PURPOSE: The aim of this study was to measure and characterize breathing-induced motion artifacts in fast helical free-breathing CT scans. METHODS: Ten lung cancer patients were scanned using fast helical CT during free breathing. In each case, 25 low-dose CT scans were acquired in alternating craniocaudal and caudocranial directions. A bellow-based breathing surrogate was simultaneously acquired. A published breathing motion model was used to estimate the diaphragm craniodaudal velocity at each CT scan. The Hounsfield unit (HU) profiles passing through a small square (7 × 7 mm2 ) portion of the diaphragm were examined and fit to error functions, which were used to characterize the motion blur. The HU profiles that intersected diaphragm-adjacent non-parenchymal tissue were excluded from analysis. The five profiles for each scan that were best isolated from non-parenchymal tissue were used to determine the amount of blurring. A convolution-based blurring model was also employed to compare against the human data. RESULTS: There was a distinct relationship between blurring and diaphragm speed. The convolution model well described the blurring behavior in the patients. Most of the CT scans were acquired at tissue velocities less than 20 mm s-1 , which was the threshold where blurring exceeded 1 mm (corresponding to the slice spacing in this study). CONCLUSIONS: Breathing motion-induced blurring occurs even for relatively fast modern helical CT scans. Measurable motion-induced blurring occurs for velocities greater than 15 mm s-1 and greater than 1 mm for velocities greater than 20 mm s-1 . Methods to manage the residual blurring artifacts will need to be developed to maximize the image quality for free-breathing CT protocols.
PURPOSE: The aim of this study was to measure and characterize breathing-induced motion artifacts in fast helical free-breathing CT scans. METHODS: Ten lung cancerpatients were scanned using fast helical CT during free breathing. In each case, 25 low-dose CT scans were acquired in alternating craniocaudal and caudocranial directions. A bellow-based breathing surrogate was simultaneously acquired. A published breathing motion model was used to estimate the diaphragm craniodaudal velocity at each CT scan. The Hounsfield unit (HU) profiles passing through a small square (7 × 7 mm2 ) portion of the diaphragm were examined and fit to error functions, which were used to characterize the motion blur. The HU profiles that intersected diaphragm-adjacent non-parenchymal tissue were excluded from analysis. The five profiles for each scan that were best isolated from non-parenchymal tissue were used to determine the amount of blurring. A convolution-based blurring model was also employed to compare against the human data. RESULTS: There was a distinct relationship between blurring and diaphragm speed. The convolution model well described the blurring behavior in the patients. Most of the CT scans were acquired at tissue velocities less than 20 mm s-1 , which was the threshold where blurring exceeded 1 mm (corresponding to the slice spacing in this study). CONCLUSIONS: Breathing motion-induced blurring occurs even for relatively fast modern helical CT scans. Measurable motion-induced blurring occurs for velocities greater than 15 mm s-1 and greater than 1 mm for velocities greater than 20 mm s-1 . Methods to manage the residual blurring artifacts will need to be developed to maximize the image quality for free-breathing CT protocols.