Ultrasound in the management of thoracic disease

Daniel A. Lichtenstein, MD

T he lung has, step by step,found a place in the field ofcritical care and emergencyultrasound. The slow develop-

ment of this discipline is not truly ex-plained, as the techniques themselves arequite simple. Similarly, the concept ofusing ultrasound as a clinical tool for theintensivist (with or without examiningthe lung) has also been surprisingly longto develop. Since 1989, the author hasused an ultrasound machine comprising1982 technology (built much before therecent explosion of technology that favorsminiaturization) and has had the plea-sure to discover a tool permitting an ac-curate whole-body imaging approach tothe critically ill. Although it is thoughtthat the delay that occurred between1982 and today will remain unexplained,this “sleepy giant” is now awake.

Traditionally, although other complextests and devices exist, the practicing in-tensivist has most commonly assessedlung function using physical examinationand auscultation (these simple tools be-ing available since 1810) (1), radiography

(available since 1895) (2), or with com-puted tomography (CT) (available since1972) (3). However, the flaws of thesefamiliar tools are increasingly acknowl-edged. Auscultation’s low accuracy in thecritically ill has recently been highlighted(4), and bedside radiography (typicallyobtained supine) has even greater limita-tions (4–10). Even CT, which has con-tributed to saving countless lives, hassome major drawbacks that may not beinitially apparent. This is further dis-cussed in part 2 (page S253). Basically,the most unstable patients are the verypatients who do not fully benefit from aCT scan. Ultrasound is a tool with at-tributes that have only recently begun tobe appreciated by the greater medical com-munity, in distinction to its use for cardiacconcerns (11). Previously, respectedsources considered that the place for ultra-sound in assessing the lung was limited(12). Despite scientific evidence provingotherwise, this opinion has persisted (13).Only recently has this begun to change.

The scientific principles of ultrasoundlargely arise from the work of Langevin(1915), with additional contributionsfrom Curie and Einstein early in the early20th century. Utilizing this technologyfor medical purposes was proposed in1946 (14) and has since been developedfor more than half a century by otherpioneers (Wild and Howry, 1951, andHenry and Griffith, 1974). Since thesepioneers, ultrasound has become an in-dispensable and cost-effective medical

tool. Since the sentinel studies like theone by Joyner et al. (15) studying pleuraleffusion, the utility of thoracic ultra-sound was largely limited to this singlediagnosis. Recent reviews of the state-of-the-art of lung investigations devoted lit-tle if any space to ultrasound (16, 17).Working in the team of François Jardin,who had pioneered the use of echocardi-ography with his ADR-4000, using a dou-ble working knowledge in intensive careand general ultrasound, and having tocope with critical situations in the heat ofthe nights, we had a privileged place toappreciate and develop hidden potentialsof ultrasound. Although appreciating thecountless advantages of applying generalultrasound for managing the critically illat the bedside (18), we noted that therewere many conditions for which lung ul-trasound proved immensely helpful. Wethus undertook the challenging task toprove that the lung should be consideredas much of a legitimate target as anyother organ with respect to the use ofbedside ultrasound.

The aim of this article will be to reviewand detail the scientific basis that sufficedto disprove the previously incorrectdogma surrounding the field of lung ul-trasound. This work relies on the analysisof the artifacts that air, in the tissues,pleural spaces, or within the lung itself,produces. Thus, the very substance thatwas previously thought to make lung ul-trasound impossible actually forms thebasis of this science. These basic physical

From Service de Réanimation Médicale, HôpitalAmbroise-Paré, Faculté Paris-Ouest, Boulogne,France.

The author has not disclosed any potential con-flicts of interest.

For information regarding this article, E-mail:dlicht@free.fr

Copyright © 2007 by the Society of Critical CareMedicine and Lippincott Williams & Wilkins

DOI: 10.1097/01.CCM.0000260674.60761.85

Using simple and standardized semiology, the lung appearsaccessible to ultrasound, despite previous opinions otherwise.Lung ultrasound allows the intensivist to quickly answer to amajority of critical situations. Not only pleural effusion but alsopneumothorax, alveolar consolidation, and interstitial syndromewill have accurate ultrasound equivalents, the recognition ofwhich practically guides management. Combined with venous,cardiac, and abdominal examination, ultrasound investigation ofthis vital organ provides a transparent overview of the critically ill,a kind of stethoscope for a visual medicine. It is believed that byusing this tool, the intensivist may more confidently manage

acute dyspnea and make emergency therapeutic decisions basedon reproducible data. Further benefits include reduced require-ments for computed tomographic scans, therefore decreasingdelay, irradiation, cost, and above all, discomfort to the patient.Thus, ultrasound of the lung can also be added to the classicarmamentarium as a clinical tool for emergency use. (Crit CareMed 2007; 35[Suppl.]:S250–S261)

KEY WORDS: chest ultrasonography; lung; ultrasound diagnosis;respiratory failure; intensive care unit; pneumothorax; alveolarconsolidation; pleural effusion; pulmonary edema; chronic ob-structive pulmonary disease; interstitial syndrome

S250 Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

properties are described in part 1. Part 2describes the comprehensive range ofacute respiratory disorders amenable todiagnosis with ultrasound. In part 3, thedaily applications of clinical lung ultra-sound are illustrated.

PART 1: ANALYSISTECHNIQUE, REQUIREDMATERIAL, AND NORMALPATTERN

Seven Principles of Lung Ultrasound.The concept of lung ultrasound can bebased on seven principles (19).

1) A simple, unsophisticated ultrasoundmachine is perfectly adequate.

2) The thorax is an anatomic area whereair and water are intimately mixed.From these interactions arise the ar-tifacts. In addition, air and water haveopposite gravitational dynamics (airrises, water descends). It is thus cru-cial to define “dependent disorders”that are water-rich, such as pleuraleffusions and alveolar consolidation,and “nondependent disorders” thatare air-rich, such as pneumothorax orthe interstitial syndrome. One maythen refer to a sky–earth axis.

3) All lung patterns arise from the pleu-ral line.

4) Lung ultrasound is largely based onthe analysis of artifacts.

5) Lung patterns are largely dynamic. A ret-rospective analysis of static images doesnot provide for an adequate analysis.

6) The majority of acute lung disordersabut the lung surface, thus explainingthe wide-ranging feasibility of lung ul-trasound.

7) As the lung surface is extensive (about1500 cm2), constituting the most vo-luminous organ, precise areas shouldbe defined, as is the norm for theabdominal examination. One may askwhere to put the probe. The answer issimple: at the same places as thestethoscope.

Choice of the Ultrasound Unit: A Crit-ical Step. For performing both lung andwhole-body ultrasound, we think thatsimplicity can be favored. The requiredimage resolution has been fully satisfac-tory since 1991. We wrote our first text-book in 1992 using only ADR-4000 fig-ures, which were already sufficient toillustrate the developing field of lung ul-trasound by using a technology from1982. We think the gray-scale analogic

resolution available since 1991 (that wecurrently use) is more than adequate toperform abdominal, cardiac, venous, andcraniofacial (optic nerve) applications.We rarely find this quality in digital sys-tems. Our unit has ideal dimensions forhospital use (30 � 38 cm footprint), be-ing easily portable from bed to bed andfrom floor to floor. A bigger unit wouldbe a hindrance, but even smaller unitscould become paradoxically more cum-bersome than our basic model once theybecome affixed to a cart. Without such acart, although it may be at risk for theft,but more importantly, the actual ultra-sound unit might be placed on the pa-tient’s bed, constituting an infection con-trol hazard. Further questions are whereto put contact product, disinfectants, andinterventional materials? It was maybeunnecessary to await for the current de-velopment of ultraminiature units to pro-vide a whole-body assessment of the crit-ically ill. The ultrasound unit we use hasexactly, in the updated versions, the sameinternal properties (notably image reso-lution) as in its original 1991 version. Wethink that the various Doppler functionsare not truly required to assess the lungsor the venous system (20, 21) or for anadequate hemodynamic assessment. Us-ing a single probe without Doppler capa-bilities simplifies a whole-body diagnosticapproach for the head, heart, and venousdiseases and their evaluation (22). Wethus think that a Doppler evaluationshould be incorporated only if the sim-pler examination does not answer theclinical question. The unnecessary reli-ance on Doppler capability producesdrawbacks, such as decreased spatial res-olution, and increases the complexity ofthe unit and the approach, increasescosts, is maybe not fully harmless (23–25), and may add physical volume to the

bedside unit. Our preferred microconvex5-MHz probe permits a full evaluation ofthe abdomen, small hard-to-access parts(such as the apex of the lungs), compres-sion of the subclavian vein or adductorarea for the femoral veins, and allows allmanner of interventional procedures(Fig. 1). We also avoid linear probes (be-cause human beings are not linear), deepstructures are not explored, and com-pression maneuvers of deep veins arehard to achieve using linear probes.

