A Preliminary Structural Analysis Of The Upper Airways Of Healthy And Moderate Persistent Asthmatic Patients

INTRODUCTION Three-dimensional reconstructions of the human lung based on medical imaging modalities such as High Resolution Computed Tomography (HRCT) are frequently used to generate anatomical models of the lung1,2,3. However, even though an important purpose of this line of research is the prediction of gas flow and aerosol deposition4,5 in diseased lungs, not much has been published on the description of airway-tree geometric remodelling due to disease3,6. This study is a preliminary report on the 3D-reconstruction of the morphology and topology of the upper airways of 7 healthy and 6 moderate asthmatic adult patients. METHODS Healthy volunteers and asthmatic patients in stable state participating in a clinical study (EudraCT #2007-003563-43) underwent an HRCT scan, producing datasets with 512x512 slice resolution and 0.75mm slice thickness. The software used for the image reconstruction and analysis is Pulmonary Workstation 2 (VIDA Diagnostics, Iowa) and Matlab v7.10. RESULTS Figure 1 is an illustration of the inter-subject variability inherent of the human lung and, more specifically, of the Right Upper Lobe (RUL). A difference in relative connectivity can easily be observed between the sub-lobar bronchi RB1, RB2 and RB3, with the existence (or not) of a relatively small intermediate airway between them. Such variability appears to be characteristic in many locations of the upper airway tree. Figure 2 displays a clinically uncharacteristic stenosis in 2 and 3D. It is unclear at this point whether the stenosis is permanently remodelled or a temporary mucous-plug due to inflammation. The number of generation-5 identifiable and segmentable airways is 77.7% for healthy and 55.2% for asthmatics, while for generation-6 this becomes 21.4% and 10.4% respectively. CONCLUSIONS The presence of intermediate airways in trifurcating or semi-trifurcating structures makes the classification of lung morphology by anatomical generation problematic. Therefore, instead of the usual Weibel7 airway classification system, either the Horsfield Order8 or a new airway-generation definition needs to be used. Furthermore, the analysis of the 3D image reconstructions seems to suggest variability in the number of sub-lobar spaces, as opposed to the 10 considered to form each lung by most anatomical atlases. In addition, even though great progress has been realized in the creation of algorithms for automatic airway recognition, the segmentation and measurement of the airway tree still requires significant user intervention. Finally, a measure of the reduced airway lumen in asthmatic patients is given by the decreased number of bronchi of generation-6 that can be delineated. (Figure presented) .