Clinical review: Bedside lung ultrasound in critical care practice

Bedside chest radiography

In the ICU, bedside chest radiography is routinely performed on a daily basis and is considered as a reference for assessing lung status in critically ill patients with acute lung injury. Limited diagnostic performance and efficacy of bedside portable chest radiography have been reported in several previous studies [3–5]. Several reasons account for the limited reliability of bedside chest radiography. First, during the acquisition procedure, the patient and the thorax often move, decreasing the spatial resolution of the radiological image. Second, the film cassette is placed posterior to the thorax. Third, the X-ray beam originates anterior, at a shorter distance than recommended and quite often not tangentially to the diaphragmatic cupola, thereby hampering the correct interpretation of the silhouette sign. These technical difficulties lead to incorrect assessment of pleural effusion, lung consolidation and alveolar-interstitial syndrome.

Lung computed tomography

Lung CT is now considered as the gold standard not only for the diagnosis of pneumothorax, pleural effusion, lung consolidation, atelectasis and alveolar-interstitial syndrome but also for guiding therapeutic procedures in critically ill patients, such as trans-thoracic drainage of localized pneumo-thorax, empyema or lung abscess. Lung image formation during CT relies on a physical principle similar to that used for image formation during chest radiography: the X-rays hitting the film or the CT detector depend on tissue absorption, which is linearly correlated to physical tissue density. In the first generation of CT scanners, the tube emitting X-rays and the X-ray detector were positioned on the opposite sides of a ring that rotated around the patient. Typically, a 1 cm-thick CT section was taken during each rotation, lasting 1 second, and the table supporting the patient had to be moved to acquire the next slice, the ring remaining in a fixed position. These conventional scanners were slow and had a poor ability to reconstruct images in different planes.

In the nineties, spiral CT scanners equipped with a slip ring were introduced, giving the possibility of scanning a volume of tissue rather than an individual slice. Acquisition time was markedly reduced and high quality reconstruction in coronal, sagittal and oblique planes became possible using a work station. Current multi-slice CT scanners, the third generation of CT scanners, are equipped with multiple X-ray detectors and the tube rotates in less than one second around the thorax while the table supporting the patient moves continuously. The multiple detectors and the decrease in rotation time allow faster coverage of a given volume of lung tissue, contributing to increased spatial resolution (voxel smaller than 1 mm³). Using specifically designed computer software offering sophisticated reconstruction and postprocessing capabilities, several hundred consecutive axial sections of the whole lung can be reconstructed from the volumetric data and visualized on the screen of a personal computer. If the computer is connected to an appropriate workstation, it is then possible to 'move into the lung' and to measure CT attenuations in any part of the pulmonary parenchyma, providing direct access to regional lung aeration. In addition, images can be reconstructed in coronal, sagittal and oblique planes, offering the possibility of a three-dimensional view of the organ. For hospitals having a computer server to store and retrieve pictures from, films are no longer necessary and physicians can derive much more accurate information on patients' lung status.

With the old generation of conventional CT scanners, obtaining contiguous 1.5 mm-thick CT sections from the apex to the diaphragm would have exposed patients to unsafe radiation levels. With the new generation of multi-slice CT scanners, the ionizing radiation is slightly greater than from a single slice spiral scanner. However, because more slices and images can be easily obtained with multi-slice CT scanners, there is a potential for increased radiation exposure [6] that has to be balanced against the total radiation dose resulting from chest radiography performed daily at the bedside.

To perform a lung CT scan, however, requires transportation to the department of radiology, a risky procedure necessitating the presence of trained physicians and sophisticated cardio-respiratory monitoring [7]. In addition, helical multi-detector row CT exposes the patient to a substantial radiation dose, which limits the repeatability of the procedure [6]. For these different reasons, lung CT remains a radiological test, access to which is limited in many ICUs, and bedside lung ultrasound appears as an attractive alternative method for deriving information on lung status.

Technical equipment

Ultrasound machines should be lightweight, compact, easy to transport and robust, allowing multiple bedside examinations. They should be equipped with a high-performance screen and a paper recorder allowing transmission of medical information and subsequent comparisons. Generally, basic models presented by manufacturers combine all these features, and have the additional advantage of being reasonably priced. Such ultrasound machines are available in many emergency wards, ICUs, units of medical transportation and even in space [8–11].

