures is the most important prerequisite before intervention. The tasks are manifold and include detailed visualisation of the anatomy, risk assessment and definition of the target. In this context, CT plays a major role, as it allows fast, highly spatial and three-dimensional imaging of the pathologies to be addressed. CT enables virtual bronchoscopies in order to plan the actual intervention, or serves as database for navigated EBUS or biopsy. CT also allows assessment of the lung parenchyma and thus determination of the eligibility of a patient and the target lobes for endobronchial lung volume reduction treatment. However, CT is only a morphological imaging technique, and functional assessment of tissue (i.e. lymph nodes) must be done by PET or magnetic resonance imaging. Post-treatment follow-up can be done by radiography for regular post-treatment assessment or by CT imaging for suspected complications.
Cite as: Ley S, Heussel CP. Imaging. In: Herth FJF, Shah PL, Gompelmann D, eds. Interventional Pulmonology (ERS Monograph). Sheffield, European Respiratory Society, 2017; pp. 49–63 [https://doi.org/10.1183/2312508X.10002617].
Various mediastinal, bronchial and lung diseases can be treated by interventional bronchoscopic techniques, which can be summarised as: tissue probing (lymph nodes, lung nodules), treatment of tracheobronchial bleeding, treatment of airway stenosis and endoscopic lung volume reduction (ELVR) [1, 2]. This chapter will focus on the radiological, or noninvasive, diagnostic modalities, which can also be used for further planning and guiding of interventions, leading to fast diagnosis with low risk, including the control and management of possible complications and success.
Patients usually present with a clinical syndrome, and (radiological) imaging is performed to further evaluate the underlying disease. Imaging therefore plays a vital role in the clinical assessment of patients with mediastinal and thoracic diseases. In particular, three-dimensional (3D) imaging techniques such as CT are frequently used techniques for assessment of tracheobronchial and pulmonary diseases. Although still more at an experimental stage, magnetic resonance imaging (MRI) is also used to address these issues and may be used for planning of interventional procedures.
Imaging techniques
For tailoring and planning interventions, high spatial resolution and/or functional information are helpful [3]. Various radiological imaging techniques are available for the assessment of pathologies of the mediastinum and lung parenchyma.
Radiography
The initial imaging procedure to work up a patient is usually a radiograph of the chest in two orientations, providing an overview of the thoracic and mediastinal structures and any pathologies to exclude severe acute comorbidities. However, this kind of projection imaging suffers from superimposition and is insufficient for guidance or planning of an interventional procedure. It is therefore used mainly as a baseline comparator for the post-interventional detection of complications (e.g. pneumothorax, effusion) and an estimation of treatment success (e.g. atelectasis, mediastinal shift, implant location) (figure 1).
Figure 1. Post-interventional chest radiograph of a patient with ILD. A cryobiopsy was performed for diagnostic purposes, and a right-sided haemothorax occurred as a complication, as shown.
Computed tomography
CT is routinely and widely used for detailed visualisation of the mediastinum, tracheobronchial tree and lungs. The spatial resolution of one CT image is typically 512×512 pixels in the x–y axis. The z-axis resolution is determined by the resolution of the detector used; usually it is in the range of 0.625 mm. Reconstruction of overlapping images is recommended (e.g. 20%) to allow optimal 3D post-processing. However, different examination protocols are applied to specific clinical situations.
Lung cancer screening
Within the European Union, lung cancer is the most frequently fatal cancer, leading to over 266 000 deaths per year and accounting for 20.8% of all cancer deaths [4]. Low-dose CT screening/scanning has evolved as the modality of choice for assessing high-risk populations [5]. These low-dose CT protocols, with a recommended dose of 0.1–0.6 mSv, aim to provide excellent visualisation of the nodular structures within the lung parenchyma (i.e. a high-contrast scenario) [6, 7]. However, these low-dose datasets inherit a lot of image noise, making evaluation of interstitial and mediastinal (low-contrast) structures difficult or even impossible. This technique is always carried out without application of intravenous contrast material. Therefore, these image datasets are often not suitable for 3D planning software tools, such as for electromagnetic navigation bronchoscopy (ENB), or as quantification tools for emphysema [8] or fibrosis [9].
