Filtered region of interest computed tomography
MetadataShow full item record
In recent years computed tomography has been established as the primary source of radiation from medical applications to the general public with more than 62 million acquisitions performed a year. In many procedures, the region of interest (ROI) is often considerably smaller than the imaged field of view (FOV), thereby subjecting the patient to unnecessary x-ray dose. For such procedures we propose reducing the x-ray dose to the patient by placing an x-ray attenuating filter in the x-ray beam; we call this approach filtered region of interest (FROI) computed tomography (CT). In FROI CT the acquired projection data consists of a high contrast/dose ROI, a low contrast/dose peripheral region and a transition region of constant gray value decrease in between. To achieve artifact free volumetric reconstruction we developed methods to efficiently and correctly equalize the attenuated gray values in the peripheral region to the gray values inside the ROI and resurrect the structural data inside the transition region. Comparison to reconstructions from conventional unfiltered data showed no difference in Hounsfield units (HU). In addition evaluations focusing on contrast-to-noise ratio (CNR) and scatter fraction were performed. Results showed that the CNR in the ROI is slightly improved inside the ROI, a result supported by the reduction in scatter fraction inside the ROI. CNR in the peripheral region though is considerably reduced, decreasing from 10% to 60% when compared to conventional data attributable to an increase in projection image noise. In addition to the good image quality inside the ROI, the proposed method offers considerable dose reduction. Theoretical and experimental dose evaluations showed that with a large ROI encompassing 70% of the FOV and a low filter thickness of 0.86 mm, integral dose reduction of about 20% can be achieved. Increasing the filter thickness for this ROI size increases the dose reduction to 30% (1.72 mm). For small ROI sizes (30% of the FOV) 60% and 80% integral dose reduction can be achieved for 0.86 mm and 1.72 mm filter thickness. In addition to centered objects of interest, clinical tasks may call for imaging of off-center objects of interest with high quality. For this circumstance we propose a double ROI approach mirrored along the axis of rotation. Evaluations of this method showed that the object of interest is inside the ROI for more than 85% of the projection images, thereby yielding good image quality for off-center objects of interest. To conclude, this dissertation thoroughly discusses the possibilities and overcomes the challenges yielded by FROI CT for both centered and off-centered objects of interest in the clinical arena. The dose reductions possible surpass all so far available options, while supplying image quality similar or even better inside the ROI compared to conventional imaging, making this method a feasible option in the clinical arena.