Deep learning‐based coronary artery motion estimation and compensation for short‐scan cardiac CT

Abstract

During a typical cardiac short scan, the heart can move several millimeters. As a result, the corresponding CT reconstructions may be corrupted by motion artifacts. Especially the assessment of small structures, such as the coronary arteries, is potentially impaired by the presence of these artifacts. In order to estimate and compensate for coronary artery motion, this manuscript proposes the deep partial angle-based motion compensation (Deep PAMoCo). Methods: The basic principle of the Deep PAMoCo relies on the concept of partial angle reconstructions (PARs), that is, it divides the short scan data into several consecutive angular segments and reconstructs them separately. Subsequently, the PARs are deformed according to a motion vector field (MVF) such that they represent the same motion state and summed up to obtain the final motioncompensated reconstruction. However, in contrast to prior work that is based on the same principle, the Deep PAMoCo estimates and applies the MVF via a deep neural network to increase the computational performance as well as the quality of the motion compensated reconstructions. Results: Using simulated data, it could be demonstrated that the Deep PAMoCo is able to remove almost all motion artifacts independent of the contrast, the radius and the motion amplitude of the coronary artery. In any case, the average error of the CT values along the coronary artery is about 25 HU while errors of up to 300 HU can be observed if no correction is applied. Similar results were obtained for clinical cardiac CT scans where the Deep PAMoCo clearly outperforms state-of-the-art coronary artery motion compensation approaches in terms of processing time as well as accuracy. Conclusions: The Deep PAMoCo provides an efficient approach to increase the diagnostic value of cardiac CT scans even if they are highly corrupted by motion.

Auto-detection of motion artifacts on CT pulmonary angiograms with a physician-trained AI algorithm

“…We achieved a similar AUC of 0.88 with a single coronal MPR image per CTPA examination. Other studies [17,19] [20] Based on the identification and estimation of motion artifact, other investigators have reported on motion artifact correction and compensation solutions for head and cardiac CT examinations [21,22].…”

“…Xu et al . reported a fully automatic AI for grading image quality (motion artifacts) of CCTA using semi-automatic labeling and tracking of the coronary arteries [20] Based on the identification and estimation of motion artifact, other investigators have reported on motion artifact correction and compensation solutions for head and cardiac CT examinations [21, 22].…”

Four‐dimensional computed tomography of the left ventricle, Part I: Motion artifact reduction

“…18,19 Some approaches use (image and/or feature) registration between a series of half -scan reconstructions. 11,20 Recently, deep learning networks have also been used effectively to estimate and correct for motion in CT. 16,21,22 Finally, the concept of using PARs has been used to facilitate robust motion estimation, with correspondences being found in conjugate pairs of PAR images. 15,23 It is interesting to note that backproject-then-warp motion compensation as well as PAR-based motion estimation both rely on essentially breaking up the sinogram data into smaller sections in time and representing these in the image domain as PAR images.…”

“…Some approaches use (image and/or feature) registration between a series of half‐scan reconstructions 11,20 . Recently, deep learning networks have also been used effectively to estimate and correct for motion in CT 16,21,22 . Finally, the concept of using PARs has been used to facilitate robust motion estimation, with correspondences being found in conjugate pairs of PAR images 15,23 …”

Simulation of random deformable motion in soft-tissue cone-beam CT with learned models

“…Motion compensation for interventional CBCT has gained significant attention, with image-based approaches including autofocus methods based on handcrafted metrics [1][2][3], and methods leveraging deep convolutional neural networks (CNNs) to directly learn motion trajectories from distortion patterns [4], or to learn features associated to motion effects that are aggregated into deep autofocus metrics [5,6]. Common to those approaches is the need for simulation methods that allow the generation of large amounts of realistic, motion-corrupted, CBCT data to enable training and evaluation.…”