Is airway diameter measured accurately on routine axial CT scans ? Comparison with true axial diameter using MPR in children with airway compression owing to pulmonary TB

Airway compression is a common complication of TB lymphadenopathy in children, and the diagnostic workup of patients with suspected tracheal or bronchial stenoses includes bronchoscopy and CT (computed tomography). This process affords the opportunity to study aspects of CT relating to airway stenosis. Axial CT scans produce excellent resolution in the horizontal plane, but the extent of airway disease may be underestimated if only axial images are obtained. An advantage of using multidetector CT (MDCT) is the use of multiplanar reconstruction (MPR) to align the image along the longitudinal axis of the airway. There is also uncertainty if window settings affect the measurement of the airway diameter.


Introduction
Airway compression is a common complication of TB lymphadenopathy in children, 1 and the diagnostic workup of patients with suspected tracheal or bronchial stenoses includes bronchoscopy and CT (computed tomography). 2This process affords the opportunity to study aspects of CT relating to airway stenosis.Axial CT scans produce excellent resolution in the horizontal plane, but the extent of airway disease may be underestimated if only axial images are obtained.
An advantage of using multidetector CT (MDCT) is the use of multiplanar reconstruction (MPR) to align the image along the longitudinal axis of the airway.There is also uncertainty if window settings affect the measurement of the airway diameter.

Aim
We wished to determine if there was a significant difference between the measurements of compressed airway diameter in the axial plane compared with measurements of diameter using MPR for determining longitudinal axis of the airway; and to evaluate how measurements on lung window settings compare with soft tissue window settings.

Patients and methods
For the study, 34 CT scans of children with central airway compression due to tuberculous lymphadenopathy were sourced from a digitally stored CT database (on CD) kept by a paediatric pulmonologist.The CT scans were all performed in one academic hospital according to a standard protocol, and were evaluated on a radiology workstation where MPR was performed.The maximum stenosis in the trachea, right main bronchus, bronchus intermedius and left main bronchus was measured, if present.The minimum diameter was taken on the axial scan as well as after MPR, providing the true longitudinal axis (Figs 1 A and 1B).All measurements were performed both in soft tissue window (Window 350/Level 50) and lung window (Window 1600/Level 500) settings (Figs 1C and 1D).Intraclass Correlation (ICC) was used for statistical analysis of the data.

Results
Scans of 18 male and 16 female children were evaluated.Their ages ranged from 1 month to 13.25 years (mean age 2.75 years).There were a total of 62 stenoses, comprising 14 in the trachea, 6 in the right main bronchus, 22 in the bronchus intermedius, and 20 in the left main bronchus.The average diameter of the narrowed airway size in the Graph 1 illustrates little difference (ICC agreement 0.950) between the sets of measurements of stenosis in the trachea, with a high consistency value (ICC consistency 0.946).Of the differences in measurements, 64.3% were <0.5 mm, 35.7% were 0.5 -1.0 mm, and none was >1 mm.
Graph 2 illustrates the values of the 6 stenoses measured in the right main bronchus.The ICC values are lower (ICC agreement 0.848 and ICC consistency 0.858) compared with the figures for the trachea, bronchus intermedius and left main bronchus.Of the differences in measurements, 33.3% were <0.5 mm, 33.3% were 0.5 -1.0 mm, and 33.3% were >1 mm.
Graph 3 illustrates little difference (ICC agreement 0.909) between the sets of measurements in the bronchus intermedius, with a high consistency value (ICC consistency 0.907).Of the differences in measurements, 54.5% were <0.5 mm, 36.4% were 0.5 -1.0 mm, and 91% were >1 mm.
Graph 4 illustrates little difference (ICC agreement 0.945) between the sets of measurements in the left main bronchus, with a high consistency value (ICC consistency 0.943).Of the differences in measurements, 60% were <0.5 mm, 35% weres 0.5 -1.0 mm, and 5% were >1 mm.Table I indicates the differences between axial measurements and those after MPR in the longitudinal axis for all the airways: of the differences, 56.4% were <0.5 mm, 35.5% were 0.5 -1.0 mm, and 8.1% were >1 mm.

Graph 3. Comparison of axial measurements v. measurements after MPR in the bronchus intermedius
Graphs 5 and 6 illustrate a high consistency (ICC consistency of 0.972 for tracheal measurements and 0.923 for right main bronchus measurements) between values on soft tissue window and lung window for the trachea and right main bronchus.Lung window measurements in the trachea were 1.58 mm (root mean square error) less than measurements in soft tissue windows and 1.52 mm (root mean square error) less in the right main bronchus.
Graphs 7 and 8 also illustrate diameter measurements taken in the bronchus intermedius and left main bronchus on lung window and were consistently less than measurements in soft tissue windows (ICC consistency 0.889 for bronchus intermedius and 0.923 for left main bronchus).Here there was less difference for smaller diameter airways A C B D Fig. 1A -D

. MDCT of the chest in a 25-month-old boy with compression of the left main bronchus owing to tuberculous lymphadenopathy. A: Standard axial image in soft tissue windows illustrating the measurement at the maximum stenosis in the left main bronchus (3.5 mm). B: Measurement of diameter after MPR along the longitudinal axis of the airway (4.3 mm). Note the calcified lymphadenopathy around the airway. C: Image A in lung windows, measurement 2.2 mm. D: Image B in lung windows, measurement 2.3 mm. SA JOURNAL OF RADIOLOGY • September 2010
(more severe stenoses) compared with differences for larger diameters.On average, the diameters were smaller by 1.4 mm on lung windows compared with soft tissue window at the same location.

