A method for the Semi-Automatic Segmentation of the human Tracheo-bronchial tree from medical images of patients with respiratory disease
Publication Date: 2014-Sep-23
The IP.com Prior Art Database
The prediction of gas flow and pharmaceutical aerosol deposition in diseased lungs and the beneficial effects of the gas and the medicine to the patient are dependent on the geometry of the bounding system (TB tree), which, especially in diseased states (e.g., asthma and COPD), is highly irregular. These irregularities need to be qualified and quantified before any predictive model for the effects of the gas and the aerosol can be developed. However, all tools currently available for the visualization and segmentation of the TB tree are built and tested with respect to healthy human lungs, displaying a distinct inability to penetrate deep enough into the TB tree and identify some prominent defects of disease on the tree geometry.
The solution to the above issues involves the following novel steps:
1) The user should check the ending of each pathway of the network (central path) to determine whether this is due to a normal decrease in bronchial tree size or due to the presence of an airway closure (see figure).
2) If the path ending is determined to be a stenosis, the closest discontinued-disconnected pathway (secondary path) should be detected in the image set.
3) A seed is automatically placed in the middle of the secondary path and the segmentation tool is reapplied..