While the sensitivity of CT is relatively high for detecting pathological structures, the specificity is relatively low. Infectious diseases in particular can mimic neoplastic pathologies, requiring further investigation. In a recent study, 116 patients were scheduled for navigational bronchoscopy for the diagnosis of a pulmonary lesion [10]. Of these, 7% had a decrease in size or resolution of their lesion at the CT done at the time of planned intervention, leading to cancellation of their procedure. For cancelled cases, the average time from initial CT prompting referral for bronchoscopy to the day of procedure scan was 53 days. Therefore, a follow-up scan before an intervention is recommended.
Standard-dose CT examination
A CT examination performed with a standard dose carries a higher radiation burden than the low-dose CT examinations used in the screening setting, usually between 2.6 and 4.3 mSv [11, 12], but allows assessment of the mediastinum, hilar and lung parenchyma structures. Use of a standard dose offers the possibility of scanning with or without i.v. contrast medium, including a CT angiogram.
An i.v. contrast-enhanced CT is recommended for mediastinal tumours/lesions or processes that infiltrate the chest wall. Contrast-enhanced CT should also be performed for assessment of haemoptysis or pulmonary embolism. The bolus timing must be chosen depending on the suspected vessel system, since the short acquisition time (e.g. 4 s) of modern scanners allows a low amount of contrast medium and therefore selective pulmonary–arterial, selective aortal or combined phases. This information must be provided to the CT technician prior to the CT scan, since individual bolus triggering enables optimal individual vessel opacification.
Mediastinal lymph nodes
Staging and evaluation of mediastinal lymph nodes by CT is based mainly on size: a node is deemed pathological with a short-axis diameter of >10 mm (sensitivity 60% and specificity 70% for malignancy) [13]. However, the excellent visualisation of CT allows precise localisation of the lymph node to plan the interventional biopsy of an enlarged lymph node [14–16].
Tracheobronchial tree
It is known that tracheobronchial variations and abnormalities occur in ∼2% of the population, although more are now being identified using CT [17] (figure 2). For almost two decades, CT imaging has been the best technique for noninvasive visualisation of the tracheobronchial tree [18]. CT datasets of the trachea can be used to plan and design individual tracheal cannulas, especially in cases of anatomical variants or previous surgical modifications (figure 3).
Figure 2. Routine CT for evaluation of ILD. The minimum intensity projection clearly demonstrates a separate origin of the right upper-lobe bronchus (bronchus suis) as a normal variant (arrow).
Figure 3. A sagittal reformated chest CT in a patient with a tracheal cannula. CT allows exact visualisation of the tracheal geometry and subsequent design of individualised tracheal cannulas.
From the trachea to the terminal bronchioles, an airway tree consists of approximately 17 generations of branches, beyond which alveoli begin to appear, ultimately terminating at the alveolar sacs.
Given the 3D nature of CT datasets, these datasets are perfectly amenable for automatic segmentation and data extraction. This is especially true for the central airways in humans, which branch in a dichotomous manner. Automatic segmentation of the tracheobronchial branching is possible down to the ninth generation (and partially down to the 11th generation) with standard PC hardware [19–21] (figure 4).
Figure 4. a) Automatic segmentation of the trachea and central airways. Visualisation in three-dimensional volume rendering is available as standard software (this example was generated with the Philips IntelliSpace Portal, COPD; Philips, Guildford, UK). b) An axial CT slice with segmented airways and a large nodule in the right upper lobe. c) Automatic lobar segmentation is also available for easy visualisation of the localisation of pathologies.
