Quantitative Lung Ultrasound: Technical Aspects and Clinical Applications

Correlations between ultrasound findings and a decrease in lung aeration are attracting attention because lung ultrasound has become easily accessible, quick, and repeatable at the bedside thanks to handheld devices. Any probe is suitable for lung ultrasound examination (fig. 1), and convex and microconvex probes are the most frequently used in emergency departments.^37,43–45  The probe is usually oriented longitudinally—perpendicular to the ribs—because this allows clear identification of the bat sign, corresponding to the upper and lower ribs, and the pleural line (fig. 2).

The scanning technique in the emergency setting comprises eight zones, four chest areas per hemithorax: upper and lower anterior and upper and lower lateral areas (fig. 3).⁴³  In an emergency department, the examination focuses on the recognition of B-positive areas, defined by the B-pattern—at least three B-lines in an intercostal space (fig. 4)—and how they change in response to procedures and medical treatments.

Lung ultrasound findings closely correlate with extravascular lung water, measured by transpulmonary thermodilution (pulse index continuous cardiac output). The A-pattern (i.e., A-lines or no more than two B-lines in an intercostal space, with visible lung sliding) predicts low extravascular lung water, whereas the number of B-positive areas correlates with extravascular lung water. However, the association with echocardiographic measures is suggested for a correct estimate of the wedge pressure.^31,37 

When combined with other scores, such as the Brest score, a systematic lung ultrasound evaluation improves the diagnosis of acute heart failure.³²  This offers a good correlation between the number of B-positive areas and left ventricular diastolic dysfunction, pulmonary artery hypertension, severity of mitral regurgitation, and amino-terminal pro-brain natriuretic peptide.³⁶  As a mirror of the air/fluid ratio in the pulmonary parenchyma, lung ultrasound can also help guide fluid resuscitation, avoiding the risks of harmful fluid overload on lung function.^4,44,46,47 

Heart failure signs and symptoms become evident late, and their disappearance is not always linked to real improvement. This may result in incorrect titration of therapy and inappropriate early discharge of the patient.⁴⁴ 

Advanced invasive methods for assessing extravascular volume, although reliable,⁴⁷  do not lend themselves well to the emergency department. Nowadays, chest x-ray films and the plasma brain natriuretic peptide concentration are considered standard noninvasive methods for staging acute decompensated heart failure. Volpicelli et al. found concordance between a decrease in B-positive areas and a reduction of extravascular lung water, as well as improvements in chest x-ray films and plasma brain natriuretic peptide.⁴⁶  Further studies in patients under hemodialysis found significant correlations between the reduction of B-positive areas and body weight loss during hemodialysis.^29,48 

In the patient presenting with acute respiratory failure, the total number of B-lines assessed in the four conventional areas per hemithorax has been used to differentiate pulmonary primary infection from overinfection in decompensated cardiac failure: the latter typically has significantly more B-lines.⁴⁹  A simplified lung ultrasound scoring system, similar to the one used in ICUs (see the section “The Critically Ill Patient”) but limited to three areas per hemithorax, can also help predict hospital mortality of these septic patients.⁵⁰ 

In the ICU, the lung ultrasound score is the most frequently used scoring system. It distinguishes four progressive steps of loss of aeration according to the artifacts visualized in a scan: score 0, normal aeration (A-lines or no more than two B-lines); score 1, moderate loss of aeration (three or more well-spaced B-lines); score 2, severe loss of aeration (coalescent B-lines); and score 3, complete loss of aeration (tissue-like pattern; figs. 4 and 5). These steps were previously referred to as N, B1, B2, and C. This score is computed in six regions per hemithorax: anterior, lateral, and posterior fields are identified by sternum, anterior, and posterior axillary lines; each field is divided into superior and inferior regions (fig. 3).

It is suggested to start in the anterior fields with a linear probe and then to switch to a phased-array or a convex probe if consolidation is visualized. A transversal approach, aligned with the intercostal space, would be preferred once the pleura has been correctly identified: it visualizes a wider pleura and more artifacts (fig. 2).⁵¹ 

The lung ultrasound score is easily learned (25 examinations) and repeatable, for both image acquisition and interpretation.^22,51  Complete examinations take 10 to 15 min.⁵²  A three-level competence pyramid was recently suggested: academic teacher with several publications in the field; more than 25 supervised procedures and 200 nonsupervised procedures; and at least 25 supervised procedures and less than 200 nonsupervised procedures).⁵³  The authors suggest daily practice at the beginning of the learning curve.

