1. Introduction

The goal of this project is to separate pulmonary arteries and veins and detect pulmonary embolism (PE) automatically from 3D thoracic Computed Tomography Angiography (CTA) images. PE is defined as a thrombus (or a clot of blood), which forms in a part of the venous system, becomes detached, and lodges in a pulmonary artery after passing through the right chambers of the heart. The position of these emboli varies with the size of the thrombus and ranges from the pulmonary trunk to the sub-segmental arteries. In the majority of cases, patients present with pulmonary emboli that are of different sizes and in multiple locations. [1]

Diagnosis of pulmonary embolism is often described as being difficult for radiologists. In spiral CTA, more than 100 images, sometimes including images obtained using multi-planar reconstruction, are viewed in each case for detection of thrombi in pulmonary arteries. Thrombi are frequently difficult to distinguish from false positives when only sectional images are used. [2] In CTA images, thrombi are generally recognized as dark regions within enhanced pulmonary arteries in CTA images, as shown in the bottom right figure.