ABSTRACT
Pulmonary nodules are the most common manifestation of lung cancer, which is the principal cause of cancer-related deaths [1]. Fast and accurate classiŽcation of the nodules is of major importance for medical computer-aided diagnostic (CAD) systems. A nodule is an approximately spherical volume of higher-density tissue visible in an x-ray lung image. Large malignant nodules (generally deŽned as greater than 1 cm in diameter) are easily detected with traditional imaging equipment and then diagnosed by needle biopsy or bronchoscopy. However, diagnostic options for small malignant nodules are limited due to difŽculties in their accessibility, especially if they are located deep in the tissue or away from the large airways. Therefore, additional imaging and CAD techniques are needed. The popular direction of research in detecting small cancerous nodules is to analyze their growth rate over time. This chapter introduces a new approach to characterize the detected nodules based on their shape.
