Modelling CTDIvol through analysis of the localiser image: Towards estimating factors influencing Automatic Exposure Control tube output

Abstract

The purpose of the Automatic Exposure Control (AEC) system in CT is to optimise radiation dose to the patient and provide adequate image quality. Prior to the CT, a 2D X-ray localiser is acquired allowing for an estimation of both the patient size and their attenuation properties; however, acquisition of the localiser is susceptible to patient positional errors which can affect an AEC algorithm misinterpretation of the subject size and attenuation. A magnifying effect, for instance, may result in an increased and suboptimal CTDIvol. The literature reports an approximately linear relationship between CTDIvol and patient offset. The testing of a newly commissioned PET/CT Siemens Biograph 16 Horizon investigated changes in CTDIvol with table height. PosteroAnterior (PA) and AnteroPosterior (AP) localisers were obtained for anthropomorphic and uniform phantoms at various table heights. A multivariate analysis was employed to investigate the relationship between the CTDIvol and parameters extracted from the localiser image. For the PA localiser dataset, irrespective of the phantom used, a recurring nonlinear relationship with patient offset was observed. Upon investigation, it appeared in some instances that the table rather than the patient was interpreted by the AEC algorithm as the subject, leading to a misinterpretation of the subject attenuation and size, and affecting an increased CTDIvol. Indeed, this was also evident for some AP localiser cases. The multivariate analysis informed on the contribution of the subject effective diameter and summed linear attenuation, identifying these as significant predictors. The resultant model accurately simulates CTDIvol from localiser image parameters.