Regional scores showed strong correlations with lung density measured by quantitative computed tomography.²²  The global lung ultrasound score is computed as the sum of regional scores; therefore, the total score ranges from 0 to 36. It showed good linear correlations with extravascular lung water measured by pulse index continuous cardiac output¹⁹  and a strong association with lung density by quantitative computed tomography,^22,28  although with a tendency to overestimate lung aeration loss in acute respiratory distress syndrome (ARDS).

A recent study suggested distinguishing moderate and severe loss of aeration (scores 1 and 2) in relation to the percentage of pleura affected by artifacts (score 2 if B-lines or subpleural consolidations in clearly 50% or more).⁵¹  This may limit overestimation of loss of aeration in patients with frequent focal coalescences and nonhomogeneous diseases such as ARDS, trauma, and ventilator-associated pneumonia. This modified score had the strongest correlation with extravascular lung water compared to all the other rating systems, and included automation⁴² ; therefore, it should be the preferred strategy for regional/global lung aeration assessment and monitoring in the critically ill.

The lung ultrasound score has been employed for monitoring lung aeration in ARDS patients receiving extracorporeal membrane oxygenation^54,55 ; progressive reduction of the score indicates gradual recovery of the injured lung. An accurate bedside tool is of particular interest in these patients because transport to a radiology department involves high risk, and standard chest x-ray films are poorly informative in almost completely deaerated lungs.

A quantitative approach can also be employed to monitor the response to treatments in the critically ill patient. In ARDS patients in shock requiring fluid resuscitation, changes in the lung ultrasound score were the earliest parameter indicating an increase in extravascular lung water as a detrimental effect of the fluid challenge.²¹  The changes in lung aeration detected by ultrasound are seen sooner than deterioration in gas exchanges, because oxygenation can initially be improved by the transient positive impact of fluid resuscitation on hemodynamics. A reaeration score computed on the basis of the same four patterns before and after treatments modifying aeration accurately detected the changes in lung aeration induced by 7 days of antibiotics in ventilator-associated pneumonia, with significant correlations with quantitative computed tomography.¹⁷ 

Similarly, a quantitative approach can be useful to guide and monitor the respiratory strategy, including pronation and positive end-expiratory pressure (PEEP) titration. A personalized approach does seem to improve survival of ARDS patients, as long as the lung pattern can be accurately distinguished as focal or diffuse.^56,57  Focal distribution, with normal aeration in anterior fields and severe or complete loss of aeration in posterior ones, indicates response to pronation^58,59 : in patients with focal distribution, the lung ultrasound score changes more when they are prone. A dedicated score for posterior fields has been proposed to identify long-term responders to pronation.⁶⁰ 

In a first study by Bouhemad et al.,¹⁸  a significant correlation was also found between the reaeration score and PEEP-induced recruitment measured by a pressure–volume curve in a decremental PEEP trial. Patients with a diffuse pattern—i.e., loss of aeration also affecting anterior fields—were more frequently PEEP responders.

The correlation between PEEP-induced recruitment and lung ultrasound scores was not confirmed in a study comparing lung ultrasound score and quantitative computed tomography: computed tomography defines recruitment as reaeration of previously not aerated tissue, whereas the lung ultrasound score detects any improvement in aeration, regardless of the initial state. This intrinsic difference implies the absence of any direct correlation between these two techniques, while not affecting the interest in a bedside ultrasound assessment of PEEP-induced changes in lung aeration.⁵²  For a better comparison, only changes in the regions with a tissue-like pattern should probably be considered. However, to date there are no clinical studies assessing this.

