CT Users Group

Katie Howard

D Tout, K Howard, PJ Julyan, PA Hulse, P Manoharan, DL Hastings

North Western Medical Physics,
Christie Hospital NHS Foundation Trust,
Manchester

Abstract

CT acquisition parameters for a GE Discovery STE8 PET-CT were initially chosen to be in line with those used in other PET-CT centres in the UK. However, a large variation in CT image quality prompted an investigation to optimise CT imaging parameters in diagnostic PET-CT to provide consistent image quality whilst minimising effective dose.

Twenty-five patient images of varying image quality were presented to two experienced PET-CT reporters who scored the images subjectively as acceptable or not acceptable based on overall image noise and structural resolution required in PET-CT exams. A quantitative measure of image noise was determined by the standard deviation of Hounsfield units (HU) in a transaxial slice through the liver (a nominally large homogeneous structure). The mean patient weight (±SD) and liver standard deviation of images scored as acceptable were 75.7 ± 16.8 kg and 28.5 ± 8.6 HU and for images scored as non-acceptable were 121.1 ± 12.5 kg and 58.8 ± 10.6 HU. This indicated that consistent image quality independent of patient size was not being achieved.

GE's automatic exposure control (AEC) uses information from the scout view to adjust the tube current during the exam, up to a pre-set maximum (mAmax), to compensate for attenuation differences along a patient. A "Noise Index" (NI) is used to set the required noise level in the resulting images. The mA tables produced by AEC for patients covering a range of weights were investigated. AEC data showed that for our current protocol the set NI was not being achieved in larger patients as the tube current was consistently being capped at mAmax. For smaller patients tube current was not reaching mAmax and a high level of image quality was being achieved.

The image scores and liver standard deviation data were used to determine a more appropriate (higher) NI that would produce CT images of acceptable image quality for PET-CT exams, and mAmax was increased so tube current was not capped. This protocol was trialled on patients undergoing PET-CT imaging. The resulting images were found to have more consistent image quality with a mean liver standard deviation (±SD) of 27.6 ± 2.0 HU and patient weight ranging from 48 - 110 kg (n = 34).

Effective dose estimates use knowledge of the absorbed energy per unit mass. For the same tube current a lower effective dose is reported for a larger patient mass [AJR 2007; 188: 540-546]. The new protocol resulted in higher tube currents for larger patients, but this will not have led to a proportional increase in the effective dose relative to smaller patients. Equally, the reduction in current for smaller patients, caused by setting a higher NI, is expected to partially compensate for the higher absorbed doses received by their smaller mass. An assessment of the effective dose and a clinician's assessment of image quality across a range of patient sizes under the new protocol will be presented.

We have optimised our CT settings in order to produce CT images of acceptable image quality for anatomical localisation in PET-CT, and to produce consistent image quality independent of patient size without greatly increasing the exposure parameters.

Click the link for a pdf of the presentation:

Back to meeting timetable