Cancer is one of the leading causes of death worldwide, with lung tumors being the most frequent cause of cancer deaths in men as well as one of the most common cancers diagnosed in woman. Analysis and evaluation of medical image data such as computed tomography (CT) scans are commonly used to support experts in their diagnosis and are crucial for the selection of further treatment.
The systematic detection of tumors by an automated system is a challenging task, as several sub-steps such as preprocessing, segmentation, feature extraction and classification have to be considered. We present state of the art methods and solutions for each of these sub-steps as well as a novel approach for a complete
automated system for tumor detection and diagnosis in lung CT-Scans. We will explain and compare the different approaches for segmentation and classification used in the context of the SPIE-AAPM Lung CT Challenge.
At last, we briefly present the methodologies and results of our approach. We explain how different image processing methods such as morphological operations, connected component labeling and marker-based watershed transformation can be combined to archive a successful segmentation of lung CT-scans using the open source software library OpenCV, while also discussing challenges and limitations of the proposed method.