A method for extracting lung lobe regions from 3D chest X-ray CT images based on 3D figure decomposition approach

This paper presents a method for extracting lung lobe regions from 3D chest X-ray CT images by utilizing a 3D figure decomposition process. Recent progress of medical volumetric scanner has enabled us to take a very precise volumetric image of a human body. Computer-aided diagnosis system that assists diagnostic process of radiologists is strongly expected to be developed. In a CAD system for the chest, it is very important to understand structures of the chest by a computer. It is required to develop algorithms that automatically segment organ regions of the chest area from input 3D chest CT images. This paper proposes a method for extracting lung lobe regions from 3D chest CT images by utilizing 3D figure decomposition technique. First, we extract lung regions by simple thresholding and morphological operations. The interlobar pleura are extracted by emphasizing sheet structures. Eigenvalues of the Hessian matrix are used for enhancement. Then, we subtract interlobar pleura regions from lung regions. Figure decomposition process based on Euclidean distance is applied for obtaining lung lobe regions. The proposed method was applied to 14 cases of 3D chest CT images. Experimental results demonstrated that the proposed method can extract lung lobe regions even for cases of incomplete fissure or over-extraction of interlobar pleura.