Data-Driven Motion Correction for Human Head in Computed Tomography

Human head motion while obtaining Computed Tomographic head scans results in motion artifacts that resemble long streaks in the CT images. These artifacts hide anatomies and hence can lead to misdiagnosis. An under-scan weighting function, that manipulates Sinogram data to reduce motion artifacts has been optimized and validated for clinical CT head images. It requires manual segregation of CT head images into motion and motion free subsets by visual analysis. The optimized under-scan weighting function is used to normalize the Sinogram of the CT head image that is affected by motion artifact. CT head images corrupted by motion artifact and the ones generated from Sinograms corrected for motion artifacts are visually and quantitively compared. Motion artifacts lead to increase in noise in the streaky regions. Motion correction algorithm results in noise reduction in the streaky regions. The same have been verified.