Automatic 3D lung nodule detection based on shape filter and region growing

Abstract

Lungs cancer is a fatal disease. However, its early detection increases the chances of survival among patients. An automated nodule detection system provides the second opinion to radiologists in early diagnosis. In the detection of pulmonary nodules, the complex vessel structure similar to lung nodules in grayscale is an important factor that interferes with the accurate detection of pulmonary nodules. In this paper, we proposed a novel automatic 3D lung nodule detection approach in computed tomography studies. Firstly, the lung region is extracted on the basis of the region growing and morphological smoothing. In the next phase, Frangi multi scale enhancement filter based on the eigenvalues of the Hessian matrix is used to enhance the vessel structures with different size and shapes, and then the fuzzy C-means clustering method(FCM) was used to segment and remove the blood vessels. We apply spherical-like filter based on 3D shape index to detect lung nodule on 3D lung parenchyma data. In the last step, Support Vector Machine(SVM) is used for classification. Experimental results show that this method can effectively reduce the impact of blood vessels on the detection of pulmonary nodules and improve the accuracy of detection of pulmonary nodules.