The need to avoid cross-contamina-tion of the critically ill and to limit nos-ocomial infections is rarely adequatelydiscussed in current practice, yet respect-ing basic principles and practices is vitalto prevent needless morbidity and mor-tality. The use of more than one probe inthe same examination should raise con-cerns regarding the ability to maintain asterile technique. Harsh disinfectantsshould be avoided because they can grad-ually damage the probe head. The flatkeyboard we use can be cleaned, whereaskeyboards full of prominent buttons can-not—a point of prime importance in theintensive care unit (ICU). Our device canalso be immediately switched on, whichis not the case of most digital units. Fi-nally, the coupling gel, an unpleasantpart of ultrasound since its advent, can beavoided with a new non-gel coupler soonavailable. This will greatly enhance thecomfort of ultrasound for both physiciansand patients.

Ultrasound Examination of the Nor-mal Lung. Lung ultrasound is a recentfield of study, and a rigorous approach isrequired to produce consistent results.To obtain the best from the examination,the operator should simply follow theseven principles sequentially.

A brief review of the normal pattern isuseful. As air rises and water descends,

Figure 1. Stage 1. Left, the probe is gently applied on the anterior chest wall in a supine patient (atthe earth level). This defines stage 1. Note the microconvex shape of the probe—a basic requirement.Stage 2 includes zone 1 and zone 2. Right, extension of the examination to stage 3.

S251Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

the position of the patient should be spec-ified. What is dependent in one position isno longer dependent in another. Oneshould define a gravitational, earth–skyaxis and specify the area where the probeis applied. The thorax should be scanneddirectly, avoiding the traditional abdom-inal route, which can lead to erroneousdiagnoses. Exclusive longitudinal scansare desirable, and thus, we think that alinear probe will be a hindrance for thispurpose. The operator must have accessto superficial and deep areas with onlyone probe. Here again, a linear probe willbe a hindrance.

As the lung is the most voluminousorgan of the body, a careful and method-ical examination comprising three basicsteps is desirable. First the thorax has tobe located (in the craniocaudal axis), thenthe lung surface located, and then zoneshave to be defined. The thorax is distin-guished from the abdomen by locatingthe diaphragm, which is a basic land-mark. Once the probe is applied to thethorax, lung sonography will largely con-sist of the analysis of artifacts becauseonly artifacts appear on the screen (Fig.2). However, the upper and lower ribs canalready be identified, casting a frank pos-terior shadow. Between two ribs and typ-ically 0.5 cm deeper (in the adult), aroughly horizontal, hyperechoic line pro-duced by the pleural interface is visible.The pleural line indicates the parietopul-monary interface (i.e., the lung surface).The ribs and the pleural line outline acharacteristic pattern, the bat sign (Fig.2). The bat sign, visible only in longitu-

dinal scan, should be recognized first inany lung examination and considered amandatory first sign to acquire. Like a Gkey in a musical partition, it is a perma-nent landmark of the lung surface.

Precise areas of interest will be definedusing clinical landmarks. The anteriorand posterior axillary lines are practicallandmarks that delineate anterior, lateral,and posterior areas. Each of these areascan be divided into smaller areas. Theseanatomic areas are considered in our ap-proach to lung sonography, which incor-porates four clinical stages of investiga-tion. Stage 1 is defined by examining theanterior chest wall in a supine patient(zone 1) at the earth level (i.e., our dailyconditions of work, under the gravityrules) and is immediately informative re-garding pneumothorax, interstitial syn-drome, or atelectasis such as can resultfrom right mainstem intubation. Stage 1prime defines this same examination per-formed when the patient is half-supine(as small pneumothoraces move towardthe apex). This does not regard trauma-tized patients. In stage 2, the lateral chestwall (zone 2) is added to the anteriorzone, until the bed physically preventsfurther lateral placement of the probe.Stage 2 gives information on substantialpleural effusions, substantial alveolarconsolidations, and phrenic nerve func-tion. A stage 3 examination is performedby slightly moving the ipsilateral shoul-der of the supine patient to position theprobe as posterior as possible withoutmoving the back (in the case of trauma)and to gain a few centimeters of sono-

graphic exploration of the posterior lungfields (zone 3). As the probe is required to“point to the sky” to perform this exam-ination, small probes are mandatory.Small pleural effusions (beginning hemo-thorax for instance) and small alveolarconsolidations, not detected by stage 1and 2 examinations, may be thus de-tected. In a stage 4 (exhaustive analysis innontrauma patient), the patient is posi-tioned laterally, or seated, to study fully theposterior chest wall. In addition, the apex isinvestigated. To optimally compare the ca-pabilities of lung ultrasound with CT, fullstage 4 examinations should be the re-quired standard, yet in most cases, stages 1,2, or 3 answer the clinical questions.

At the pleural line, two important dy-namic and static signs can be described.First is the dynamic normal sign of lungsliding. This is the basic sign of normal-ity. Lung sliding is a kind of dynamictwinkling movement visible at the pleuralline and synchronized with respiration. Itcorresponds to the displacement of thelung along the craniocaudal axis. For ob-jectifying and documenting lung sliding,M mode yields a simple pattern, the sea-shore sign (Fig. 2). With these simplesigns, the use of Doppler is not required.Much could be written about lung slid-ing. Basically, the 2.5-MHz probes equip-ping many echocardiographic-Dopplerunits usually have insufficient image res-olution. Modern units also have dynamicnoise filters or persistence filters. Thesefilters, designed to provide a flatteringimage, can render lung sliding hard orimpossible to detect and must be by-passed. Lung sliding is a relative move-ment alongside the superficial chesttissues, which are motionless. The ampli-tude of lung sliding is maximal at thebases. Very discrete lung sliding shouldcarefully be sought, as any degree of lungsliding has the same meaning (all-or-nothing rule). Further, lung sliding canbe detected even with mechanical venti-lation, morbid obesity, advanced age, orlung emphysema (even with giant bul-lae). It should be noted, however, that ina dyspneic patient, muscular contrac-tions can make lung-sliding analysis dif-ficult, unless the seashore sign is sought.

The second sign is the normal staticsign. Air artifacts normally arise from thepleural line. In general, two diametricallyopposed types can be described: eitherhorizontal or vertical. Several clinicallyrelevant kinds of artifacts exist, and apractical alphabetic classification is re-quired to avoid long descriptions (19).

Figure 2. Normal lung pattern. Left, longitudinal scan of an intercostal space. Only artifacts (ribs andair) are visible. However, between two ribs (vertical arrows), strictly half a centimeter below in theadult, the pleural line is located (upper horizontal arrows). Upper rib, pleural line, and lower riboutline the bat sign. The horizontal lines (lower horizontal arrows) that arise from the pleural linehave clinical applications (the A lines). Right, seashore sign (M mode). A flagrant difference in patternappears on either side of the pleural line (arrows). The motionless superficial layers generatehorizontal lines—the waves. The deep artifacts follow the lung sliding, hence the sandy pattern.

S252 Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

The basic normal sign is a horizontalrepetition of the pleural line recurring atregular intervals, called ultrasound A-linesign (Fig. 2). The B line and some otherartifacts will be further described in thesection on pathologic conditions. Otherartifacts (C, I, J, N, O, S lines . . .) andother subtle signs will not be further de-tailed here.

The normal lung pattern combineslung sliding with a predominance or to-tality of A lines. In a ventilated patientwithout respiratory concerns, the cupolasare usually located one or two spacesbelow the mamillary line. They move to-ward the abdomen at inspiration, with anamplitude of around 10–15 mm.

PART 2: ULTRASOUNDSEMIOLOGY AND CLINICALAPPLICATIONS OF THE MAINACUTE LUNG DISORDERS

According to the second principle oflung ultrasound, the image and artifactpatterns produced are a function of theair/fluid ratios. Pleural effusion containspure liquid. Alveolar consolidation con-tains mainly liquid and very little air.Interstitial syndrome contains mainly airand little liquid. Pneumothorax containspure air.

Pleural Effusion

Fluid pleural effusion is a disordercontaining exclusively fluid and no air.Although detecting this entity with ultra-sound was imagined in 1946 (14) andassessed in 1967 (15), this simple appli-cation is not fully exploited in all institu-tions.

Maybe this application was not exten-sively exploited because radiologists haveeasy access to CT. Whereas pleural effu-sion can be obvious in echogenic pa-tients, it needs standardized diagnosticcriteria in others. Using some not well-known signs, ultrasound accuracy provesnearly as efficient as that of CT (26). Bothtests have better accuracy than the su-pine chest radiograph (4).

