Certain investigations on lung nodule detection and classification in ct images using computer aided diagnosis techniques

Abstract

The detection of lung nodules in Computed Tomography (CT) newlineimages is very important in the computer-aided diagnosis of lung cancer. The newlinelung segmentation from CT images plays a significant role in the screening newlineprocess for lung cancer. This process adds robustness to the changes in the newlineanatomy of lung region within a CT image. Even though there are various newlinesystems available for diagnosing lung cancers, Computer Aided Diagnosis newline(CAD) method is gaining importance. However, automated lung nodule newlinedetection and lung nodule classification based on various stages, face certain newlineproblems due to inhomogeneities, anatomical shape variations and uneven newlinelung boundary conditions that arise in CT Imaging. newlineThe primary objective of this research is to develop efficient CAD newlinemethodologies for fully automated lung nodule detection and classification. It newlinehas been observed from the review of literature that the active contour newlinemethods fail to produce encouraging results for the effective segmentation of newlinelung region. The simple active contour methods are not capable enough to newlinesegment lung regions from the partially homogeneous background pixels. newlineThis constitutes the boundary discontinuities, which require more memory newlinespace and processing time. At times, this method produces unrealistic results newlinein computer based disease diagnosis process. Although assorted solutions newlinehave been provided for filling the semantic gap, they have not yielded newlinesatisfactory results. So it continues to be open for further research. newlineThe core objectives of the research work are (i) to investigate the newlinelimitations of current lung nodule classification methods and to develop new newlineapproaches to overcome the restrictions (ii) to analyze the performances of the newlinethree CAD based techniques proposed for lung nodule detection and newlineclassification. newline