Lung Nodule Segmentation Method Based On Hybrid Imaging

The incidence rate of the lung cancer is the highest mortality rate of malignant tumors in the world today. The new case of lung cancer in china at2011is2.26million and the deaths number of the cancer is1.6million. It can be seen that the mortality of the disease is more than3/4. So we can say the mortality rate of the lung cancer is the highest levels in cancer. In recent years, The use of hybrid imaging for the diagnosis of lung cancer medical diagnosis is appeared. The hybrid imaging overcome the lack of nodules diagnostic information of a single image, But this mixed data have huge amount of information, so radiation doctor’s workload increased dramatically. This article intends to address the Hybrid imaging lung nodules problem and propose an automatic segmentation pulmonary nodules method. The method can effectively positioning nodule location, and to lay the foundation for nodular disease diagnosis.This article is relying on the National Science Foundation project based on hybrid imaging of solitary pulmonary nodule computer-aided diagnostic methods. The research content of the articles is mainly from the following two aspects. First of all, the noise elements in a limited amount of photons maximum likelihood expectation maximization (MLEM) reconstruction of PET/CT (positron emission tomography) image research influence and even increase the image features brightness changes. In order to reduce these impacts, this paper presents the anatomical side information in the value of non-the local average algorithm (AMNLM). The algorithm is applicable to the low signal-to-noise ratio of image denoising non-local averaging algorithm (NLM). The algorithm is applied clinical lung lesions, PET/CT image analysis research. Parameter inversion results show that the trade-off in the value of the visual assessment of lesion contrast and noise filtering method, wavelet filtering method compared to traditional NLM method the method has better denoising effect, which proves this method effectiveness.Second, to assist doctors in clinical diagnosis, a solitary pulmonary nodule segmentation method which based on PET/CT (positron emission tomography/computed tomography) was proposed. Firstly, image registration algorithm is used to complete the appropriate process. Then the processed images are segmented and extracted respectively in PET and CT. Having located the metabolic imaging area of the modules in PET, the high metabolic center of the region is regarded as seed points to be mapped to the corresponding positions on the CT using the registration algorithm. Then applying the regional growth algorithm, the nodule images are extracted. The method was used to analyze the PET/CT data. The findings suggest that complete functional and structural images of pulmonary nodules can be obtained, thus demonstrating the accuracy of the method even in the absence of supervision.In this paper, we study the pre-processing of medical images and proposed a denoising method based on the Hybrid imaging. experiments show that the algorithm effectively remove noise while retain lesions detail features; we also proposed a diagnostic PET/CT lung nodule segmentation method followed by experimental achieve the same desired effect, by the processing of the two processes preliminary extraction of the lung nodules region, we can provide a basis for the semantic modeling of the latter step.