Automated Detection of Respiratory Movements for Image Quality Assurance

Abstract

In this study, we aimed to develop a rapid quantification technique that can circumvent radiodiagnostic errors that occur during respiratory movements in chest imaging. We also intended to use this technique to improve data interpretation, diagnosis, and image quality assurance. Chest X-ray imaging was performed in a total of 45 patients between 2006 and 2011 at a general hospital. During imaging, the respiratory movements of the patients were visually evaluated by radiologists and radiological technicians. The radiographic images were retrospectively analyzed. A statistical analysis was conducted using equal-variance Student’s t-test. A significantly distinct pattern in the frequency domain was observed in the power spectra of radiographs of patients without and with respiratory movements. Similarly, 3D power spectrum images showed a decreased power intensity and a wider base region in images of patients with different disease states that reflected different patterns. These results suggest that the proposed method can easily detect respiratory movements; hence, it may be used for image quality assurance and improving data interpretation and diagnosis.