Research On The Three-Dimensional Visualization And Feature Extraction Of Solitary Pulmonary Nodules

In lung disease, lung cancer is one of the diseases which the morbidity and mortality are still increasing currently. Lung cancer’s early performance is solitary pulmonary nodules, usually isolated pulmonary nodules due to its smaller radius easily ignored by doctors. Visible, the early detection of lung cancer plays a vital role on improving the cure rate. So the three-dimensional visualization of lung nodules and feature extraction become the new research hotspot and difficulty. The key is the result of three-dimensional visualization can be more intuitive, more image appear in patients with lung in front of the doctor, more convenient for the doctor from the perspective of the three-dimensional comprehensive lesions of the patients with characteristics were extracted.This topic in the research of national natural science fund "the isolation of pulmonary nodules based on hybrid imaging computer aided diagnosis methods and the structure and function based on medical image computer-aided diagnosis of mixed characteristics of peripheral lung cancer", on the basis of the lesions of the commonly used three-dimensional visualization technology and feature extraction aspect has carried on the exploration, and in the visual aspects of isolated pulmonary nodules on predecessors’ research methods and make improvements to carry on the thorough research, in terms of three-dimension feature extraction, mainly for isolation of pulmonary nodule size characteristics were studied. On this basis, according to the actual situation of the patient makes a deep exploration and experiment research.First, the traditional methods are usually firstly segment of lung nodules from two-dimensional images, then use the method of three-dimensional construction for visualization. These methods have some problems when the pulmonary nodule images began to appear or at the end of the pulmonary nodules in CT image. Information is less easy to be ignored and cause pulmonary nodule segmentation is incomplete, and each image segmentation caused by the error is bigger, so that when the three-dimension visualization of isolated pulmonary nodules are incomplete and miss features. The main solution is: first of all, in lung CT image clustering segmentation method is used to segment the lung parenchyma, after then the segmentation of lung parenchyma adopt the method of surface rendering in three-dimensional visualization, finally on the basis of three dimensional lung parenchyma, improved three-dimensional segmentation method was adopted to realize the three-dimensional visualization of isolated pulmonary nodules, effectively reduce the visual after isolation of pulmonary nodules information loss problem. In addition, in order to make three-dimension visualization after isolation of pulmonary nodules more complete, this article first to the original conducted a series of CT and PET image preprocessing, including image denoising, image enhancement and image interpolation, etc.Then, aiming at the present stage in the process of lung nodules feature selection and classification which are mostly based on two-dimension image characteristics, and ignored the three-dimensional characteristics. In this paper, the three-dimension characters of pulmonary nodule has carried on the extraction and quantification, including volume, surface area, spherical and convex degree. At the same time the character of volume is an important indicator of diagnose benign and malignant. So we propose a more accurate method of measuring pulmonary nodule volume. First, due to the class spheres of isolation pulmonary nodules, calculating the centroid coordinates, then compare the distance between the points in pulmonary nodule and centroid coordinates, take the maximum distance for external sphere radius, with centroid location for, its maximum distance as the radius of circumscribed sphere; Then in the sphere using Monte Carlo method to generate uniform pseudo random points, and pseudo random point position is in the pulmonary nodules, inside and outside, and the number of statistics. The usable floor area ratio method for the isolation of pulmonary nodule size. And were compared through the experiment, the results show that the accuracy of the algorithm in computing the size of the pulmonary nodules and the error of superior performance, etc.