In terms of outcome prediction, the lung ultrasound score after 3 days of mechanical ventilation in ARDS patients predicted mortality well.^19,61  In postsurgical patients, a score higher than 14 at admission to the ICU identified the frailest patients with 90% specificity and predicted a prolonged need for postoperative respiratory support.⁶² 

Quantitative lung ultrasound can also be used to make weaning from mechanical ventilation more successful: computed at the end of a successful spontaneous breathing trial, the lung ultrasound score can identify patients at high risk of extubation failure.^63,64  It has been suggested that the increase in lung ultrasound after a spontaneous breathing trial corresponds to weaning-induced lung edema, thus serving as an indirect measure of increased left ventricle filling pressure.⁶⁵  However, in a cohort of elderly patients with frequent underlying diastolic dysfunction, the increase in lung ultrasound score after a spontaneous breathing trial was not necessarily caused by increased left ventricle filling pressure, and in fact an increase in left ventricle filling pressure was not necessarily associated with an increase in extravascular lung water as measured by lung ultrasound.⁶⁶  The physiopathological mechanisms underlying a worsening of lung aeration, and therefore weaning failure, include weaning-induced lung edema but also lung derecruitment caused by unsolved underlying lung disease or inadequate muscle strength. The lung ultrasound score helps detect any deterioration in lung aeration, regardless of the etiology, and is therefore a very accurate tool to identify the failing patient. A simplified approach focused on anterolateral fields has been proposed. A flowchart for approaching the patient to be weaned from mechanical ventilation is set out in figure 6.

In patients with COVID-19, the lung ultrasound score predicts the severity of the disease accurately when compared to computed tomography⁵³ ; a score lower than 13 corresponds to mild disease, whereas a score higher than 23 identifies severe pneumonia at admission. Lung ultrasound score can be used to monitor the progression of the disease and to predict the need for mechanical ventilation.⁶⁷  Pocket-size devices also offer good accuracy in the assessment of lung ultrasound scores⁶⁸  in COVID-19,⁶⁹  becoming a valuable tool as part of the clinical assessment and helping optimize the allocation of clinical resources in this pandemic.

Although it was initially used only to rule out pneumothorax as a complication of central line placement, lung ultrasound is now attracting attention in anesthesia in clinical practice, both inside and outside the operating theater, for quantitative analysis of lung aeration. Lung ultrasound has been used to assess a wide range of perioperative pulmonary complications and to monitor lung aeration changes during general anesthesia. The technical approach is similar to that in the critically ill patient. Perioperative pulmonary complications—such as atelectasis, pneumothorax, bronchospasm, pleural effusions, and aspiration pneumonitis—result in fact in higher mortality and greater need for health care.^70,71  Therefore, early detection and prompt treatment are recommended.

The computation and monitoring of lung ultrasound scores permit early detection of postoperative pulmonary complications after major abdominal surgery.⁷²  Compared to chest x-ray films and clinical examination, lung ultrasound has performed better for detecting respiratory complications (atelectasis, consolidations, alveolar–interstitial syndrome, pleural effusion, and pneumothorax) after cardiothoracic surgery.^73,74  Among these, intraoperative atelectasis is the substratum for postoperative lung dysfunction and further respiratory complications such as pneumonia, weaning failure, and postextubation respiratory failure.⁷⁵  Subclinical atelectasis has also been reported in the dependent parts of the lung immediately after induction of anesthesia. Accordingly, Monastesse et al.⁷⁶  found that the worsening of oxygenation after the induction of general anesthesia was associated with an increase in lung ultrasound score, mainly in the dependent lung regions. A modified lung ultrasound score taking more account of subpleural consolidations showed greater sensitivity in following aeration loss after pneumoperitoneum.⁷⁶  Similar results were obtained in children, where typical lung ultrasound signs (tissue-like pattern, subpleural consolidations, and abolished lung sliding with the presence of the lung pulse) accurately identified anesthesia-induced atelectasis, assessed by magnetic resonance imaging.⁷⁷ 

Although randomized clinical trials have given variable results, it has been suggested that to limit volutrauma and reduce atelectrauma and thus prevent postoperative complications, mechanical ventilation settings must be finely tuned—i.e., based on the use of PEEP, protective lung ventilation, and recruitment maneuvers.⁷⁸  Aeration quantified by repeated lung ultrasound serves to monitor the state of the lung during general anesthesia. In a randomized clinical trial, Généreux et al.⁷⁹  randomized patients to either lung protection or zero PEEP during open gynecological surgery with the primary outcome set as the difference in lung ultrasound score at different perioperative time points. There was significantly less deterioration of lung aeration during general anesthesia in the protective ventilation group.⁷⁹  Moreover, among children under general anesthesia, lung ultrasound–guided recruitment maneuvers significantly reduced the incidence of anesthesia-induced atelectasis.⁸⁰ 

