INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The bedside chest radiograph (CXR) remains the most frequent radiologic examination conducted in critically ill patients even though its utility remains a matter of some debate (1, 2). Hall and colleagues (3) reported that only 8% of routine CXRs in ventilated patients revealed major new findings (requiring treatment within 24 h), whereas 30% revealed minor new findings (not requiring intervention within 24 h). Changes in preexisting infiltrates and intravascular volume status accounted for more than 75% of the new findings, two thirds of which were not anticipated by physical or laboratory examinations. The frequency of unanticipated CXR findings may be relatively low; however, it is sufficient to encourage daily bedside CXR, at least for intubated if not for most patients in the intensive care unit. Indeed, multiple studies indicate that the daily CXR is frequently used as a component of the day-to-day evaluation of changes in patient status (1, 4, 5).

Unfortunately, there are very little data that establishes a correlation between changes in the bedside CXR and a change in the patient's clinical condition. This correlation is limited by several technical (nonphysiologic) factors that impact on the appearance of the CXR. Among these, the ventilatory parameters play a prominent role. Changes in lung volume (6) and pressure (6) during mechanical ventilation are known to substantially alter the appearance of the bedside CXR. Zimmerman and colleagues (10) found that increases in tidal volume or the addition of PEEP to the ventilation parameters produced an apparent improvement in the infiltrates on the CXR. More recently, studies have shown that CXRs taken during pressure-support breaths were judged to be indicative of worsened disease compared with CXRs taken during intermittant mandatory ventilation (IMV) breaths (11).

Ventilation conditions have a prominent effect on the appearance of the bedside CXR (12); however, they are seldom noted and more rarely controlled during portable CXRs, despite suggestion by several authorities that CXRs should be interpreted only in the context of the ventilatory parameters (15). Little information is available to the radiologist regarding changes in ventilation or the patient's clinical condition, which only complicates the correlation between the appearance of the CXR and the patient's condition.

The degree of inflation and the pressure in the lung at the instant the CXR is taken is totally dependent on when the radiographer exposes the film during the respiratory cycle. Exposing the radiograph film at the exact peak of inflation may be quite difficult because the inspiratory pause may be exceedingly brief, especially in seriously ill patients who require complex ventilatory modes, or in pediatric patients who typically receive rapid respiratory rates and small tidal volumes. Motion artifact and submaximal inflation are particularly problematic in ventilated patients who often require frequent CXRs (18, 19), prompting us to evaluate the usefulness of tightly controlled timing of the radiograph during the breathing cycle.

Variations in power, exposure time, and focal distance have traditionally disadvantaged portable CXRs. To improve their resolution, manufacturers have increased the power of portable radiographic equipment. Exposure times are shorter* and recent design changes permit more degrees of freedom in positioning the source, which has improved focal distances. These improvements have minimized the errors to an extent that they can frequently be overcome by digitization technology (20). Hence, the correlation between apparent changes on serial CXRs and changes in clinical status is now limited by technical inconsistency attributable to conditions encountered by the radiographer at the bedside rather than by the equipment available. These conditions are marginally within the control of the radiographer, but they nevertheless exert significant impact on the appearance of the CXR, and therefore on its correlation with the condition of the patient.

We used an electronic interface to synchronize the CXR with the inspiratory pause, which controlled many of the variables that impact on the appearance of the bedside radiograph in mechanically ventilated patients. This device can be programmed to maintain consistent ventilatory parameters between serial films by triggering film exposure during a single breath delivered at previous ventilatory parameters. Its technologic development is described elsewhere (26). This study was designed to evaluate the impact on portable CXR appearance when the film exposure was synchronized with the end-inspiratory pause. The interface was used only to synchronize the timing of the film exposure.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

After obtaining approval from the Food and Drug Administration and the Institutional Review Board, paired portable CXRs were obtained on 25 endotracheally intubated patients. The only inclusion criterion was that patients be intubated and mechanically ventilated; no exclusion criteria were applied. Two of the 25 patients enrolled were spontaneously breathing (pressure support ventilation) whereas the remainder received synchronized intermittent mandatory ventilation (SIMV).

