Background:
National guidelines recommend sampling unilateral pleural effusions to characterize the fluid as either transudative or exudative. Currently, the only method to characterize pleural fluid is by collecting a sample through an invasive drainage procedure that has risks of complications and is uncomfortable for the patient. Few noninvasive methods have been studied to differentiate exudative from transudative pleural effusions. Higher pleural fluid echogenicity by ultrasound has been associated with a higher likelihood of an effusion being exudative. By determining if a quantitative measurement of pleural fluid echogenicity can differentiate exudative vs. transudative effusions, it may be possible to avoid unnecessary thoracenteses. However, it is unknown how ultrasound image gain settings may affect measurement of pleural fluid echogenicity.
The purpose of this pilot project was to assess the reliability of measuring pleural fluid echogenicity on a small sample of images using different gain settings. Based on our findings, an image processing protocol may be developed to measure pleural fluid echogenicity and tested on a large sample of patients to assess the correlation between pleural fluid echogenicity and transudative vs. exudative effusions.
Methods:
We retrospectively measured the pleural fluid echogenicity on images with low, moderate, and high gain settings collected from 10 subjects. To quantify the echogenicity, the pixel density was measured using an image processing software (ImageJ) by selecting a gated area manually to isolate the pleural fluid. First, the pixel density was measured for a set of images to assess the frame-to-frame variability. Second, the pixel density of the low, moderate, and high gain images was calculated for all ten sets of images, and the mean differences between the high and low gain images were calculated.
Results:
The mean pixel density did not vary significantly per frame for different gain settings on the first set of test images (Figure 1). However, a significant difference was seen in the mean pixel density for the 10 sets of images with low, moderate, and high gain (Table 1). The mean difference in pixel density between high and low gain settings was 54.50.
Conclusions:
The mean pixel density of pleural effusions as a quantitative measurement of echogenicity varies significantly based on gain settings. Although past studies have concluded that higher pleural fluid echogenicity correlates with a greater likelihood of an exudative effusion, the gain settings must be controlled to test this hypothesis. The next steps are to develop an image processing protocol that controls for gain and determine whether the pixel density correlates with transudative vs. exudative pleural effusions.