In postanesthesia care, lung ultrasound accurately detected perioperative pulmonary complications in patients presenting hypoxemia. It was reliable for atelectasis (diagnostic accuracy 97.2%), pneumothorax (diagnostic accuracy 96.7%), and pleural effusion (diagnostic accuracy 95.1%), showing good agreement with thoracic computed tomography scans.⁸¹  There was also good correlation between the lung ultrasound score and the quantitative volumetric data of atelectasis after thoracic surgery.⁸² 

Consistent with this, lung ultrasound findings and scores in the postanesthesia care unit after major surgery are associated respectively with patient outcome and perioperative pulmonary complications. Zieleskiewicz et al.⁸²  showed that the incidence of perioperative pulmonary complications was higher in patients with alveolar consolidation and pleural effusion, immediately after extubation in the ICU. Furthermore, pleural effusions were associated with intrahospital mortality, and alveolar consolidation was associated with postoperative mechanical ventilation.⁸³ 

Quantitative lung ultrasound is a fairly recent technique in pediatric and neonatal critical care and uses a scoring system very similar to the original one described in the section “The Critically Ill Patient.”⁸⁴  The pediatric score has been simplified by the reduction of scanned areas from six to three per hemithorax, creating a single lateral zone on each side without investigating posterior fields, but it shares exactly the same lung ultrasound semiology. Thus, this simplified lung ultrasound score goes from 0 to 18.²⁰  This choice was made because the pediatric score was originally proposed for neonates and infants who have a smaller chest, making the divisions used for adults difficult. In addition, they are frequently moved from the supine to prone position (and back) as routine nursing. This makes scanning the posterior lung regions less valuable, at least when the patient has been recently moved. These zones are, however, important for toddlers and older children and also for infants who have been positioned differently after a given time or on purpose for respiratory care reasons. The simplified lung ultrasound score has been validated against several oxygenation measures,⁴⁵  such as computerized gray scale analysis³⁶  and the amount of surfactant available.⁸⁵ 

The lung ultrasound score is calculated using a high-frequency small linear or a microlinear “golf-stick” probe, given the smaller thorax. However, this must be clearly adapted to patients’ sizes, the amount of subcutaneous tissue, and the equipment available.

The lung ultrasound score computation has high interobserver agreement regardless of the operator’s experience.²⁰  Moreover, the interpretation of basic lung ultrasound signs on which the score is based is not affected by the use of different probes and loses only slightly when convex probes are used on neonates by novice operators.⁸⁶ 

The visual interpretation of basic lung ultrasound signs by clinicians and the computerized evaluation have also been compared, and interestingly, the computer analysis was no more accurate than the expert’s.²³  The simplified lung ultrasound score has the same intrinsic limitation as the original and also needs improvement as regards the following: (1) the number and extension of chest zones to be scanned, (2) the size of the tissue-like pattern to be considered, and (3) the possible modifications to be applied in relation to the patient’s greater weight/age.

Like for adult medicine, simpler scores have been proposed, based on the B-lines count or on simple pattern recognition in a whole newborn chest scan. They may be useful respectively to follow up infants with pulmonary edema caused by congenital heart defects³⁸  or for a first evaluation of respiratory failure early in life,⁸⁷  when a more detailed score is not available for lack of training or equipment.

The lung ultrasound score has significant accuracy in predicting the need for surfactants in preterm²⁰  and extremely preterm⁸⁸  neonates with respiratory distress syndrome. For this purpose, an ultrasound-guided surfactant policy called ESTHER (Echography-guided Surfactant THERapy) has been proposed.⁸⁹  Successive studies reported similar findings,^90–93  and a meta-analysis finally confirmed the accuracy of the lung ultrasound score to guide surfactant replacement.⁹⁴ 

The score has also been successfully used to guide respiratory care in infants with evolving bronchopulmonary dysplasia (recently also called chronic pulmonary insufficiency of prematurity)⁹⁵  and to predict bronchopulmonary dysplasia.⁹⁶  Lung ultrasound scores served for monitoring lung aeration and function in extremely preterm infants; when combined with gestational age, they accurately predicted bronchopulmonary dysplasia, starting from the seventh day of life.⁹⁷ 

