Research And Implementation Of True And False Lung Nodules Classification Algorithm Based On Swarm Intelligence

Lung cancer is one of the highest cancer mortality rates. Lung nodule is an early manifestation of lung cancer, and pulmonary nodule detection is an important way of the use of computer-aided diagnosis of lung cancer. Because of the complex of lung tissue and diversity of pulmonary nodules, after CT image pre-processed there is still a lot of nodules. For the difficulty of distinguishing between true and false nodule issues, this paper focuses on the detection process, introduces swarm intelligence optimization methods, designs and implements the classification algorithm of pulmonary nodule. And the classification of pulmonary nodules is discussed from the following aspects.Pulmonary Nodule morphology and texture are diverse, causing the effect of a single feature’s distinguishing bad. This article extracts multiple types of features for pulmonary nodule, including gray level character, texture, gradient characteristics and shape features, and combines two-dimensional and three-dimensional features, resulting in comprehensive description of image attributes the dataset for pulmonary nodules is unbalanced and characteristics dimension is high, so the cost-sensitive support vector machines (SVM) is used, and its RBF kernel function is used to make multidimensional data be mapped to a higher-dimensional space, so the original data inseparable in the low-dimensional space can be divided in higher-dimensional space. And using combination of multiple classifiers for classification, further improves the effect of classification result.Swarm intelligence optimization methods are applied to nodule classification problems. Make use of genetic algorithms, particle swarm algorithms, artificial bee colony algorithm to realize parameter adjustment and feature selection, effectively improve the classification accuracy. This article designs and implements the true and false pulmonary nodule classification algorithms, ensuring the effect of true and false pulmonary nodule detection classification of nodule, with good usability.