Discussion
4][5] The introduction of MDCT scanners has made it possible to acquire high-resolution images of the upper, central and segmental airways within a short acquisition time. 6In patients with airway compression, the advantage of CT over bronchoscopy is that the cause of airway compression can be diagnosed and the airway distal to the obstruction can be evaluated in areas that are usually inaccessible to bronchoscopy.Additional areas of stricture beyond an area of narrowing can also be evaluated. 5,7Chest CT is therefore an indispensable tool for identifying airway stenoses caused by TB lymphadenopathy.
Although axial CT images are sufficient for evaluating most airway abnormalities, there are inherent limitations of axial images for assessing the airways: Fig. 4. Illustration of a partially compressed segment of a tubular structure.There is a higher possibility that cuts through the compressed segment might not be through the centre; therefore, an inaccurate diameter will be measured.There are at least 2 slices that may give a realistic diameter on the non-compressed segment and none in the compressed segment in this theoretical example.
• difficulty in assessing the interfaces and surfaces of airways that lie parallel to the axial plane. 8oth axial CT and CT with MPR are accurate in detecting airway stenoses. 5,9Remy-Jardin et al. stated that MPR did not provide information that was not evident on transverse images.They did, however, find that mild airway narrowing was easier to recognize on MPR and the length of airway narrowing was easier to measure on MPR than on transverse CT images. 9Quint et al. found that CT with MPR occasionally provided information that was not evident on axial images, and advocated the use of MPRs in imaging of the central airways. 5PR to produce images of each of the major airways into its longitudinal axis, so as to obtain a true 'axial' diameter, is timeconsuming as it must be performed manually and individually for each bronchus as these are oriented uniquely in each patient.Measurements in the axial scan plane are more convenient and rapid as this is the routine image plane provided; i.e. there is no standard practice.
From the results of this study, no significant statistical difference was found between axial measurements and measurements taken after MPR providing the longitudinal axis, even in the relatively horizontal left main bronchus.The majority (56.4%) of differences were <0.4 mm with only 8.1% of the differences being >1 mm.Therefore, central airway stenoses are accurately detected and also not significantly overestimated or underestimated on axial images without MPR (Figs 5 A and 5B).It was, however, easier to evaluate mild stenoses and measure the length of stenoses with MPR.
Measurements taken in lung windows consistently measured the diameter of stenoses of the central airways less than soft tissue (1.4 mm on average) (Figs 5C and 5D).The wide window and low-level setting of lung window should make it the desired setting for evaluating the airfilled airways.It is therefore more likely that soft tissue underestimates stenoses.In small-calibre airways, it is therefore important to measure stenoses on lung window to prevent underestimation.
Currently, a coarse parameter of 50% luminal obstruction of a main stem or lobar bronchus is considered a significant clinical indicator for management of children with airway obstruction.In the future, when there is less reliance on bronchoscopy and more reliance on imaging for airway narrowing, more refined measurement may become important to management decisions.

Conclusion
The degree of central airway stenoses is accurately assessed on axial images, obviating the need to perform MPR to obtain a true axial diameter of the airways in children.Measurements taken on soft tissue window consistently underestimate stenoses.

Graph 4 .Graph 5 .Graph 6 .
ICC(agreement)=0.909 ICC(consistency)=0.Comparison of axial measurements v. measurements after MPR in the left main bronchus.Graph 4. Comparison of axial measurements v. measurements after MPR in the left main bronchusICC(agreement)=0.945ICC(consistency)=0.Comparison between measurements in soft tissue and lung windows in the trachea.Graph 5. Comparison between measurements in soft tissue and lung windows in the trachea ICC(agreement)=0.652ICC(consistency)=0.Comparison between measurements in soft tissue and lung windows in the right main bronchus.Graph 6.Comparison between measurements in soft tissue and lung windows in the right bronchus ICC(agreement)=0.558ICC(consistency)=0.

•Graph 8 .Fig 2 .Fig. 3 .
Fig 2. Illustration of axially imaged central airways.The cuts demonstrated are through the trachea, carina and main bronchi.An axial cut through the vertical trachea provides an accurate diameter of this tubular structure.If a more oblique structure is cut axially, the shape is oblong and may give an inaccurate measurement if there is asymmetrical compression.
Fig. 5A -D.MDCT of the chest in a 52-month-old boy with compression of the bronchus intermedius owing to tuberculous lymphadenopathy.A: Standard axial image in soft tissue windows illustrating the measurement of the maximum stenosis in the bronchus intermedius (4.1 mm).Note the subcarinal lymphadenopathy.B: Measurement of the true diameter after MPR along the longitudinal axis (5.0 mm).C: Image A in lung windows measuring 2.3 mm.D: Image B in lung windows measuring 2.3 mm.