Another technical characteristic should be required for the use of lung ultrasound in the ICU: the probes and the ultrasound machine should comply with repeated decontamination procedures since they serve multiple patients, and can be the vector for resistant pathogens that could be disseminated in the ICU [12–24]. The efficiency of the decontamination procedure is facilitated by a compact ultrasound machine equipped with a waterproof keyboard. This latter characteristic is present on a few ultrasound machines only, restricting choice.

Ideally, an emission frequency of 5 to 7 MHz is desirable for optimizing ultrasound visualisation of the lung. The probe should be small with a convex tip so it can be easily placed on intercostal spaces, which offer an acoustic window on the lung parenchyma. Generally, a convex array probe (3 to 5 MHz), as available on multi-purpose ultrasound machines, combines these advantages and allows a good visualization of lung.

Lung ultrasound examination

The patient can be satisfactorily examined in the supine position. The lateral decubitus position offers, however, a better view on dorsal regions of lower lobes. A complete evaluation of both lungs requires a systematic protocol of examination. First, the operator should locate the diaphragm and the lungs. Lung consolidation or pleural effusion are found predominantly in dependant and dorsal lung regions and can be easily distinguished from liver or spleen once the diaphragm has been located. Using anterior and posterior-axillary lines as anatomical landmarks, each chest wall can be divided into six lung regions that should be systematically analyzed: upper and lower parts of the anterior, lateral and posterior chest wall. In a given region of interest, all adjacent intercostal spaces offer acoustic windows that allow the assessment of the lung surface by moving the probe transversally. Dorsal lung segments of upper lobes, located behind the scapula, are the only regions that cannot be explored by lung ultrasound. To provide an exhaustive assessment of lung aeration and pleural effusion, the ultrasound examination should cover both lungs, just as for auscultation. To be comprehensive, a chest ultrasound examination should take around 15 minutes, although with enough knowledge and skills, users can perform lung examination more quickly.

Normal ultrasound pattern and basic abnormalities

Normally, ultrasounds are not transmitted through anatomical structures filled with gas and the lung parenchyma is not visible beyond the pleura. The injured lung is characterized by a marked increase in tissue extending to lung periphery that produces ultrasound artifacts resulting from the abnormal gas/tissue interface. A number of recently published studies have demonstrated the ability of bedside lung ultrasound to accurately assess lung aeration in patients with acute lung injury.

When the loss of aeration is massive and results in lung consolidation, or when a pleural effusion is present, ultrasounds are transmitted to deep intra-thoracic structures. As a consequence, intramediastinal organs like the aortic arch can be visualized in the presence of consolidation of upper lobes [26]. Several studies have clearly established the value of lung ultrasound for detecting and quantifying pleural effusion and lung consolidation. For physicians beginning their lung ultrasound training on critically ill patients on mechanical ventilation, the detection of pleural effusion and lung consolidation in dependant lung regions is the easiest part and the basic skill is generally acquired over a very short period of time [27].

Normal pattern

For each considered intercostal space, the probe should be positioned perpendicular to the ribs. Using a longitudinal view, the ribs, characterized by a posterior shadowing, should be identified. A hyperechoic and sliding line, moving forward and back with ventilation, is seen 0.5 cm below the rib line, and is called the 'pleural line'. In time-motion mode, a 'seashore sign' is present, characterized by motionless parietal tissue over the pleural line and a homogeneous granular pattern below it [28]. The pleural line results from the movement of the visceral pleura against the parietal pleura during the respiratory cycle. Beyond this pleural line, motionless and regularly spaced horizontal lines are seen: they are meaningless and correspond to 'artifacts of repetition'. Thus, a normal ultrasound pattern is defined by 'lung sliding' associated with artifactual horizontal A-lines (Figure 1, Additional file 1). In one-third of patients with normal lungs, however, isolated vertical B-lines can be detected in dependant lung regions and are devoid of any pathological significance. B-lines move with the pleural line and efface A-lines.

Alveolar-interstitial syndrome

In the presence of injured lung characterized by an increased amount of lung tissue extending to lung periphery [29], vertical artifacts arising from the pleura and extending to the edge of the screen [30] are detected and called vertical 'B-lines' or 'comet tails'. They appear as shining vertical lines arising from the pleural line and reach the edge of the screen. The number of these vertical B-lines depends on the degree of lung aeration loss, and their intensity increases with inspiratory movements [2, 31]. As mentioned above, less than one or two vertical artifacts can be detected in dependant lung regions in normally aerated lungs [31].