These 3D datasets form the basis for ENB, an image-guided approach that uses 3D-reconstructed CT scans and sensor location technology to guide a steerable endoscopic probe to peripheral lung lesions that may be beyond the reach of conventional bronchoscopes [22]. While actual bronchoscopy offers inherent valuable options for treatment and tissue assessment, virtual bronchoscopy based on CT image datasets provides several additional advantages: 1) the ability to pass stenoses virtually, 2) the view of the stenosis is not limited to a central-to-peripheral view but is also possible from a peripheral-to-central view, and 3) extralumenal organs such as lymph nodes, and tumours, vessels, struma and atelectasis, for example, can be displayed in the same procedure, enabling direct assessment of a mass causing a compression or obstruction; a multiplanar reformat orthogonal to the bronchial centre line can be useful in this situation.
With respect to malignant bronchial stenosis, squamous cell carcinoma, accounting for ∼25–30% of all lung cancers, is the most common cause arising in central airways. The progression from normal bronchial epithelium to squamous metaplasia followed by dysplasia, carcinoma in situ and finally invasive carcinoma has been well described. Approximately 11% of patients with moderate dysplasia and 19–50% with severe dysplasia will develop invasive carcinoma. Therefore, reporting of central airway obstruction on radiology reports can have an impact on bronchoscopic airway interventions and patient outcomes [23] (figure 5). HERTH et al. [24] compared CT assessment of thoracic tumour invasion into the bronchial wall versus EBUS assessment of the same condition in 131 patients and concluded that EBUS had a far better specificity (100%), sensitivity (89%) and accuracy (94%) compared with a CT scan (28%, 75% and 51%, respectively). The ability of chest CT and EBUS to distinguish between compression and infiltration was measured against the histology results [24].
Figure 5. Minimum intensity projection of a CT dataset in the coronal orientation. Narrowing of the trachea (arrow) has resulted from tuberculosis infection. The tuberculosis is also located in the left upper lobe (arrowhead).
Prior to an actual bronchoscopic intervention, rendering of a virtual bronchoscopy based on the CT dataset is possible to plan the procedure [25]. This might help to identify features such as lymph nodes for TBB, which might be relevant in staging purposes. In 32 patients with thoracic malignancies, CT scans and virtual bronchoscopy were performed and compared with actual bronchoscopy [26]. The sensitivities of CT scanning and virtual bronchoscopy for the detection of endoluminal, obstructive and mucosal lesions were 90%, 100% and 16%, respectively.
Prior to stent placement, thin-section CT with both 2D and 3D reconstructions along the bronchial centreline are helpful to evaluate the relationship of the airway to adjacent structures such as major vessels [27, 28].
Multiplanar reconstruction and 3D volume-rendered images aid diagnostic interpretation and help in communicating results to referring physicians. CT can accurately evaluate the aetiology, location and length of airway obstruction, and can determine the type, length and sizing of airway stents. Landing zones can be accurately defined, and anatomical variations (e.g. tracheal bronchus) can easily be considered.
The CT protocol utilised for stent evaluation should be tailored to the specific patient. In cases of malignant central obstruction, i.v. contrast should routinely be administered to delineate the relationship between the obstructing mass, bronchi and central vascular structures.
CT is also the modality of choice for planning of any endobronchial volume reduction treatment. In this setting, CT allows combined visualisation of the lung parenchyma and airways. Many factors are now recognised as needing to be evaluated for estimation of success of ELVR [2]. First, it was shown that heterogeneous emphysema is a better predictor for outcome than bullous or other forms of emphysema [29]: a target lobe is destroyed by emphysema. Second, the lung lobe must be affected by emphysema in a fixed manner, meaning that there is no change in air content during inspiration and expiration [30]. This information can be achieved by imaging the patient in both respiratory settings and comparing visually, or preferably quantitatively, the density changes between both examinations. Furthermore, treatment with obstructive endobronchial valves is only successful if there is no collateral ventilation to the treated lobe. The major contributors to the integrity of the lobes are the lobar fissures (figure 6). A computer-assisted analysis of 573 CT examinations recently showed that ∼90% of all examined persons had incomplete fissures regardless of whether COPD was present or not [31]. Optionally, functional information, including lung lobe perfusion, provides additional information on which lobe to treat as a target.