The availability of lung ultrasound at the bedside, the lack of irradiation, and multiple potential applications for diagnosis and follow-up of chronic disorders make it eminently suitable for internal medicine, cardiology, and rheumatology departments. In the chronic setting, with less time-dependent patients, a more comprehensive approach has been proposed to detect and quantify B-lines,⁴³  with examination of anterior, lateral, and posterior regions of the chest, scanning every available intercostal space, on both sides, reaching up to 28 scans for the anterolateral chest⁹⁸  and up to 72 scans when including posterior fields.¹⁴  Posterior scanning is specially indicated to assess interstitial fibrosis, whereas in all other situations, anterolateral scanning is usually enough, particularly for heart failure.

In chronic settings, too, simplified schemes have been proposed with 8 regions for the anterolateral chest⁴³  and 14 for the global chest.⁹⁹  A quantitative score based on the sum of the visualized B-lines serves to quantify lung congestion, the so-called B-lines score. Usually when there are up to three to four B-lines in a scan, they can be counted one by one. When there are more, they tend to be confluent, and counting becomes difficult; in this situation, the percentage of the video screen below the pleural line occupied by B-lines should be estimated and then divided by 10 (i.e., if approximately 70% of the screen is occupied by B-lines, it would count as seven B-lines; fig. 4).¹⁰⁰  Phased-array and convex probes are the most widely used, depending on availability, with linear probes usually needed only to visualize the pleural line better. It is easy to learn how to detect B-lines: in a recent trial, an entirely web-based program focused on lung congestion, achieving a high correlation coefficient between each trainee and the expert.¹⁰¹ 

The B-lines score strongly correlates with the invasive lung gravimetry wet/dry ratio of the lung tissue in vitro²⁵  and correlates well with quantitative computed tomography and extravascular lung water in the acute patient. It also correlated with laboratory (brain natriuretic peptide) and echocardiography findings in chronic heart failure.

When a cross-evaluation was applied to the lung ultrasound score and amino-terminal pro-brain natriuretic peptide in an outpatient heart failure population, a 15–B-line cutoff reliably assessed decompensation, confirmed by amino-terminal pro-brain natriuretic peptide values, chest x-ray films, and a 6-min walking test.²⁷  This has also been employed to assess prognosis.¹⁰² 

Patients with end-stage chronic kidney disease undergoing hemodialysis have a high prevalence of lung congestion because of the increase in extravascular lung water. The number of B-lines before and after treatment can change: B-line resolution seems to occur in real time as fluid is removed from the body^24,35  and is also related to weight loss and smaller vena cava diameter after dialysis.²⁶  In these patients, the decrease in the number of B-lines correlates with the increase of thoracic impedance, indicative of a drop in lung water content.²⁹  Thus, the B-lines score can safely guide a strategy for dry weight reduction,¹⁰³  to assess pulmonary congestion²⁹  and prevent fluid overload.¹⁰⁴  The score is also associated with clinical outcome in end-stage renal disease patients with cardiovascular disease.¹⁰⁵ 

Lung ultrasound can detect interstitial lung diseases in a variety of conditions. Many articles report good correlation with computed tomography, especially in patients with connective tissue disease.¹⁴  The literature shows considerable differences in probes, scanning techniques, scoring systems, and interpretation.¹⁰⁶  However, overall there is evidence that lung ultrasound helps in the diagnosis of interstitial lung diseases, and the number of B-lines correlates well with computed tomography,^33,80  pulmonary function tests, and disease severity scales.^30,107  Lung ultrasound offers 100% sensitivity and negative predictive value in both early and advanced systemic sclerosis, making it a useful screening tool for interstitial lung disease. The B-lines score correlates with the radiological score and can predict deterioration in interstitial lung disease.³⁴ 

Computer-aided diagnosis is rapidly becoming part of lung ultrasound equipment. The computer output is used as a “second opinion,” assisting physicians in image interpretation. The computer algorithm generally consists of several steps, which may include image processing, image feature analysis, and data classification using tools such as artificial neural networks.¹⁰⁸  With lung ultrasound, objective diagnostic assistance could facilitate the interpretation of images. This has the potential to be advantageous in terms of better interobserver agreement, faster data analysis, and applicability to large sets of data.⁴² 

Some seminal studies have examined the feasibility of computer-aided diagnosis for reliable quantification of B-lines in a sample set of lung ultrasound clips^39,109  and provided valuable insight on the clinical meaning of B-lines being strongly correlated with extravascular lung water but weakly with physical density,³⁹  in contrast to computed tomography (fig. 7).

