Non-Invasive Microstructure and Morphology Investigation of the Mouse Lung: Qualitative Description and Quantitative Measurement

Zhang, Lu; Li, Dongyue; Luo, Shuqian

The ribs are bone structures, which had a distinct gray scale in PCI reconstructed CT slices. Threshold-based method was used for the segmentation of ribs. Then an image segmentation method was applied to separate the lung tissues from the background. Since the histograms of our lung images had two peaks and a valley between them, the threshold for image segmentation can be chosen automatically at the bottom of this valley [30] . The other pixels with intensity lower than the threshold were set to zero. After this step, the ribs associated with many background signals considered as noise were removed. The whole lung airways can be segmented from this volume dataset by manually set a gray-level threshold. Due to the complexity of the lung structures and their irregular shapes and similar gray levels on the images, threshold-based image segmentation method was inadequate to separate the bronchial tree from the whole lung airways. In this study, we developed an image segmentation method based on 3D region growing to separate bronchial tree which was a semi-automatic procedure. It started in manually placing the seed point in the section of bronchus as well as setting the intensity threshold. Under the control of the intensity threshold, the growing would stop when the bronchial tree were separated.