Deriving Lung Perfusion Directly from CT Image Using Deep Convolutional Neural Network: A Preliminary Study

摘要

Functional avoidance radiation therapy for lung cancer patients aims to limit dose delivery to highly functional lung. However, the clinical functional imaging suffers from many shortcomings, including the need of exogenous contrasts, longer processing time, etc. In this study, we present a new approach to derive the lung functional images, using a deep convolutional neural network to learn and exploit the underlying functional information in the CT image and generate functional perfusion image. In this study, ~(99m)Tc MAA SPECT/CT scans of 30 lung cancer patients were retrospectively analyzed. The CNN model was trained using randomly selected dataset of 25 patients and tested using the remaining 5 subjects. Our study showed that it is feasible to derive perfusion images from CT image. Using the deep neural network with discrete labels, the main defect regions can be predicted. This technique holds the promise to provide lung function images for image guided functional lung avoidance radiation therapy.