Signs. Customarily, a pleural effusionis detected during abdominal examina-tions, using a subcostal approach. We donot use this traditional access. We find itsafer to analyze the pleura directlythrough the intercostal spaces with ashort probe. The effusion should first besought in a stage 2 examination (i.e., lat-erally in a supine patient) at bed level.Substantial effusions are immediately vis-ible. If no effusion is visible, and if moreinformation is required, the examinermay proceed to a stage 3 (posterior chest)examination to detect minimal effusions.The classic anechoic pattern is not a per-fect criterion, although it can be nondi-agnostic in critical cases. Apart from theobvious diagnostic criteria of a dependentfluid image located above the diaphragm,it is possible to add two more subtlesigns, one static and one dynamic, thatwill greatly help in the difficult cases(Fig. 3). One static sign is the sharp sign.A pleural effusion is limited by four reg-ular borders forming the shape of asharp. These borders consist of the pleu-ral line, from where it arises, the upperand lower shadows of the ribs, and thedeep border, which is always regular. Thisborder is assumed to be the visceralpleura and was called the lung line. The

dynamic sign is the sinusoid sign. Itshows the respiratory variation of the in-terpleural distance with inspiratory de-crease (Fig. 4). The sinusoid sign indi-cates the centrifugal shifting of the lungtoward the wall during inspiration. As thelung moves toward a “core–surface” axis,the pattern, on M mode, is a sinusoid.The sharp and sinusoid signs confirm thepresence of pleural effusion with a spec-ificity of 97% when the gold standardused is withdrawal of pleural fluid (26).

With CT as a gold standard, sensitivityand specificity of ultrasound are 93% (4).Minimal effusions can be detected usingultrasound, provided the probe is appliedover the adequate area of the chest. Ex-tremely small effusions are not detectedusing CT, raising the problem of the per-tinent gold standard. An aerated lunglobe will float over the effusion. A consol-idated lobe will swim within the effusion(the jellyfish sign).

Clinical Applications. The rapid bed-side diagnosis of pleural effusions has ob-vious diagnostic and therapeutic implica-tions for the critically ill. Mattison et al.(27) described a prevalence of 62% inmedical ICUs, with 41% of effusions be-ing present at admission. Ultrasound issuperior to radiography in all respects.Ultrasound will detect the effusion, eval-uate its volume, provide information onits nature, and indicate the appropriatearea for a thoracentesis, with better ac-curacy than radiography. Bedside radiog-raphy rarely detects small effusions andcan also miss effusions of up to 525 mL(28). It can also prompt false-positive di-agnoses. Ultrasound is acknowledged asthe method of choice to detect an effu-

Figure 3. Pleural effusion in an intercostal ap-proach. Note one basic static sign, the sharp sign,as the effusion (E) is outlined by four regularborders: pleural line, shadow of ribs, visceralpleura. Note that the lung at this area is notconsolidated, as air artifacts are visible.

Figure 4. Pleural effusion, a basic dynamic sign, the sinusoid sign. The image provided to the left (asin Fig. 3) is not specific to pleural effusion, can be difficult to see in poorly echoic patients, and canbe echoic. In addition, it does not provide any information about its viscosity. The image at right (Mmode) highlights the sinusoid sign, a sign specific to liquid pleural effusion, and indicates a lowviscosity. E, expiration.

S253Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

sion in a supine patient (29). In our ob-servations, one third of ultrasound-visible and easy-to-puncture effusions inventilated patients remained occult to su-pine bedside radiography (26). Bedsideradiography does not provide reliable in-formation on the volume of an effusion.We have no special ultrasound techniquefor measuring the exact volume either,estimating one effusion as between 500and 1000 mL and another as between 15and 30 mL. We think these approxima-tions are sufficient in clinical practice.Other approaches are available (30).However, ultrasound provides informa-tion about the nature of the pleural effu-sion, data that we do not expect fromradiography. The main causes of pleuraleffusions in the ICU are heart failure(35%), atelectasis (23%), pneumonia(11%), and empyema (1% of cases) (27).Theoretically, a transudate is anechoic,an exudate echoic. A liquid with mobileparticles (plankton sign) or septa is sug-gestive of exudate, hemothorax, or puru-lent pleurisy and should be aspirated andformally analyzed (see Fig. 1 on p. S263).When faced with an anechoic effusion, webelieve that it is prudent to perform ul-trasound-assisted thoracentesis wheneverknowledge of the nature of the effusionmight improve the prognosis. This verysimple procedure may make long discus-sions of differential diagnoses irrelevant.

Diagnostic or therapeutic thoracente-sis is not often routinely performed on acritically ill, ventilated patient because ofconcerns regarding the risks. We thinkthat with practice, care, and ultrasoundguidance, thoracentesis can become rou-tine in this situation. With ultrasounddetection, even radio-occult effusions inventilated patients may be safely aspi-rated. The principle is based on the visualapproach rules (26). Precise and repro-ducible criteria are mandatory. Briefly,one must check for an inspiratory en-largement of the interpleural space of�15 mm, with effusion visible at the ad-jacent upper and lower intercostal spaces.The patient can remain in the supineposition in half of the cases. One checksfor the absence of respiratory interposi-tion of a vital organ (lung, heart, liver,spleen). The sinusoid sign has the at-tribute of clearly indicating low viscosityof the pleural fluid, in other words, thepossibility of using a fine needle to min-imize procedural trauma. Thoracentesisshould be done immediately after ultra-sound localization, with the patient re-maining in the same position. A clinical

landmark made in the radiology depart-ment and followed by an aspiration oncethe patient is back in the ICU seems in-adequate. Skinfolds can displace the cu-taneous landmark. All these precautionsare easy to follow, and in our experience,the success rate in ventilated patients is97% (26). Using these criteria is like driv-ing a car: with open eyes and an attentivebrain behind, the risk of an accident islow, whereas the converse is true as well.Complications such as a pneumothoraxvary between rarely (31) and nil (26).Typically, �10 secs are needed to obtain aliquid sample in �88% of cases. With-drawal of pleural fluid should improvethe ventilatory mechanics (32) and assistweaning from the ventilator, amongother benefits.

Alveolar Consolidation

Alveolar consolidation containsmainly fluid and little air. This daily con-cern in the ICU is not always accuratelydetected by bedside radiography. Auscul-tation is sometimes superior to radiogra-phy (4). These limitations may necessi-tate use of a CT scan. However, 98.5% ofcases of alveolar consolidation abut thepleura (33), a mandatory condition for itsultrasound detection. Whereas the loca-tion of pleural effusion, pneumothorax,or interstitial syndrome are rather stan-dardized, the location of an alveolar con-solidation varies with pathogenesis. Alve-olar consolidation is usually dependent,thus being demonstrated by a stage 3examination, often lateral, thus beingdemonstrated with a stage 2 examination,and sometimes anterior, being detectablewith a stage 1 examination. Those casesamenable to expedient diagnosis with thesimple stage 1 examination roughly cor-respond to the middle and upper lobes. Itshould be noted that a subcostal ap-proach often yields ghost artifacts of theliver or spleen (mirror artifacts throughthe diaphragm). This is why, amongother reasons, we do not use this route.Detecting alveolar consolidation is not anew application for ultrasound (14, 34).However, despite these previous descrip-tions, ultrasound has been seldom usedfor this purpose in general.

Signs. Using basic but rigorous termi-nology to define alveolar consolidation,we found a sensitivity of 90% and a spec-ificity of 98% using ultrasound corrobo-rated by CT as the gold standard (33).Apart from some obvious criteria (imagelocated in the thorax, that is, above the

diaphragm, image arising from the pleu-ral line or from an associated pleural ef-fusion, tissue-like pattern, reminiscent ofthe liver), two specific criteria of interestcan be defined (Fig. 5). Analogous to thecritical criteria for pleural effusion, thereis both an important static and dynamiccriterion. The static criterion states thatan alveolar consolidation usually has ir-regular deep boundaries. The superficialboundary is the pleural line or the deepboundary of a pleural effusion, if present.The deep boundary is irregular, as in con-nection with the aerated lung, a patterntherefore different from the lung line.Only when the whole lobe is involved willthe deep boundary be regular. The dy-namic criterion requires an absence ofany dynamic sinusoidal component, thusexcluding pleural effusion as a cause. Inthe case of alveolar consolidation, cranio-caudal inspiratory movement is presentor even impaired (in the most severecases), but no inspiratory centrifugalshift (i.e., from the bottom to the top ofthe screen) should occur in the core–surface axis. This is mandatory for distin-guishing alveolar consolidation frompleural effusion, which are potentially as-sociated but distinct entities and diag-noses.

