Methacholine-Induced Temporal Changes in Airway Geometry and Lung Density by CT: EBCT

This study utilized the FASTRAC C-100 CT scanner (Imatron; South San Francisco, CA), an electron-beam scanner, employing the “high-resolution” mode. The following parameters were used: 512 X 512 matrix, 12.7-cm reconstruction circle (in plane pixel dimension = 0.248 mm), 3 mm-slice thickness, 2-mm-slice increments, 125 kilovoltage peak and 310 mA, and 200-ms scan times. Multiple slices are acquired with table motion between slice scans. Each scan protocol took 25 s to complete. There was no synchronization of scanning to the cardiac cycle.
For each study, the pig was placed in a supine position and vertically centered within the imaging field. With respiration suspended at FRC, a scout scan was obtained. A preliminary 40-level data set was acquired from just above the carina to the approximate dome of the diaphragm. From these 40 levels, 10 levels of interest were selected in the lower lobes to use in the remainder of the study. Pulmonary
Image Selection: The EBCT hard copy images, filmed at lung window level of — 450 Hounsfield units (HU) and a window width of 1,350 HU, were used to select the slices for analysis. By using prominent and defined parenchymal landmarks, such as vascular or bronchial branching points, the same anatomic levels before and after methacholine administration could be matched, ensuring that the same airways at the same levels and the same regions of the lungs were analyzed throughout each experiment.
Luminal Area Measurement: To quantitate the luminal crosssectional area (CSA) of the airway on a CT image, analysis of the matched airways in the selected images was performed using an objective, semiautomatic edge-detection algorithm in a custom-made software package called Volumetric Image Display and Analysis,2 running on a Sun Sparc II workstation (Sun Microsystems; Palo Alto, CA). The use of this algorithm has been fully described and validated previously. In brief, it applies the “half-maximum” principle to the regional CT density values around the perimeter of a user-drawn estimate of the airway edge to adjust the contour to the correct position.
To compensate for potential through plane motion of the airway location from scan to scan, we evaluated the stack of CT slices gathered at each premethacholine and postmethacholine time point to locate the same anatomic landmarks as previously described. To further compensate for any potential error in misselection of slice, we also used one slice above and one slice below the “matched” level (in addition to the matched level) for airway luminal measurements. Airway area was measured five times at each level for the three levels; thus, each airway CSA value reflects an average of 15 measurements. In each pig, five different-sized airways were chosen at baseline scan. These airways were grouped by size (baseline diameter): (1) 8 to 10 mm, (2) 6 to 8 mm, (3) 4 to 6 mm, (4) 2 to 4 mm, and (5) 1 to 2 mm.
Lung Density Measurements: To assess the effects of metha-choline-induced bronchoconstriction on lung density, we measured the density (in HU) of selected regions in the lung. Using a region of interest (ROI) module within the Volumetric Image Display and Analysis software,2 we carefully selected 50 ROIs in one of the baseline slices obtained for each of the seven studies from three pigs. The scan level for analysis was chosen at baseline to have maximal lung CSA and a good selection of airways appropriate for cross-sectional analysis. Each ROI consisted of a 4-mm by 2-mm rectangle that was carefully drawn on the CT image to avoid inclusion of visible large blood vessels and airways and to achieve broad-based sampling distributed throughout the ventral-dorsal and medial-to-lateral aspect of the right lung section (Fig 2). As with airway CSA measurements, in order to compensate for possible positional shifts, one slice above and one slice below the matched slice also were analyzed. Thus, a total of 150 (50 X 3) ROIs were analyzed for each time point (30 s, 2 min, and 4 min) in each experiment. The ROIs were copied from the baseline slices and pasted to postmethacholine scans, so that for each experiment the same ROIs were analyzed at each of the different time points (baseline, 30 s, 2 min, and 4 min). Position of copied regions were adjusted slightly in cases where a major blood vessel or airway migrated into the boundaries of the region so that the final position of the ROI again avoided these structures. The density measurements, which were corrected to measured 100% air content and 100% blood content levels (see below), were then compared before and after methacholine injection.
Figure 2. Fifty ROIs were selected in each baseline scan for the density measurements. Regions were carefully placed to avoid inclusion of visible blood vessels and airways. To compensate for possible positional shifts, one slice above and one slice below also were analyzed.

Category: Lung

Tags: bronchoconstriction, CT, electron beam CT, image processing, lung density, methacholine