Abstrait

Automatic Lobar Segmentation Algorithm for Pulmonary Lobes from Chest Ct Scans Based On Fissures and Blood Vessels

Poonkodi R, Geetharani M and Gunasekaran R

Lobewise examination of the pulmonary parenchyma is of scientific significance designed for make a diagnosis and monitoring of pulmonary diseases pathologies becomes important . In medical applications the segmentation of the lung lobe segmentation becomes very important to identify the lung lobes and make a treatment for medical applications of particular patient. Image segmentation is the procedure in which the original image is partitioned into homogeneous regions and plays an important role in medical image processing. In lung lobe segmentation methods segmentation of pulmonary becomes a very difficult and challenging issues for CT images, since changeable reflection resolution, noise and feature obtained through dissimilar CT scanners. Furthermore, difference in examination is numerous, and imperfect fracture is frequent, particularly in rigorous lung disease cases. In recent work automatic lung lobe segmentation the segmentation is performed based on the watershed transformation with the intention of obtain an examination of lobar airways and vasculature addicted to description .But segmentation result are fused based on the Euclidean distance measure which becomes less results, in order to solve this issue in this work, an automated novel lung lobe segmentation method is applied to computed tomography (CT) which separates the lung lobes into pulmonary fissure , Vessel and voxles. Measure of medium truth degree (MMTD) is employed for automatic segmentation of lung lobe. MMTD measure the similarity between the original pixels and make use of correlation of the pixel. In initial stage, the lobar markers are determined by the calculation of labeled bronchial tree. In initial stage of the work the pulmonary vessels are detected based on MMTD. In the second step of the work pulmonary fissures is segmented based on the fissure enhancement where eigen values of is determined from Hessian matrix . In third stage two pre-processing steps is applied such as Gaussian smoothening and bronchial tree . Gaussian smoothening is mainly performed to reduce the noises in the input image samples. The bronchial tree is mainly applied to enhance the quality of the input image samples. Cost image is calculated through mixing the information of fissures, bronchi, and pulmonary vessels distance results. The experimentation results of the proposed MMTD- Lobar segmentation is compared with earlier methods and it applied for 20 CT scans through rejection or mild disease.