Many subtle findings can be describedusing ultrasound. The volume can be as-sessed. Abscesses or necrotizing areaswithin the consolidation can be detected(see Fig. 2 on p. S263). Hyperechoicpunctiform or linear images are possiblypresent and indicate air bronchograms(34). These air bronchograms can be mo-tionless or have an intrinsic inspiratorycentrifugal movement, called dynamic airbronchogram, as opposed as static airbronchograms. The dynamic air bron-chogram allows distinction between non-

Figure 5. Massive alveolar consolidation of thelower lobe. Note the air bronchograms (dynamicin dynamic acquisition). Note the homogeneouspattern, indicative of noncomplicated pneumonia(compare with Fig. 2 on p. S263).

S254 Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

retractile (pneumonia) and retractile (at-electasis) consolidations with 100%specificity for diagnosing nonretractileones (35, 36). The absence of satellitelung rockets is suggestive of aspirationpneumonia. Lung sliding is frequentlyabolished.

Late-stage atelectasis yields alveolarconsolidation with static air bron-chograms, shift of neighboring organs,pinching of intercostal spaces, and aboli-tion of lung sliding. The lung pulse is asign available early after single-lung in-tubation, when the lung is still aerated, asin the seconds immediately after intuba-tion. The lung pulse is a vibration visibleat the pleural line, in rhythm with theheart beat, clearly visible because lungsliding is abolished, and objectified withM mode. The heart vibrations are usuallydominated by the lung expansion. Withmarked atelectasis, abolition of lung slid-ing allows the heart beat to be observedmore readily. The lung pulse had a sen-sitivity of 90% for the diagnosis of one-lung intubation in one study (37).

Why Use Ultrasound? The value of ul-trasound follows from the inadequacy ofradiography (4–10). Radiography gives arough summation of consolidation, pleu-ral effusion, and abscesses, whereas ultra-sound accurately distinguishes each dis-order. Ultrasound can have diagnostic(immediate diagnosis of pneumonia in apatient with fever, pain, and normal ra-diograph, for instance), monitoring (pro-gression of acute respiratory distress syn-drome, indication for prone positioning,positive end-expiratory pressure setting),or even therapeutic effect (see “Point-of-

Care Ultrasound: Infection Control in theIntensive Care Unit” in this supplement).

Interstitial Syndrome

Despite being described back in 1994(38) and confirmed since 1997 (39), thisis an area of bedside diagnosis that will benew to many practicing clinicians. Usingthis approach provides information thatis not provided on a bedside chest radio-graph and that has no auscultatory equiv-alent when using the stethoscope. Inter-stitial syndrome seen in the critically ill ismostly due to thickening of interlobularsepta, which generate Kerley lines, andground-glass areas, which are visible onCT scans. The major causes are cardio-genic pulmonary edema and infectiousprocesses. In the interstitial syndrome,predominant air components are mingledwith a minimal amount of fluid. The ul-trasound study of an aerated organ, withdiagnoses based exclusively on the anal-ysis of artifacts, requires the examiner tothink in an abstract manner. We will seehow to diagnose interstitial syndromeand, above all, why.

Signs of the Interstitial Syndrome.The sign is a vertical artifact (a comet-tailartifact) with special features. It arisesfrom the pleural line, is a well-definedand laser-like beam, is dominant (iterases the A lines), spreads up withoutfading to the edge of the screen, and issynchronous with lung sliding. This arti-fact, as described, has been called B line.Several B lines visible in a single view arereminiscent of a rocket at liftoff and havebeen termed “lung rockets,” or B� lines(Fig. 6). Diffuse lung rockets dissemi-nated all over the anterolateral wall de-fine diffuse interstitial syndrome. Thetest is defined as negative when such Blines are absent, isolated, or exclusively

confined to the last intercostal spaceabove the diaphragm, a variant observedin 27% of healthy subjects (39). Diffuselung rockets have a sensitivity and spec-ificity of 93% for the diagnosis of inter-stitial syndrome when compared with ra-diography, and the concordance iscomplete when the gold standard is CT(39). A separation of the artifacts of about7 mm indicates thickening of the inter-lobular septa (B7 lines), whereas a sepa-ration of 3 � 1 mm (B3 lines) is corre-lated with ground-glass lesions (39). Oneor two B lines visible in a single view aredubbed b lines (lower case) and seem tohave no pathologic meaning.

The B line must critically be distin-guished from two other artifacts: the Eand the Z lines (Fig. 7). E lines (E forsubcutaneous emphysema) are long butdo not arise from the pleural line. The Zlines arise from the pleural line like the Blines, but four features allow easy distinc-tion. They are ill-defined, quickly vanish,are independent from lung sliding, anddo not erase the A lines. Z lines are veryfrequent, visible in 80% of patients (40).They should be considered as a parasiteartifact devoid of clinical meaning (40).This generates an important basic rule.Lung artifacts have the characteristic fea-ture that A lines and B lines cannot bevisible at the same location. Lung arti-facts are either A lines or B lines.

What is the structure detected by ultra-sound? The B lines are generated by ele-ments with a high acoustic impedance gra-dient from the surrounding structures,such as fluid surrounded by air (water is anexcellent transmitter, whereas air im-pedes ultrasound). The detected elementsare smaller than the resolution of ultra-sound. They are present at and all overthe lung surface. They are separated fromeach other by �7 mm. They are present

Figure 6. Interstitial syndrome. These verticalcomet-tail artifacts have the specific peculiaritiesof strictly arising from the pleural line, beingwell-defined and laser-like, moving with the lungsliding, spreading to the edge of the screen with-out fading, and erasing normal A lines. This pat-tern defines B lines. Several B lines in a singleview define lung rockets. Diffuse lung rocketsindicate interstitial syndrome. This patient hascardiogenic pulmonary edema.

Figure 7. Some artifacts: E and Z lines. Left, these well-defined comet tails descend to the edge of thescreen. However, the bat sign is absent (as with Fig. 6). This pattern cannot be due to B lines. Thepatient has subcutaneous emphysema with extensive collections of gas between anatomic struc-tures—a condition generating E lines. Right, the ill-defined comet-tail artifacts (three visible here,arrowheads) arise from the pleural line but do not erase the physiologic A lines (arrows) and quicklyvanish without reaching the edge of the screen. These are Z lines.

S255Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

in pulmonary edema, but labile, resolvingon its treatment. All these features (andsome others) are characteristic of thick-ened subpleural interlobular septa, whichperfectly fulfill this description. CT cor-relation has proven that B lines corre-spond to thickened interlobular septa. Anormal septum has a width of 300 �mand cannot be seen using ultrasound. Thethickened septum has a width of 700 �m,a size that remains under the power ofultrasound but allows generation of theartifact. Ultrasound B lines are thus anultrasound equivalent of the familiar Ker-ley B lines (41). The superficial septaalone can be detected using ultrasound.They are indicative of the deeper septalthickening. Acute interstitial syndrome isgenerally diffuse, especially from cardio-genic cause. This explains why the diag-nosis is immediate, the moment theprobe is applied to the chest wall.

Why Use Ultrasound? Initially, the in-tensivist may question the relevance ofdetecting interstitial syndrome (using ul-trasound or any other method). Devoid ofstethacoustic or radiologic signs allowingdiagnosis, the intensivist has likely be-come accustomed to practicing withoutthis information. This does not discountthe fact that this application of ultra-sound and this information may have animmediate effect on the critically ill. Therecognition of diffuse interstitial syndromein emergency situations is virtually equiv-alent to diagnosing acute pulmonaryedema (cardiogenic or permeability re-lated). Detecting B lines rules out pneu-mothorax (42). In a dyspneic patient, de-tecting lung rockets allows immediatedifferentiation between cardiogenic pul-monary edema and exacerbation ofchronic obstructive pulmonary disease.Only a few seconds are required, and apermanent digital record may be ob-tained of the examination; somethingthat is impossible with simple ausculta-tion. The sensitivity of the ultrasounddetection of pulmonary edema in this set-ting is 100% and specificity is 92% (43).Lung rockets are unusual in pulmonaryembolism. Their absence is found with a92% sensitivity (44). Other uses such asdistinction between lesional and cardio-genic pulmonary edema, morphologicanalysis of acute respiratory distress syn-drome, qualitative assessment of the oc-clusion pressure, and measuring lungfluid or lung compliance are under inves-tigation.

Pneumothorax

Pneumothoraces contain pure air andno fluid. Can ultrasound detect air (aclassic foe to ultrasound) within an air-containing area? Numerous studies havenow conclusively proven the answer to beyes, provided one more step is made to-ward abstraction and provided that arti-facts are accepted as providing clinicalinformation. This indication for immedi-ate bedside diagnosis has a marked ad-vantage in both an accuracy and timeli-ness to that of radiography, especially inthe supine patient. After adequate train-ing, any intensivist will be able to ruleout pneumothorax in a few seconds andwill need �1 min to rule it in.