When a CXR was ordered for an enrolled patient, two radiographs were taken. One was taken using conventional techniques in which the radiographer attempts to time the exposure with the peak lung inflation; another was taken using the interface, which automatically synchronized exposure of the film with the inspiratory pause. The sequence of films was randomized by coin toss and there was no selection for radiographer (i.e., various radiographers took the radiographs). The radiographer positioned the radiograph machine, the film cartridge (40 inches from the tube), and the patient, and set the exposure time and the intensity. The radiographic plates used were the standard 14 × 17-inch screen (Kodak Lanex, Rochester, NY) with 400-speed film (Konica MGSR, Tokyo, Japan). The radiographic technique used a beam energy (kVp; range, 75 to 85) believed to be optimal for each patient and a 1milliamper second (mA/s) exposure, but no grid (21, 22) was used. The second film was taken within 3 min of the first and movement of the patient was minimized while the cartridge was changed. The radiograph machine was not moved or adjusted in any way between exposures. Hence, the only variable between the radiographs was the method of timing the film exposure (manual versus electronic).

The CXR taken with the interface was synchronized with the reversal of airflow and the change in pressure in the breathing circuit. Airflow is positive on inspiration but becomes negative during exhalation; positive pressure during inspiration becomes less positive during exhalation. These transitions occur precisely at the inspiratory pause of the ventilatory cycle, making it easy to expose the radiograph film at this precise instant. Ventilatory parameters were exactly the same for both radiographs, and patients served as their own controls in this study.

The study was triple-blinded. Patients were not aware of which CXR was obtained using the interface. The interface is equipped with a firing handle identical to that provided with the radiograph machine used for this study (GE Model AMX4; General Electric, Milwaukee, WI), and so the radiographer was also unaware of which technique was used. Paired radiographs from each patient were placed side by side on a viewing scope and reviewed by five board-certified radiologists.

The radiologists were blinded to technique, and were asked to rate each film on a visual analogue scale (0 to 10) in five categories: clarity of lines and tubes, definition of the pulmonary vasculature, degree of inflation, clarity of the diaphragm and tissue margins, and definition of the mediastinum. Each radiologist was also asked to choose which of the two films they preferred.

Preference for conventional versus synchronized films is reported as the number of radiologists who preferred each technique and the total number of films obtained by each technique that were rated as superior by a majority of the five radiologists. Data pertaining to specific criteria are presented for individual radiologists as mean ± SEM. Film preference based on technique was statistically analyzed using Student's paired t test at the 99% confidence interval, whereas each criteria for individual radiologists was statistically evaluated at the 95% and 99% confidence intervals.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Synchronized radiographs were judged to be superior by a majority ( 3) of radiologists in 21 of 25 patients, which was highly significant (p = 0.0001) (Table 1). Interestingly, all five radiologists agreed that overpenetration, exacerbated by increased inflation, explained why the radiographs of three patients taken with the electronic interface were inferior to those taken without it, arguing against a bias among the radiologists for electronically synchronized films, which were obvious because the lungs were better inflated. In another patient in whom the electronic interface was judged to worsen the appearance of the radiograph, some rotation occurred when the film cartridges were changed between the paired radiographs. This rotation actually improved the visibility of one hemidiaphragm on the second film, which was taken conventionally.

A majority of the five radiologists found that four of the five criteria evaluated were significantly improved (p < 0.05) by Student's paired t test (Table 2). Although an ANOVA might seem more appropriate for this analysis, an ANOVA was specifically not used because of the heterogeneous evaluation by the five radiologists. Therefore, although two reviewers rated the same film better for a specific criteria, the discrepancy between the reviewers was great enough to obscure the difference between paired films. Surprisingly, even though 21 of 24 films were felt to be improved by a majority of radiologists, averaged differences across radiologists by criteria were not significantly different (Table 1) because of variability between reviewers (see DISCUSSION).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Electronically synchronized radiographs were rated as superior by a majority of board-certified radiologists in 21 of 25 cases (Table 1). In one of the remaining four sets of radiographs, the patient was inadvertently moved between films. The repositioning resulted in increased clarity of the left hemidiaphragm on the second film, which happened to be the one taken conventionally (randomized sequence). Although we tried to limit patient movement when the film cartridges were changed, we were not successful in this case. We chose not to repeat this set of radiographs because including it was a bias against our hypothesisthat synchronization of the film exposure with the inspiratory pause would be of benefit.

