Computed Tomography (CT) scans are nowadays one of the most important diagnostic tools able to detect medical conditions that conventional X-rays cannot. Indeed, CT scans can show a three-dimensional (3D) view of the section of the body under study but to do so, being the CT scan images initially in two-dimension (2D), it is needed a system for automatic three-dimensional (3D) reconstruction of the section. Based on that matter, in this work we formulate realistic numerical models of inflated and deflated lungs in normal and Acute Respiratory Distress conditions using CT images. We performed a 3D reconstruction of the lung through an automatic recognition of the useful area, done in order to help researchers to implement phantom models and to develop non-hazardous and non-invasive microwave imaging techniques for accurate diagnosis of Acute Respiratory Distress Syndrome (ARDS).