Pneumothoraces remain common inthe critically ill, from initial traumaticinjuries, iatrogenic procedures, or ac-quired from illness or from barotrauma.They may quickly be life threatening (45).Bedside radiography misses a large per-centage of cases (9, 46–48), even tensionpneumothoraces (49); thus, this situationoften requires CT for confirmation, timepermitting. Bedside radiographs, evenwhen they show the pneumothorax, are apoor indicator of its volume. However, CTcannot be routinely used for this indica-tion either. The excessive use of CT willlead to over-irradiation and increasedcosts and will subject patients to the risksof medical transport, whereas the seriousconsequences may occur if pneumotho-rax is overlooked. We believe that usingultrasound is an extremely simple way toresolve this quandary.

Signs. Pneumothorax semiology mayappear abstract, as it refers exclusively to

artifact analysis. It may also appear com-plex, as several signs have to be investi-gated. However, after minimal training inacknowledged centers, the signs are per-fectly reproducible. Pneumothorax is a“nondependent” semiotic. It should besought first at the anterior and lowerarea, as 98% of significant pneumothora-ces are at least anterior and inferior insupine patients (50). This easy-to-investi-gate location is fortuitous. Many signs areavailable, three covering the majority ofsituations. All our studies have been per-formed with CT as a gold standard.

Lung sliding should be sought in area1. Absent lung sliding is a basic and ini-tial step for the diagnosis, which was ac-tually first described in horses (51). Lungsliding allows pneumothorax to be confi-dently discounted, in a matter of seconds,because the negative predictive value is100% for the diagnosis of pneumothorax(52). With a pneumothorax, a strikingabsence of motion arising from the pleu-ral line is observed instead of the familiarlung sliding. Sensitivity is 100% (whennonfeasible cases are not considered).The abolition of lung sliding can be ob-jectified in M mode, which gives a char-acteristic pattern, the stratosphere sign(Fig. 8). One can use Doppler, but it isnot essential. One can also use a linearprobe, but our microconvex probe, com-bined with our black-and-white technol-ogy, consistently permits a full investiga-tion of this area (and of the deep lung andwhole body).

The role of ultrasound for detectingabolition of lung sliding is increasinglybeing described (53, 54). However, absent

Figure 8. Pneumothorax and stratosphere sign. The complete abolition of lung sliding can beobjectified using M mode (right). Exclusively horizontal lines are displayed, indicating completeabsence of dynamics at the level of, and below, the pleural line (arrowheads), a pattern called thestratosphere sign. Note the absence of a B line at left.

S256 Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

lung sliding is extremely frequent in crit-ically ill patients. Specificity, which is91% in a general population (52), falls to78% when patients are all critically ill(55) and occurs in up to 60% when acuterespiratory distress syndrome patientsare studied. In dyspneic patients seen inthe emergency room, abolished lung slid-ing has a positive predictive value of only27% for the diagnosis of pneumothorax,(unpublished data). Thus, absent lungsliding does not mean pneumothorax.Countless other situations yield abolishedlung sliding: jet or high-frequency venti-lation, massive atelectasis (includingone-lung intubation), acute pleural sym-physis (inflammatory adherences), severefibrosis, phrenic nerve palsy, cardiopul-monary arrest, simple apnea, and inap-propriate operator technique (abusive useof persistence filter, inappropriate probe,etc.). Paradoxically, lung sliding is mostoften abolished precisely in the patientwho both is at maximal risk for pneumo-thorax and who will not tolerate it phys-iologically. We should reiterate that ab-sent lung sliding is not specific topneumothorax. By combining the assess-ment of lung sliding with other signs,however, the effectiveness of ultrasoundis improved.

From the pleural line in stage 1 ariseexclusively horizontal artifacts, A lines(Fig. 8). No B line is visible, a patterncalled the A-line sign. It should be noted,however, that Z lines are very frequentlyvisible. We believe that the use of linearprobes usually prevents correct recogni-tion and distinction of B, Z, and A lines, aserious concern. In fact, this distinctionrelies on deep analysis, which linearprobes usually do not achieve, whereasthe probe we use adequately studies su-perficial and deep areas. The A-line sign,which is 100% sensitive for the diagnosisof complete pneumothorax, is in no casespecific. Specificity is 60% (42). Whatmatters is that the slightest B line allowsprompt ruling out of pneumothorax (42).As we saw that B lines arise from the lungalone, this finding is logical. This is pre-cious information in numerous cases inwhich lung sliding is absent.

The lung point is a specific sign thatallows pneumothorax to be confirmedand that confidently indicates those pa-tients who will benefit from chest tubeplacement in an extreme emergency.When a profile suggestive of pneumotho-rax (A lines with absent lung sliding) isdetected on stage 1, the probe graduallymoves to the lateral areas, until it finds a

fleeting, sudden inspiratory visualization,of either lung sliding or B lines, in anarea where abolished lung sliding andexclusive A lines were previously re-corded. This is an all-or-nothing law, cor-responding to whether the lung is in con-tact with the chest wall (Fig. 9). Thespecificity of the lung point is 100%. Itsoverall sensitivity is 66% and falls withmajor pneumothoraces with completelung retraction (55). Interestingly, sensi-tivity for occult pneumothorax is high:79% of pneumothoraces not visible onbedside radiographs are definitely diag-nosed using ultrasound (40). Once again,ultrasound seems to be more accuratethan bedside radiography.

The lung point is a critical sign be-cause it confirms that the abolition oflung sliding is real and not due to tech-nical defaults. In addition, the lung pointprovides indication about the pneumo-thorax volume and evolution if nottreated. A lateral lung point was corre-lated with a 90% need for drainage vs. 8%with anterior lung point (40). Briefly, ananterior lung point indicates moderatepneumothorax (generally radio-occult),whereas a very posterior or absent lungpoint characterizes massive pneumotho-races with complete retraction.

Some Applications. The clinical appli-cations are multiple. Ultrasound hasproven superior to radiography (55). Ul-trasound can thus complete or replaceradiography and decrease use of CT. Therecognition of pneumothorax in an emer-gency is the main application. This haslong been feasible in prehospital medi-

cine (56). We believe that in extremeemergencies, ultrasound will replace ra-diography. Drainage, previously doneblindly in most unstable patients, can bedone confidently using visual approach.The other basic application is ruling outpneumothorax in a few seconds whenmanaging acute dyspnea or cardiac ar-rest, after any chest procedure (subcla-vian catheterism, thoracentesis), or evenroutinely in a ventilated patient. For suchapplications, a bulky device can do moreharm than good. An elegant application isthe possibility of monitoring a pneumo-thorax based on ultrasound alone. All inall, the need for repeated radiographicstudies, which thicken the medical file,increase the hospital budget, and contin-uously irradiate the patient, will be de-creased.

Changes in the volume of a small andespecially occult pneumothorax can bemonitored if the intensivist chooses tomanage the patient conservatively but notinsert a tube. This logic can be pushed to itsextreme when radiographs are wholly un-desirable (pregnancy, children). Concernsabout irradiation, especially in pediatric pa-tients, are being increasingly discussed(57–59). Thus, it is questionable whetherusing CT scanning to follow conservativelymanaged pneumothoraces offers a goodbalance between therapeutic benefit and ir-radiation (60).

Airway Control

The dynamic real-time nature of lungultrasound allows the immediate diagno-

Figure 9. Lung point. On the right (time-motion), a sudden change is visible at the precise locationwhere the collapsed lung, subject to a slight increase in volume during inspiration, reaches the wall.The “sandy” pattern generated by lung sliding instantaneously replaces a pattern formed by horizontallines (arrow).

S257Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

sis of complete atelectasis, such as occursimmediately after one-lung intubation.The detection of absent lung sliding andthe sole presence of a lung pulse has 90%sensitivity for the diagnosis of immediatecomplete atelectasis after one-lung intu-bation (37). From the diagnosis of one-lung intubation, absence of lung pulseallows check radiography to be postponedafter intubation. The ability of ultrasoundto help such procedures is beginning tobe appreciated (61). Other ultrasoundsigns have also been described that mayassist in safely managing the airway.When the patient is correctly intubated,both cupolas of the diaphragm shouldhave the same amplitude. When rightmainstem intubation occurs, the left cu-pola remains motionless, whereas theright one has an abnormally increasedamplitude, often �15–20 mm with usualtidal volumes. The position of the endo-tracheal tube within the upper tracheagives a characteristic pattern using subtleto-and-fro movements of the tube (�1mm to avoid mucosal damage). Visualperception is sufficient (and M mode canobjectify it, yielding a variant of the sea-shore sign). Doppler is also not requiredfor these diagnostic adjuncts. Ultrasoundis also useful for guiding a percutaneous(or surgical) tracheostomy by preciselydefining the anatomic structures thatshould be avoided (62).