In the other three instances in which the conventional technique resulted in a superior CXR, the radiographs were overpenetrated and the increased lung inflation associated with use of the interface exacerbated this. The fact that all five radiologists in these three cases found synchronized radiographs to be worse than those taken conventionally argues against a bias for electronic interfacing that they could discern because electronically synchronized radiographs were obviously better inflated. Therefore, we believe that the panel of radiologists was adequately blinded despite our concerns that they would know which film was taken with electronic assistance because the inflation of the lungs was consistently better.

A majority of radiologists rated the synchronized CXRs taken with the interface as superior in all but one criteria. We chose Student's single-tailed, paired t test as a statistical measure comparing the interface with conventional technique because there was so much variability between reviewers and between patients. That is, some radiologists might give both CXRs a high rank, whereas others might rank both poorly. However, when the two CXRs from the same patient were compared with one another, those taken with the interface were consistently ranked better by a majority of radiologists in 4 of 5 categories. Hence, although the average of the five radiologists' evaluations of individual criteria by technique differed insignificantly this resulted from wide discrepancy between reviewers. In other words, one radiologist might evaluate a pair of films as poor, whereas another felt they were good. This resulted in a reduction in the average difference between the films as the number of reviewers increased.

Despite the small sample size and large variability between individual radiologists' ratings, the interface improved four of the five criteria evaluated: definition of pulmonary vasculature, the degree of inflation, the clarity of the diaphragmatic margins, and definition of the mediastinum. Visibility of lines and tubes was not altered in the opinion of any of the radiologists. The number of patients with pulmonary pathology making the parenchyma visible on CXR was too small in this series (only four patients) to confirm that synchronizing the film exposure with the inspiratory pause improves this.

Portable CXRs are conventionally done by a radiographer who attempts to synchronize the film exposure with peak inflation. This is relatively simple in patients who can "take a deep breath and hold it." However, in patients who cannot comply with these instructions (e.g., mechanically ventilated patients), synchronizing the exposure with the ventilatory cycle may be difficult. Patients receiving complex modes of ventilation, rapid ventilatory rates, or shallow tidal volumes further complicate this line of sight technique for synchronization because the inspiratory pause may be exceedingly brief, unpredictable, and nearly imperceptible.

There is also a variable delay between depression of the firing button on the radiograph machine and the actual film exposure because the rotors for the anode must come up to speed before the cathode will discharge. This delay is inconsistent between individual portable CXR machines and even between exposures on the same machine, presenting an additional dilemma for the radiographer, who must not only time the film exposure to the peak of inspiration, but also estimate the delay between depressing the firing button and execution of the exposure. Electronic interfacing eliminates this delay completely. When the firing button is pressed, the rotor comes up to speed and a microprocessor begins tracking the ventilatory cycle. The cathode is then discharged at the instant the inspiratory pause occurs. This eliminates any delay between the time the button is depressed and the actual film exposure.

Any device or technique that standardizes the timing of the film exposure with ventilation should improve the consistency if not the appearance of the bedside CXR. During ventilation, the amount of lung parenchyma is constant, but the volume of the lung changes. Because density is defined as mass/volume, the density of the lung tissue will vary during the ventilatory cycle; minimized when the volume is maximized (i.e., at the end of inspiration). Furthermore, the distribution of airflow within the lung is dependent on regional compliance. Therefore, the density of the tissue in any area will be inversely proportional to the compliance in that area (2).

Bed-ridden patients, especially those requiring ventilatory support, must frequently remain supine. In the supine position, lung compliance is essentially uniform in all regions and, therefore, the inspired volume is distributed equally throughout the normal lungs. Synchronization of the exposure with the end of inspiration ensures that the exposure occurs when the lungs are fully inflated and ensures that all of the uninvolved tissue (with similar compliance) has the same radiographic density and the greatest disparity with diseased tissue (which will have a reduced compliance).

The tissue density, and therefore the X-ray penetration, will differ across lung fields whenever the relative inflation varies, which is dependent on the regional compliance. Therefore, the density of the parenchyma in different lung fields will be most homogeneous and minimized at elevated lung volumes. This is true regardless of the patient's position (i.e., supine versus reclined). When the density of normal lung tissue is minimized and homogeneous, areas of the lung in which there is a higher water content (e.g., edema) or less compliance (e.g., consolidated infiltrates) will be more distinct from the surrounding uninvolved parenchyma. Likewise, increased intravascular volume is readily discernible from the relatively engorged pulmonary hili and the prominence of the pulmonary vasculature.

