Automatic selection of tube potential for radiation dose reduction in CT: a general strategy.

摘要

PURPOSE: To optimize radiation dose efficiency in CT while maintaining image quality, it is important to select the optimal tube potential. The selection of optimal tube potential, however, is highly dependent on patient size and diagnostic task. The purpose of this work was to develop a general strategy that allows for automatic tube potential selection for each individual patient and each diagnostic task. METHODS: The authors propose a general strategy that allows automatic adaptation of the tube potential as a function of patient size and diagnostic task, using a novel index of image quality, "iodine contrast to noise ratio with a noise constraint (iCNR_NC)," to characterize the different image quality requirements by various clinical applications. The relative dose factor (RDF) at each tube potential to achieve a target image quality was then determined as a function of patient size and the noise constraint parameter. A workflow was developed to automatically identify the optimal tube potential that is both dose efficient and practically feasible, incorporating patient size and diagnostic task. An experimental study using a series of semianthropomorphic thoracic phantoms was used to demonstrate how the proposed general strategy can be implemented and how the radiation dose reduction achievable by the tube potential selection depends on phantom sizes and noise constraint parameters. RESULTS: The proposed strategy provides a flexible and quantitative way to select the optimal tube potential based on the patient size and diagnostic task. The noise constraint parameter alpha can be adapted for different clinical applications. For example, alpha = 1 for noncontrast routine exams; alpha = 1.1- 1.25 for contrast-enhanced routine exams; and alpha = 1.5-2.0 for CT angiography. For the five thoracic phantoms in the experiment, when alpha = 1, the optimal tube potentials were 80, 100, 100, 120, 120, respectively. The corresponding RDFs (relative to 120 kV) were 78.0%, 90.9%, 95.2%, 100%, and 100%. When alpha = 1.5, the optimal tube potentials were 80, 80, 80, 100, 100, respectively, with corresponding RDFs of 34.7%, 44.7%, 54.7%, 60.8%, and 89.5%. CONCLUSIONS: A general strategy to automatically select the most dose efficient tube potential for CT exams was developed that takes into account patient size and diagnostic task. Dependent on the patient size and the selection of noise constraint parameter for different diagnostic tasks, the dose reduction at each tube potential, quantified explicitly with the RDF, varies significantly.