Measuring CT Reconstruction Quality with Deep Convolutional Neural Networks

Mayank Patwari, Ralf Gutjahr, Rainer Raupach, Andreas Maier

Abstract: With the increasing use of CT in diagnostic imaging, reducing the clinical radiation dose is necessary for ensuring patient safety. Reduced radiation dose results in quantum noise which adversely affects image quality and diagnostic value. Moreover, obtaining high quality images to act as reference images for image quality assessment is difficult. Therefore, automatic no-reference quality assessment of reconstructed images is necessary to preserve diagnostic image quality, while controlling radiation dose. In this work, we investigate the use of a deep convolutional neural network to measure CT image quality. Our developed metric shows concordance with conventional metrics of CT image quality (\(|r|>\) 0.75, \(|\rho |>\) 0.75). Our metric ranks images in terms of quality highly accurately (\(\tau \) = 0.98). We measure noise textures and levels not present in our training dataset. Furthermore, the proposed metric shows the improved quality in high dose iteratively reconstructed images, and the reduced quality in low dose images.