Comparing images taken at different points during the ventilatory cycle (i.e., degree of inflation and tissue density) complicates the diagnosis of interval changes and may in part explain why CXRs often fail to correlate with alterations in the clinical condition of the patient. There is virtually no way for the radiologist to know for sure whether an increase in tissue density from one film to the next has resulted from a change in technique (23), a change in patient position (23), a change in ventilation (3, 6), a change in the disease process (24, 25), or a change in the intravascular volume (8, 26, 27). Clearly, radiologists may note that a change in rotation or inflation has occurred, but there is no way to know if these conditions fully account for the difference in CXR appearance or if the patient's condition has also changed, contributing to the interval change. Knowledge of change in the ventilatory parameters would help to address this, but it presupposes that radiologists are aware and cognizant of the implications of ventilator adjustments on the patient and the radiograph. Very little work has been done to quantify the effect of ventilation changes on the appearance of the CXR, and it is rare that such information is even available to the radiologist.

The interface can be programmed to adjust the ventilator to previous settings for one breath and trigger the film exposure at a specific point during that breath (usually peak inflation). This ensures that serial CXRs are truly technically equivalent with respect to the ventilation parameters or that changes are available for consideration when the radiograph is interpreted. However, this study was designed to evaluate whether simply synchronizing the radiograph with the breathing cycle improved the portable CXR. The interface was not used to adjust the ventilator's settings because radiographs were taken within minutes of each other, but only to time the radiograph with the inspiratory pause in this initial trial. Hence, additional improvement might be gained if the device were used in patients receiving serial CXRs over a prolonged period with different levels of ventilatory support.

In this study, X-rays were triggered by the reversal of gas flow through the endotracheal tube. However, airway pressure, CO₂ content in the exhaled gas, or the electrical signal from the serial port of the ventilator may also be used. Using airflow, pressure, or capnography as the trigger to synchronize the film exposure with respiration can result in an inappropriate exposure (e.g., if the patient coughs). This possibility is circumvented if the film exposure is triggered using the electronic signal from the serial port of the ventilator, which exactly indicates the apneustic pause of a normal ventilatory cycle. Triggering as we did actually executes the exposure at only 98.5% of full inflation.

Use of the serial port has an additional advantage because the ventilator can be programmed to deliver a breath at specified conditions (e.g., sigh) in patients in whom there is no contraindication, analogous to a spontaneously breathing patient taking a deep breath. This allows the portable CXR to be obtained at near total lung capacity rather than at the end-inspiratory volume. Radiographs taken at the total lung capacity minimize tissue density and also the amount of the lung parenchyma obscured by the ribs. Although this might be advantageous, it might also result in more overpenetrated radiographs. Furthermore, even the most transient interruption of optimal ventilatory parameters can result in a profound loss of pulmonary function in patients requiring mechanical ventilation.

Motion artifact, which reduces the quality of radiographs, is also limited when the film exposure is synchronized with the inspiratory pause. Motion caused by ventilation is especially problematic in CXRs because the tissue density changes as the tissue moves. This compounds the distortion of the image because of movement with changes in density. Clearly the brevity of the exposure reduces the impact of motion artifact on the quality of the image. Nevertheless, any distortion that occurs will be worsened by the simultaneously changing lung volume (i.e., tissue density), the rate of which may be regionally dissimilar because of differing compliances in different lung fields. If beam alignment (21, 22) and optical density by digitization were optimized (24), and used together with the interface, the quality of the portable radiograph would be expected to improve dramatically. However, this remains unconfirmed pending further investigation.

Synchronizing the film exposure with the breathing cycle is applicable to imaging techniques that rely on repetitive exposures such as tomography, computed tomography scanning, and magnetic resonance imaging. Motion artifact and variable lung inflation between cuts reduce the quality of imaging considerably in all of these modalities, and rudimentary attempts to overcome this problem have met with limited success (28, 29). Electronic synchronization should improve these methods considerably, but further studies are needed to confirm this.

Conclusions

Synchronizing the film exposure with the inspiratory pause significantly improves the appearance of portable CXRs in mechanically ventilated patients, as judged by a majority of board-certified radiologists, over several criteria. This may be useful in other imaging techniques as well.