This paper presents the implementation of deep learning back propagation algorithm (BPA) for the segmentation of lung images. Lung images are obtained the internet resource (LIDC database). An artificial neural network with many hidden layers is considered. The training of the ANN is performed by using the steepest descent BPA algorithm. Lung images of many slices of different patients are considered for training the proposed algorithm. Training of the algorithm is done by using the features extracted from the image by using statistical features. The trained algorithms is used to segment the lung image. The segmentation accuracy of the deep learning BPA is more than 95%.
Volume 11 | 07-Special Issue
Pages: 1529-1534