A soft computing-based B-line analysis was initially developed for automated quantitative classification of the severity of pulmonary interstitial syndrome¹¹⁰  but could not differentiate pulmonary edema and fibrosis. A MATLAB (MathWorks, USA)–based algorithm was then developed to process image loops after ultrasound acquisition in a region of interest defined by the user.¹¹¹  That early study confirmed that B-mode images vary considerably in B-line features such as number, brightness, intensity, and persistence during the respiratory cycle, indicating how an algorithm-based approach was able to process considerably more data than was manually possible.

In 2015, a histogram-based software of textural features from echo signals was developed. In spontaneously breathing patients, this method was considerably more accurate than either chest x-ray film or visual ultrasonography in the diagnosis of community-acquired pneumonia, independently of size and distance from the pleural line, also compared with computed tomography.⁴⁰  In mechanically ventilated patients, this analysis provided better diagnostic accuracy for detecting increased extravascular lung water than visual scoring, also compared to the EV1000 clinical platform.⁴¹  In 2018, an algorithm to detect, characterize, and count B-lines was presented.¹¹²  It discriminated healthy from unhealthy lung scans. In 2019, a further algorithm was developed for automatic detection and quantification of B-lines.⁴²  That study demonstrated that a computer-aided scoring algorithm of the percentage of the pleural line affected by B-lines gave the best interobserver agreement among various scoring systems. Finally, in 2020, a new algorithm for specific analysis of the pleural line and the immediate subpleural space, based on the gray-level co-occurrence matrix with a second-order statistical method of texture analysis, showed the potential for helping in the differential diagnosis of pulmonary edemas.¹¹³ 

These studies illustrate an initial step toward the goal of developing an algorithm to compute an automatic, operator-independent score for B-lines. The added value of computer-aided diagnosis lies in the use of a pattern recognition based on an artificial neural network to create a new decision support system tool for processing lung ultrasound images.

Every ultrasound examination is operator-dependent, and adequate training is mandatory. The learning curve for lung ultrasound is steep, so even relatively short training can achieve high concordance between expert and nonexpert operators.^52,101  Automation may further improve this.

The penetration of ultrasound to the pleura is prevented by the interposition of a chest drain, dressing and air, subcutaneous emphysema, and wounds or burns preventing direct contact with the skin. Moreover, lung ultrasound assessment takes only a few minutes and is normally part of the clinical assessment. Finally, it does improve patient management, in terms of time to diagnosis and time to advanced examinations.⁵ 

Lung ultrasound is a powerful tool to integrate the clinical assessment of the lung in multiple fields. Different scoring systems have now been validated to improve the use of lung ultrasound not only as a diagnostic tool but also for monitoring lung aeration. Each approach responds to different clinical needs and should be applied in the appropriate clinical contexts. Computer-aided technology has the potential to make quantitative lung ultrasound assessment faster and more reliable, particularly in nonexpert hands, facilitating the further spread of this technique.

Support was provided solely from institutional and/or departmental sources.

Dr. Mojoli received fees for lectures from GE Healthcare (Boston, Massachusetts), Hamilton Medical (Bonaduz, Switzerland), and SEDA Società Per Azioni (Trezzano sul Naviglio, Italy), outside the current work. A research agreement is active between University of Pavia and Hamilton Medical. Dr. Mongodi received fees for lectures from GE Healthcare, outside the current work. Dr. De Luca received travel grants and lecture fees from Philips Healthcare (Amsterdam, The Netherlands) and research assistance from GE Healthcare, outside the current work. Dr. Gargani received consultancy fees from GE Healthcare and had financial relationships with Philips Healthcare and Caption Health (Brisbane, California), outside the current work. The other authors declare no competing interests.