Chest radiography plays an important role in the detection and diagnosis of disease related to lungs. The chest radiograph specifies the thoracic anatomy and provides high yield, at the low cost [1]. There are some challenges in processing Chest X-ray images. For example, in lung segmentation, the strong edges at the rib cage and clavicle region cause local minima for most minimization approaches. Segmenting the lung apex is also a nontrivial problem because of the changing intensity at the clavicle bone [11]. Examples of normal Chest X-ray and abnormal Chest X-ray i.e., with and without Tuberculosis infection are shown in Figure.1 and Figure.2 respectively.
Computer Aided Diagnosis (CAD) has been popular for Tuberculosis detection and lung …show more content…
These methods have mostly heuristic assumptions and compute approximate solutions that can be far from the global optimum. Therefore, rule based methods are generally used as an initialization stage of more robust segmentation algorithms. Pixel classification-based methods are more general than rule-based segmentation methods []. The segmentation in pixel classification-based method is based on the feature vector for each pixel in the input image and the intensities of the lung region. Deformable models are used in medical image segmentation due to its shape flexibility. Active Shape Models (ASM) and Active Appearance Models (AAM) have been applied to segment the lung region. Although Active Shape Models and Active Appearance Model approach have become popular for biomedical applications, they have several limitations and shortcomings including: (i) they can become trapped at local minima in chest x-rays due to high contrast and strong rib cage edges, (ii) segmentation performance relies on the accuracy of the initial model to the actual boundary, and (iii) they have many internal parameters which produces highly variable
Computer Aided Diagnosis (CAD) has been popular for Tuberculosis detection and lung …show more content…
These methods have mostly heuristic assumptions and compute approximate solutions that can be far from the global optimum. Therefore, rule based methods are generally used as an initialization stage of more robust segmentation algorithms. Pixel classification-based methods are more general than rule-based segmentation methods []. The segmentation in pixel classification-based method is based on the feature vector for each pixel in the input image and the intensities of the lung region. Deformable models are used in medical image segmentation due to its shape flexibility. Active Shape Models (ASM) and Active Appearance Models (AAM) have been applied to segment the lung region. Although Active Shape Models and Active Appearance Model approach have become popular for biomedical applications, they have several limitations and shortcomings including: (i) they can become trapped at local minima in chest x-rays due to high contrast and strong rib cage edges, (ii) segmentation performance relies on the accuracy of the initial model to the actual boundary, and (iii) they have many internal parameters which produces highly variable