Detection of Nodule and Lung Segmentation Using Local Gabor XOR Pattern in CT Images

In recent times, for detection of nodule and lung segmentation many computer aided diagnosis systems have been designed to assist the radiologist. The two main factors which affect the accuracy and effectiveness of the detection of lung cancer are nodules that have similar intensity and that they are connected to a vessel and nodule with typical weak edges. Hence, it is difficult to define the boundaries. In the present work, the main objective is to handle the two above mentioned problems by the CADe system for segmentation and detection of nodules from CT Scan images using LGXP method and Morphological Image Processing. The advantage of using Local Gabor XOR Pattern (LGXP) and modified region growing algorithms are using extensive feature sets such as texture contrast, correlation and shape are extracted. The present work has been analyzed using the data of different subjects with varying ages to help lower the number of omissions and to decrease the time needed to examine the scan by a radiologist. There are five problems that can be associated with CADe for lung cancer detection. The first problem is the detection of noises and loss-less information in the CT image. Due to the presence of noise the performance of the system may degrade. We have taken 10 CT scans or subjects from LIDC-IDRI, which includes 5 different cases and the results found vary in satisfaction.