Detecting the Inferior Thoracic Aperture using Statistical Shape Models

Inferior thoracic aperture (ITA) is a 3D closed contour near the periphery of the diaphragm and the accurate detection of ITA from CT images has many important applications. However, it is very difficult to identify certain parts of the ITA from CT images due to the properties of adjacent anatomic structures and image noise. In this paper, we develop a new method to detect the full ITA contour by combining the following information: (a) parts of ITA that can be easily and accurately detected from CT images, (b) several anatomically-meaningful landmark points near or on the ITA, and (c) a statistical shape model of the ITA. Experiments are conducted on 14 real cases using the leave-one-out technique. We show that the results produced by the proposed method are better than the ones produced by a simple spline-based interpolation method.