Abstract
The authors propose a new filter for enhancing cavity regions of arbitrary size and shape in three-dimensional grayscale images. This filter, which is called a "cavity enhancement filter", enhances the airways in the bronchial region when it is applied to chest X-ray CT images. The authors also propose a new method for extracting airway trees from chest X-ray CT images using the cavity enhancement filter. In this method, airway trees are extracted using a two-stage region-growing method. First, the initial airway trees are extracted using the first region-growing method. Second, the cavity enhancement filter is applied to the voxels in regions neighboring vessels. Finally, an iterative region-growing method is applied to obtain the final airway trees using the CT numbers and the output of the cavity enhancement filter as the conditions for extraction. The proposed method was applied to chest CT images of nine cases. The results showed that the number of extracted airway branches was increased by 1.3-5.3 times in comparison to a region-growing method using CT numbers and that the percentage of FPs in the extracted volumes was less than 0.7%.