Other Applications

Multiple other disorders can be de-tected using ultrasound, providing ascope for this modality limited only bythe imagination of the clinician. The am-plitude of lung sliding is informative re-garding correct lung compliance, pres-ence of pleural symphysis, or as seen,massive atelectasis. Markedly diminishedlung compliance is frequent in acute re-spiratory distress syndrome or massivepneumonia, and ultrasound yields a dy-namic pattern that no other test can de-tect, namely, abolition of lung sliding.The analysis of the interstitial syndrome,areas of ground-glass, consolidated lungareas, and pleural effusions will help indistinguishing cardiogenic from perme-ability-related pulmonary edema. In pul-monary embolism, a normal pattern (“A”profile) in a dyspneic patient is expected.This is the equivalent of the traditionallynormal chest radiograph. In a patientwith acute respiratory distress and nohistory of asthma or chronic obstructivepulmonary disease, this is immediately

suggestive of a pulmonary embolism. Wehave found a sensitivity of 92% for de-tecting the A profile as opposed to the Bprofile, and this accuracy increases to100% if only the B3 profile is considered(44). Some authors have further de-scribed pulmonary infarction (63), a signthat we rarely observe, perhaps becausethe patients seen by the intensivist havesevere pulmonary embolism, a setting inwhich pulmonary infarction has littletime to develop. Lung abscess is also ac-cessible to ultrasound (see “Point-of-CareUltrasound: Infection Control in the In-tensive Care Unit” in this supplement).

Countless disorders can also be foundin the mediastinum. An experienced usercan often avoid immediate referrals toinvasive or time-consuming techniques.Aortic aneurysm or dissection of the tho-racic aorta can often be detected usingour probe that has a small footprint. Bothsimple and complex conditions such asmediastinitis, tracheal stenosis, or accu-mulation of secretions above an ET tubeare nicely documented in many cases(22). Bedside ultrasound can be concep-tualized as comprised of both diagnosticand interventional attributes. For diagno-sis, several signs such as the swirl sign,plankton sign, and lung pulse improveultrasound accuracy in the diagnosis ofconditions like pneumothorax and pleu-ral effusion. Interventional ultrasoundplays a major part in managing condi-tions such as pneumothorax, lung ab-scess, and pleural effusions The dia-phragm is also amenable to study byultrasound. The location, amplitude anddirection of movement, and degree of in-spiratory thickening are all easily as-sessed by ultrasound, and ultrasoundalone. Even a ruptured diaphragm will bebetter documented on ultrasound thanon CT.

PART 3: CLINICALCONSIDERATIONS ARISINGFROM THE USE OF LUNGULTRASOUND

Interesting applications are accessibleby combining the potentials describedabove. To respect the word count, we willsee the role of a simple black-and-whiteunit when compared with radiography andCT, we will investigate a dyspneic patient,define who is interested (which patients,which operators?), and appreciate strongand weak points of the method.

Lung Ultrasound: Answer to the Tra-ditional Quandary of Radiography or CT

in the ICU. CT is often requested by cli-nicians due to the poor accuracy of bed-side radiography (10). Although CT re-mains an invaluable diagnostic tool incritical care medicine, it is currentlytime-consuming and requires patienttransport. Scientific analysis of the po-tential of lung ultrasound shows that ithas an intermediate role between CTscanning and radiography. Ultrasound ingeneral has a near 90–100% accuracy,depending on the application (4, 40). Al-though CT has major advantages of pro-viding detailed, relatively easy to inter-pret images, a good regional overview,and that its use has saved many lives (17),it does have significant drawbacks (22).Because the main acute chest disorderscan be assessed using ultrasound, thequestion arises as to whether it is re-quired to transport such critically ill pa-tients to a CT scan. Apart from the needfor transportation, the time spent (typi-cally 1 hr, all in all, even if the imageacquisition is done in 10 secs), and thedrawbacks of iodine injection, one partic-ular disadvantage should be borne inmind: the irradiation. One CT scan cre-ates as much irradiation as �100 chestradiographs, an increasing concern inyoung women and children (57–59). Thehigh cost of CT is also not an insignifi-cant issue; in daily practice, how manypatients on Earth have access to thismethod? Finally, we think that the ultra-sound detection of occult pregnancyshould be routine and can take a fewseconds. Positive findings may further in-fluence decision making regarding fur-ther ultrasound studies and radiographicimaging.

In the future, bedside radiographymay become redundant; its indicationsshould gradually decrease. Overall, theslight inferiority of ultrasound com-pared with CT in some applications canbe balanced with others for which ul-trasound clearly seems to be superior.This regards spatial resolution, which al-lows detection of septations within pleu-ral effusion (which are never seen in a CTscan) or necrotizing areas in consolida-tions (64). Also, CT is unable to detect allreal-time dynamic signs such as lungsliding, dynamic air bronchogram, phrenicdynamics, among others. By minglingthe points of slight inferiority with thoseof slight superiority, one can envisageultrasound as a credible alternative tochest CT.

Approach to a Dyspneic Patient. Aspulmonary edema, chronic obstructive

S258 Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

pulmonary disease, pneumothorax, pul-monary embolism, and pneumonia yieldparticular patterns, these signs can beinvaluable at the bedside of a dyspneicpatient. Currently, when investigatingacute dyspnea with the usual tools, incor-rect initial diagnoses are frequent (65).The idea of performing lung ultrasoundhere may seem peculiar. However, if anultrasound unit is readily available, theultrasound data will complete or correctboth the clinical and radiologic findingsthat are often nondiagnostic or even mis-leading. In emergencies, ultrasound dataobtained using a simple device withoutDoppler have enabled the physician togive a correct etiological diagnosis in85% of cases, whereas the traditional ap-proach (clinical examination comple-mented by laboratory tests and chest ra-diography) was accurate in only 51% ofcases (65). Timeliness may be life saving,however, and any delay that waiting for aultrasound machine creates cannot be ac-cepted. Therefore, the intensivist shouldhave mastered the ultrasound profiles ofsimple lung, two-dimensional cardiac,and venous pathology (22). The best ex-ample is the challenge of chronic ob-structive pulmonary disease vs. pulmo-nary edema.

Obviously, traditional information canand should be combined with the ultra-sound findings to take the best of each.Physical examination and thorough re-view of the radiographic studies (if per-formed) remain necessary if time per-mits. We believe, however, that withincreased experience, the intensivist willcome to increasingly rely on ultrasoundin the sickest patients.

Some reminders will be useful to ap-preciate the place of lung ultrasound.

A Field to Be Defined: For Whom?Although the critically ill patient in anICU will be the first to benefit from lungultrasound, these benefits also extend tothe emergency room (66), trauma room,and even the prehospital environment.Lung ultrasound has proven its feasibilityin remote areas (56). We believe thatonce these benefits become apparent, pa-tients in cardiology, pneumology, pedia-try of course, anesthesiology, chest sur-gery, and even internal or familymedicine will benefit from these tech-niques.

A Field to Be Defined: By Whom?Lung ultrasound is an opportunity forthose who practice intensive care medi-cine. As no discipline has legitimatelyclaimed, studied, or adopted this field, it

will rightly belong to those who first ded-icate themselves and prove benefit totheir patients. Intensivists are in closephysical contact with the patient, and ap-propriate training should enable us tomaster the technique (67). Clearly, on-site availability of an ultrasound devicewithin the ICU will simplify patient man-agement. We also think that althoughmedicolegal issues must be considered inmany practices, when the patient’s life isin danger, medicolegal issues should berelegated to a secondary concern.

Lung Ultrasound: A Space for Simplic-ity. An issue that likely slowed the recog-nition and acceptance of general ultra-sound is the perception that this is adifficult exercise. As regards lung ultra-sound, paradoxically, we think this is amarked misconception. Appreciatinglung sliding or lung rockets has an ex-tremely short learning curve (4, 33, 67).B lines and the stratosphere sign are es-sentially the most simple signs one canimagine in ultrasound (or even in medi-cine). This focus on simplicity can also beapplied to the equipment required. Theunsophisticated equipment we describeadequately covers whole-body applica-tions. Doppler functions are unnecessary.A cardiac analysis in a dyspneic patientcan be reduced to assessing left ventriclecontractility alone (or to no analysis insome cases), a major advantage for phy-sicians unfamiliar with this discipline.Lung ultrasound feasibility is 98% in ourobservations (68).

Versatility: An Access to the Neighbor-ing Organs. Although the focus of thisarticle, assessing the lung is only a singlefirst step in assessing the critically ill(22). Combining cardiac and lung ultra-sound results in the thorax being consid-ered as a whole. Unexpected diagnoseswill be made in the abdominal (pneumo-peritoneum, mesenteric infarction, etc),cephalic (maxillary sinusitis, intracranialhypertension), and venous areas. Withthe same system, interventional ultra-sound can be liberally performed (22).