It has been demonstrated that multiple B-lines 7 mm apart are caused by thickened interlobular septa characterizing interstitial edema (Figure 2a, Additional file 2). In contrast, B-lines 3 mm or less apart are caused by ground-glass areas characterizing alveolar edema (Figure 2b, Additional file 3).

Lung consolidation

Massive lung edema, lobar bronchopneumonia, pulmonary contusion and lobar atelectasis all induce a massive loss of lung aeration that enables ultrasounds to be transmitted towards the depth of the thorax. Lung consolidation appears as a hypoechoic tissue structure that is poorly defined and wedge-shaped [32]. Within the consolidation, hyperechoic punctiform images can be seen, corresponding to air bronchograms (air-filled bronchi) [33]. Penetration of gas into the bronchial tree of the consolidation during inspiration produces an inspiratory reinforcement of these hyperechoic punctiform images. The ultrasound size of the consolidation is not influenced by respiratory movements (Figure 3a,b, Additional file 4). Several studies have demonstrated that lung ultrasound has a high performance in diagnosing alveolar consolidation and is helpful for guiding percutaneous lung biopsy [2, 34–37].

Peripheral lung abscesses with pleural contact or included inside a lung consolidation are also detectable by lung ultrasound [32, 35, 38, 39]. They appear as rounded hypo-echoic lesions with outer margins (Figure 4, Additional file 5). If a cavity is present, additional non-dependant hyperechoic signals are generated by the interface gas/tissue. By analogy with percutaneous drainage of abdominal collections, ultrasound-guided percutaneous drainage of lung abscesses has proved to be a safe and effective alternative to CT-guided drainage [32, 35, 38, 39].

Ultrasound assessment of alveolar recruitment and lung re-aeration

Lung ultrasound has been recently shown to provide the possibility of assessing quantitatively the lung re-aeration resulting from antimicrobial therapy in 24 critically ill patients with ventilator-associated pneumonia [40]. At the bedside, the whole lung was examined as described above, and each region of interest was attributed a score according to four stages of lung aeration before and after antimicrobial therapy: normal, interstitial syndrome (B lines 7 mm apart), alveolar-interstitial syndrome (B lines less than 3 mm apart) and alveolar consolidation. A tight correlation was found between pulmonary re-aeration measured by lung CT and the change in the 'ultrasound score'. Further studies are required to confirm whether lung ultrasound, using similar principles, provides the possibility of measuring alveolar recruitment resulting from positive end expiratory pressure (PEEP) or recruitment maneuver.

Pleural effusion

Pleural effusion should be sought on a longitudinal view, in dependant lung regions delineated by the chest wall and the diaphragm. It appears as a hypoechoic and homogeneous structure with no gas inside and is present during expiration and inspiration [41]. In other words, it appears as a dependant dark zone free of echo (Figure 3a,b, Additional file 4). Since pleural effusion acts as an acoustic window, lung can be seen as a bright pleural line if it remains aerated. If the pleural effusion is abundant enough to be compressive, the lung is seen consolidated and floating in the pleural effusion (Figure 5, Additional file 6). Assessment of pleural effusion requires attention to spleen or liver and diaphragm, especially when pleural puncture is considered. Pleural effusion can be easily distinguished from spleen or liver by using color Doppler that shows intrasplenic and intrahepatic blood vessels; or by visualization of a sinusoidal inspiratory movement of the visceral pleura from depth to periphery [42]. The skills required to detect pleural effusion are easy to acquire, as suggested by several publications [43–45].

The lung ultrasound approach has been proposed for quantifying pleural effusion volume [45–48]. In the supine position, an interpleural distance at the lung base, defined as the distance between the lung and the posterior chest wall, ≥50 mm is highly predictive of a pleural effusion ≥500 ml [45, 48]. Measurement of the interpleural distance can be performed at either end-expiration or end-inspiration [46], with no difference between them, and seems less reliable when measured on the left side [46]. All studies agree that ultrasound measurement of the interpleural space at the lung base is not accurate enough to quantify small (≤500 ml) and very large (≥1,000 ml) pleural effusions [45–47]. Recently, another ultrasound approach has been proposed for quantifying pleural effusion: by multiplying the height of the pleural effusion by its transversal area, measured half-way between upper and lower limits. An excellent correlation was found between the volume of pleural effusion assessed by CT of the whole lung and the ultrasound determination [49].

Although the nature of pleural effusion (transudate or exsu-date) cannot be accurately assessed on ultrasound examination only, some ultrasound patterns are evocative. Transudates are always anechoic but exsudates appear often to be echoic and loculated [50].

