Premium
Automatic multiorgan segmentation in thorax CT images using U‐net‐ GAN
Author(s) -
Dong Xue,
Lei Yang,
Wang Tonghe,
Thomas Matthew,
Tang Leonardo,
Curran Walter J.,
Liu Tian,
Yang Xiaofeng
Publication year - 2019
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1002/mp.13458
Subject(s) - segmentation , ground truth , discriminator , artificial intelligence , computer science , deep learning , convolutional neural network , generator (circuit theory) , sørensen–dice coefficient , pattern recognition (psychology) , image segmentation , nuclear medicine , computer vision , medicine , physics , telecommunications , power (physics) , quantum mechanics , detector
Purpose Accurate and timely organs‐at‐risk ( OARs ) segmentation is key to efficient and high‐quality radiation therapy planning. The purpose of this work is to develop a deep learning‐based method to automatically segment multiple thoracic OAR s on chest computed tomography ( CT ) for radiotherapy treatment planning. Methods We propose an adversarial training strategy to train deep neural networks for the segmentation of multiple organs on thoracic CT images. The proposed design of adversarial networks, called U‐Net‐generative adversarial network (U‐Net‐ GAN ), jointly trains a set of U‐Nets as generators and fully convolutional networks ( FCN s) as discriminators. Specifically, the generator, composed of U‐Net, produces an image segmentation map of multiple organs by an end‐to‐end mapping learned from CT image to multiorgan‐segmented OAR s. The discriminator, structured as an FCN , discriminates between the ground truth and segmented OAR s produced by the generator. The generator and discriminator compete against each other in an adversarial learning process to produce the optimal segmentation map of multiple organs. Our segmentation results were compared with manually segmented OAR s (ground truth) for quantitative evaluations in geometric difference, as well as dosimetric performance by investigating the dose‐volume histogram in 20 stereotactic body radiation therapy ( SBRT ) lung plans. Results This segmentation technique was applied to delineate the left and right lungs, spinal cord, esophagus, and heart using 35 patients’ chest CT s. The averaged dice similarity coefficient for the above five OAR s are 0.97, 0.97, 0.90, 0.75, and 0.87, respectively. The mean surface distance of the five OAR s obtained with proposed method ranges between 0.4 and 1.5 mm on average among all 35 patients. The mean dose differences on the 20 SBRT lung plans are −0.001 to 0.155 Gy for the five OAR s. Conclusion We have investigated a novel deep learning‐based approach with a GAN strategy to segment multiple OAR s in the thorax using chest CT images and demonstrated its feasibility and reliability. This is a potentially valuable method for improving the efficiency of chest radiotherapy treatment planning.