The hemodynamic control of an un-stable patient is a classic in the ICU. Manytools (too many?) are available. Some ap-plications (under submission) of lung ul-trasound will help the physician in im-mediate decisions in this field.

Harmlessness, Cost Savings. Irradia-tion is an increasing concern in the ra-diologic literature. Its deleterious side ef-fects in the child and the young womanare now acknowledged. Lung ultrasoundis an elegant way to circumvent this is-

sue. The indication for radiographs andCT scans should progressively decrease.This decrease should yield cost savings.Costs should decrease both due to re-duced immediate complications (such aspneumothorax due to thoracentesis) andfrom remote ones (such as neoplasia as aconsequence of irradiation). In addition,it has been shown that one single appli-cation allows reimbursement of the unitwe describe in �3 yrs (37).

Limitations of Lung Ultrasound. Acomprehensive understanding of the lim-itations of ultrasound is required to makeultrasound the safe and high-precisiontool it is. Prudent operators will promptlyrecognize a limitation and rely on thetraditional diagnostic tools that sufficedin the past. Hindrances to ultrasound canbe organic or artificial. Organic obstaclescan be innate (a poorly echoic patient) oracquired (subcutaneous emphysema,pleural calcifications, obese body habi-tus). Artificial obstacles, mostly dressingsand tubes, can be limited by smart policyand generally dealt with by the attendingclinician who is not afraid to redress awound after the ultrasound assessment.Intraparenchymal lesions (pneumatocele,deep abscess, rare cases of central consol-idations) will escape surface ultrasound.Paradoxically, obesity is not a major hin-drance to lung ultrasound. Operator skillis a familiar limitation. Insufficient train-ing will result in avoidable pitfalls. Non-longitudinal scans, incorrect or looselyheld probes, use of the subcostal route(which can create ghost artifacts mimick-ing effusions or consolidations), disre-gard of the sky–earth axis in terms ofsearching for pleural effusion in a non-dependent area, abusive use of the dy-namic noise filter, and incorrect loca-tion of the diaphragm will all result inerrors that should be decreased withadequate training.

Each application has its limitations.Pleural effusion, if loculated, will notyield the sinusoid sign, and then the di-agnostic criteria for liquid will depend onthe operator’s judgment. The dark ultra-sound lung is a rare pattern in whichnone of the numerous discriminativesigns is available. It usually correspondsto a white radiologic lung and is mostoften due to massive pleural effusion.Here, CT can be useful. B lines alone donot discriminate between acute andchronic interstitial syndrome. Subphrenicfat can mimic alveolar consolidation if carewas not taken to locate the cupola. Lastly,confusing B, Z, and E lines can occur. In

S259Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

cases of subcutaneous emphysema, whichis not always associated to pneumothorax,experienced users will sometimes identifyconserved lung sliding (22). Subcutaneousemphysema can generate E lines, a com-mon pitfall for the beginner who risks con-fusion with B lines. Posterior pneumotho-races will escape anterior analysis but areaccompanied by suggestive anterior pat-terns (22). In dyspneic patients, lung slid-ing is sometimes hard to detect because ofmuscular contractions, but an answer isgenerally forthcoming with careful study.Major dyspnea with intense muscular effortis rare in spontaneous pneumothorax andin that occurring in a ventilated patient. Alung point is not always present, butclearly, a patient who displays chest pain,dyspnea, tympanism, A lines, and absentlung sliding after an invasive procedure islikely the victim of a pneumothorax.

Training in Lung Ultrasound: Issue orStrong Point? Clinical medicine can onlybe mastered by dedicating years to carefulmedical study. Likewise, critical care ul-trasound, which is a discipline unto itself,cannot be learned in a few hours. Wethink the future of lung ultrasound train-ing should be the responsibility of theuniversity, which should involve studentsas early as possible for maximal societalbenefit. The practicing intensivist whodesires to save lives more easily will seethat training in lung and emergency ul-trasound is paradoxically easier to learnthan standard general ultrasound (4, 33,67). There is nothing in common be-tween the recognition of a B line (a fewminutes of training) and a fetal malfor-mation. In practice, the best way for mas-tering lung ultrasound is to spend a pe-riod as resident or fellow in the ICUsregularly practicing lung ultrasound.

The aim of our training center is togive to the operator the keys for usingultrasound alone for diagnosis and forperforming therapeutic actions. Withpractice, life-saving drainage of compres-sive pleurisy or pneumothorax or correctmanagement of an acute dyspnea shouldbecome routine.

CONCLUSIONS

Lung ultrasound constitutes a visualmedicine and provides a transparent ap-proach to the acutely ill, guiding man-agement and care. Although the use ofthis modality has been largely neglectedby the critical care community, its valueto patients is being increasingly demon-strated. Provided minor limitations are

accepted, lung ultrasound seems to haveonly advantages; noninvasive, immedi-ately implemented, highly feasible, easyto execute, and versatile (from bedside toaircraft, from head to feet), it providesdiagnoses with an accuracy superior tothat of radiography and is time saving ina dyspneic patient. Substantial cost sav-ings are possible, global irradiation de-creases, and patients’ comfort increases.Everyone wins, there is no loser. Scien-tific considerations aside, we like tohighlight again this basic advantage: sim-plicity. Sometimes answering with dis-concerting ease questions for which onlysophisticated approaches were hithertoindicated, ultrasound elegantly simplifiesdaily problems encountered in extremeemergencies (69). Symbolizing for somethe stethoscope of tomorrow, ultrasoundis actually a genuine stethoscope of todayif we consider the etymology: scopein (toobserve) and stethos (the chest wall).

ACKNOWLEDGMENTS

I thank, one more time, but neverenough, Enrico Storti and Luca Neri fortheir outstanding achievement in thissmall revolution and also for a pricelessgift, their friendship; and Alan Sustic,who will also find a devoted place in thissection.

REFERENCES

1. Laënnec RTH: Traité de l’auscultation médi-ate, ou traité du diagnostic des maladies despoumons et du coeur. New York, Hafner,1962; Paris, JA Brosson and JS Claudé, 1819

2. Williams FH: The Roentgen Rays in Medicineand Surgery. New York, Macmillan, 1901

3. Hounsfield GN: Computerized transverse ax-ial scanning. Br J Radiol 1973; 46:1016–1022

4. Lichtenstein D, Goldstein I, Mourgeon E, etal: Comparative diagnostic performances ofauscultation, chest radiography and lung ul-trasonography in acute respiratory distresssyndrome. Anesthesiology 2004; 100:9–15

5. Greenbaum DM, Marschall KE: The value ofroutine daily chest X-rays in intubated pa-tients in the medical intensive care unit. CritCare Med 1982; 10:29–30

6. Janower ML, Jennas-Nocera Z, Mukai J: Util-ity and efficacy of portable chest radiographs.AJR Am J Roentgenol 1984; 142:265–267

7. Peruzzi W, Garner W, Bools J, et al: Portablechest roentgenography and CT in critically illpatients. Chest 1988; 93:722–726

8. Wiener MD, Garay SM, Leitman BS, et al:Imaging of the intensive care unit patient.Clin Chest Med 1991; 12:169–198

9. Tocino IM, Miller MH, Fairfax WR: Distribu-tion of pneumothorax in the supine and

semi-recumbent critically ill adult. AJR Am JRoentgenol 1985; 144:901–905

10. Ivatury RR, Sugerman HJ: Chest radiographor computed tomography in the intensivecare unit? Crit Care Med 2000; 28:1033–1039

11. Jardin F, Dubourg O: L’exploration échocar-diographique en médecine d’urgence. Paris,Masson, 1986, pp 3–154

12. Friedman PJ: Diagnostic procedures in respi-ratory diseases. In: Harrison’s Principles ofInternal Medicine. 12th Edition. New York,McGraw-Hill, 1992, p 1043

13. Weinberger SE, Drazen JM: Diagnostic pro-cedures in respiratory diseases. In: Harri-son’s Principles of Internal Medicine. 16thEdition. New York, McGraw-Hill, 2005, pp1505–1508

14. Dénier A: Les ultrasons, leur application audiagnostic. Presse Méd 1946; 22:307–308

15. Joyner CR, Herman RJ, Reid JM: Reflectedultrasound in the detection and localisationof pleural effusion. JAMA 1967; 200:399–402

16. Desai SR, Hansel DM: Lung imaging in theadult respiratory distress syndrome: Currentpractice and new insights. Intensive CareMed 1997; 23:7–15

17. Wyncoll DL, Evans TW: Acute respiratorydistress syndrome. Lancet 1999; 354:497–501

18. Lichtenstein D, Axler O: Intensive use of gen-eral ultrasound in the intensive care unit: Aprospective study of 150 consecutive pa-tients. Intensive Care Med 1993; 19:353–355