Last but not least, lung ultrasound is increasingly used for guiding thoracocentesis at the bedside [42, 51]. It provides the possibility of detecting pleural adherences that may hamper efficient thoracic drainage and transform thoraco-centesis into a risky procedure (Figure 6, Additional file 7). It enables the safe thoracic drainage of small and/or loculated pleural effusions. It may reduce the risk of intrafissural or intraparenchymal placement of thoracic tubes [52].

Pneumothorax

Pneumothorax is defined by the interposition of gas between visceral and parietal pleural layers. As a consequence, lung sliding is abolished, ultrasounds cannot be transmitted through the injured lung parenchyma and comet tails (vertical B-lines) are no longer visible. Only longitudinal reverberations of motionless pleural line (horizontal A-lines) can be seen [53]. In some circumstances, such as the presence of a thoracic tube, pleural adherences, bullous emphysema and advanced chronic obstructive pulmonary disease, lung sliding can be abolished in the absence of pneumothorax. The diagnosis remains uncertain in patients with normal lung aeration whereas in patients with lung injury, the presence of vertical B-lines rules out the diagnosis.

The ultrasound diagnosis of pneumothorax is the most difficult part of training: long experience is required to acquire appropriate skills that rely on the ability to recognize lung sliding and its abolition [42]. When possible, the use of higher emission frequencies (5 to 10 MHz) facilitates the recognition of lung sliding abolition. The diagnosis is even more difficult in the presence of partial pneumothorax. The patient should lie strictly supine to allow location of pleural gas effusion in non-dependant lung regions. To confirm the diagnosis of partial pneumothorax, examination should be extended lo lateral regions of the chest wall to localize the point where the normal lung pattern (lung sliding and/or the presence of vertical B-lines) replaces the pneumothorax pattern (absent lung sliding and horizontal A-lines). This point is called the 'lung point' [54] (Additional file 8). Utilization of the time motion mode can facilitate detection of the lung point (Figure 7).

The ultrasound pattern characterizing pneumothorax was described in the early 1990s [55–57]. Several studies have demonstrated that bedside lung ultrasound is more efficient than bedside chest radiography for diagnosing pneumothorax in emergency conditions if rapidly performed by the clinician in charge [28, 58–60]. Recently, interest in lung ultrasound for diagnosing pneumothorax in emergency and trauma patients has been reported [61]. Using a portable ultrasound device and a 3.5 to 7 MHz probe, three emergency physicians, having received formal 28 hour training for emergency bedside ultrasound, systematically performed lung ultrasound in 135 trauma patients admitted either to the resuscitation or the Emergency ICU of the Second Affiliated Hospital of Hangzhou (China). At admission, all patients had frontal chest radiography and 131 a thoracic computed scan of the whole lung, which served as gold standard for the diagnosis of pneumothorax. The sensitivity and specificity of lung ultrasound for diagnosing pneumothorax were 86% and 97%, respectively, whereas conventional chest radiography had sensitivity and specificity of 28% and 100%, respectively. Lung ultrasound over-diagnosed pneumothorax in two patients with pleural adherences. Bedside chest radiography missed all partial pneumothoraces whereas lung ultrasound detected the majority of them by identifying the lung point. In addition, lung ultrasound allowed the detection of pneumothorax within 2 to 4 minutes compared to 20 to 30 minutes for the chest radiography.

Limitations of lung ultrasound

When adopting lung ultrasound as a routine monitoring tool in the ICU, physicians should be aware of its limitations. Lung ultrasound examination and correct interpretation of the resulting images require formal training aimed at acquiring the necessary knowledge and skills. If several lung ultrasound examinations are performed on a daily basis, the learning curve for acquiring skills for diagnosing pleural effusion, lung consolidation and alveolar-interstitial syndrome is short, less than six weeks. The intra- and inter-observer variability is small, less than 5% [2]. The learning time for acquiring skills required for diagnosing pneumothorax is probably longer due to its low incidence in the critical care environment. In fact, the acquisition of the skills for diagnosing pneumothorax is the most difficult part of lung ultrasound training.

Lung ultrasound has intrinsic limitations that are not operator dependent but patient dependent. Obese patients are frequently difficult to examine using lung ultrasound because of the thickness of their rib cage. The presence of subcutaneous emphysema or large thoracic dressings alters or precludes the propagation of ultrasound beams to the lung periphery. Last but not least, it has to be pointed out that lung ultrasound cannot detect lung over-inflation resulting from an increase in intrathoracic pressures.