19. Lichtenstein D: Lung ultrasound in the crit-ically ill. In: Yearbook of Intensive Care andEmergency Medicine. Heidelberg, Springer,2004, pp 625–644

20. Cronan JJ: Venous thromboembolic disease:The role of ultrasound, state of the art. Ra-diology 1993; 186:619–630

21. Lichtenstein D, Jardin F: Diagnosis of inter-nal jugular vein thrombosis. Intensive CareMedicine 1997; 23:1188–1189

22. Lichtenstein D: General Ultrasound in theCritically Ill. Third Edition. New York,Springer-Verlag, 2005, pp 1–200

23. Taylor KJW: A prudent approach to Dopplerultrasonography. Radiology 1987; 165:283–284

24. Miller DL: Update on safety of diagnosticultrasonography. J Clin Ultrasound 1991; 19:531–540

25. Guidelines of the British Medical UltrasoundSociety, 2000

26. Lichtenstein D, Hulot JS, Rabiller A, et al:Feasibility and safety of ultrasound-aidedthoracentesis in mechanically ventilated pa-tients. Intensive Care Med 1999; 25:955–958

27. Mattison LE, Coppage L, Alderman DF, et al:Pleural effusions in the medical ICU: Preva-lence, causes and clinical implications. Chest1997; 111:1018–1023

28. Collins JD, Burwell D, Furmanski S, et al:Minimal detectable pleural effusions. Radiol-ogy 1972; 105:51–53

29. Doust B, Baum JK, Maklad NF, et al: Ultra-sonic evaluation of pleural opacities. Radiol-ogy 1975; 114:135–140

S260 Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)

30. Vignon P, Chastagner C, Berkane V, et al:Quantitative assessment of pleural effusionin critically ill patients by means of ultra-sonography. Crit Care Med 2005; 33:1757–1763

31. Mayo PH, Goltz HR, Tafreshi M, et al: Safetyof ultrasound-guided thoracentesis in pa-tients receiving mechanical ventilation.Chest 2004; 125:1059–1062

32. Talmor M, Hydo L, Gershenwald JG, et al:Beneficial effects of chest tube drainage ofpleural effusion in acute respiratory failurerefractory to PEEP ventilation. Surgery1998; 123:137–143

33. Lichtenstein D, Lascols N, Meziere G, et al:Ultrasound diagnosis of alveolar consolida-tion in the critically ill. Intensive Care Med2004; 30:276–281

34. Weinberg B, Diakoumakis EE, Kass EG, et al:The air bronchogram: Sonographic demon-stration. AJR Am J Roentgenol 1986; 147:593–595

35. Lichtenstein D, Meziere G, Seitz J: [The dy-namic air bronchogram: An ultrasound signof nonretractile alveolar consolidation]. Ab-str. Réanimation 2002; 11(Suppl 3):98s

36. Lichtenstein D, Meziere G: Ultrasound diag-nosis of atelectasis. Int J Intensive Care 2005;12:88–93

37. Lichtenstein D, Lascols N, Prin S, et al: Thelung pulse: An early ultrasound sign of com-plete atelectasis. Intensive Care Med 2003;29:2187–2192

38. Lichtenstein D: Diagnostic échographique del’oedème pulmonaire. Rev Im Med 1994;6:561–562

39. Lichtenstein D, Meziere G, Biderman P, et al:The comet-tail artifact: An ultrasound sign ofalveolar-interstitial syndrome. Am J RespirCrit Care Med 1997; 156:1640–1646

40. Lichtenstein D, Meziere G, Lascols N, et al:Ultrasound diagnosis of occult pneumotho-rax. Crit Care Med 2005; 33:1231–1238

41. Kerley P: Radiology in heart disease. BMJ1933; 2:594

42. Lichtenstein D, Meziere G, Biderman P, et al:The comet-tail artifact, an ultrasound signruling out pneumothorax. Intensive CareMed 1999; 25:383–388

43. Lichtenstein D, Meziere G: A lung ultrasound

sign allowing bedside distinction betweenpulmonary edema and COPD: The comet-tailartifact. Intensive Care Med 1998; 24:1331–1334

44. Lichtenstein D, Loubiere Y: Lung ultra-sonography in pulmonary embolism. Chest2003; 123:2154

45. Steier M, Ching N, Roberts EB, et al: Pneu-mothorax complicating ventilatory support.J Thorac Cardiovasc Surg 1974; 67:17–23

46. Hill SL, Edmisten T, Holtzman G, et al: Theoccult pneumothorax: an increasing diag-nostic entity in trauma. Am Surg 1999; 65:254–258

47. Kurdziel JC, Dondelinger RF, Hemmer M: Ra-diological management of blunt polytraumawith CT and angiography: An integrated ap-proach. Ann Radiol 1987; 30:121–124

48. McGonigal MD, Schwab CW, Kauder DR, etal: Supplemented emergent chest CT in themanagement of blunt torso trauma.J Trauma 1990; 30:1431–1435

49. Gobien RP, Reines HD, Schabel SI: Localizedtension pneumothorax: Unrecognized formof barotrauma in ARDS. Radiology 1982;142:15–19

50. Lichtenstein D, Holzapfel L, Frija J: [Cutane-ous projection of pneumothorax and impacton the ultrasound diagnosis]. Abstr. RéanUrg 2000; 9(Suppl 2):138s

51. Rantanen NW: Diseases of the thorax. VetClin North Am 1986; 2:49–66

52. Lichtenstein D, Menu Y: A bedside ultra-sound sign ruling out pneumothorax in thecritically ill: Lung sliding. Chest 1995; 108:1345–1348

53. Kirkpatrick AW, Sirois M, Laupland KB, et al:Hand-held thoracic sonography for detectingpost-traumatic pneumothoraces. J Trauma2004; 57:288–295

54. Blaivas M, Lyon M, Duggal S: A prospectivecomparison of supine chest radiography andbedside ultrasound for the diagnosis of trau-matic pneumothorax. Acad Emerg Med 2005;12:844–849

55. Lichtenstein D, Meziere G, Biderman P, et al:The lung point: An ultrasound sign specificto pneumothorax. Intensive Care Med 2000;26:1434–1440

56. Lichtenstein D, Courret JP: Feasibility of ul-

trasound in the helicopter. Intensive CareMed 1998; 24:1119

57. Brenner DJ, Elliston CD, Hall EJ, et al: Esti-mated risks of radiation-induced fatal cancerfrom pediatric CT. AJR Am J Roentgenol2001; 176:289–296

58. Kalra MK, Maher MM, Toth TL, et al: Strat-egies for CT radiation dose optimization. Ra-diology 2004; 230:619–628

59. Berrington de Gonzales A, Darby S: Risk ofcancer from diagnostic X-Rays. Lancet 2004;363:345–351

60. Sahn SA, Heffner JE: Spontaneous pneumo-thorax. N Engl J Med 2000; 342:868–874

61. Chun R, Kirkpatrick AW, Sirois M, et al:Where’s the tube? Evaluation of hand-heldultrasound in confirming endotracheal tubeplacement. Prehospital Disaster Med 2004;19:366–369

62. Sustic A, Kovac D, Zgaljardic Z, et al: Ultra-sound-guided percutaneous dilatational tra-cheostomy: A safe method to avoid cranialmisplacement of the tracheostomy tube. In-tensive Care Med 2000; 26:1379–1381

63. Bitschnau R, Mathis G: Chest ultrasound inthe diagnosis of acute pulmonary embolism.Radiology 1999; 211:290

64. Lichtenstein D, Peyrouset O: Lung ultra-sound superior to CT? The example of a CT-occult necrotizing pneumonia. IntensiveCare Med 2006; 32:334–335

65. Lichtenstein D, Meziere G: Ultrasound diag-nosis of an acute dyspnea. Abstr. Crit Care2003; 7(Suppl 2):S93

66. Rose JS, Levitt MA, Porter J, et al: Does thepresence of ultrasound really affect com-puted tomographic scan use? A prospectiverandomized trial of ultrasound in trauma.J Trauma 2001; 51:545–550

67. Lichtenstein D, Meziere G: [Training of gen-eral ultrasound by the intensivist]. Abstr.Réan Urg 1998; 7(Suppl 1):108s

68. Lichtenstein D, Biderman P, Chironi G, et al:[Feasibility of general ultrasound in the in-tensive care]. Abstr. Réan Urg 1996; 5, 788:SP50

69. van der Werf TS, Zijlstra JG: Ultrasound ofthe lung: Just imagine. Intensive Care Med2004; 30:183–184

S261Crit Care Med 2007 Vol. 35, No. 5 (Suppl.)