Dose reduction for children is widely promoted in the Image Gently campaign by the Alliance for Radiation Safety in Pediatric Imaging. For adult imaging, a similar effort has been made in the Image Wisely campaign by the Joint Task Force on Adult Radiation Protection. From pediatric to bariatric patients, patient sizes differ across a substantial range. The proper implementation of dose reduction for such a wide range of patient sizes remains challenging. In principle, the degree of dose reduction should adhere to the preservation of image quality; hence, it is a process of dose optimization. However, different definitions of the image quality metric have been proposed from past studies, including constant pixel noise or constant contrast-to-noise ratio (CNR). The former has been implemented on many commercial scanners for automatic tube current modulation. Constant detector energy deposition was also studied. Nevertheless, neither of these metrics adequately addresses task-specific low-contrast detectability.
In previous work, we utilized statistically defined low-contrast detectability (SD-LCD) as the task-specific image quality metric for an abdomen phantom of medium size. Based on those studies, using two different scanners, the relationship of the dose to SD-LCD and lesion size was found to be in the form of a power law. However, several important questions remained unanswered. First, a power law form identified using a medium-size phantom may or may not be valid for the phantoms of other sizes, because the influences of multiple bow-tie filters and scatter are unknown. More important, the explicit patient size dependency of the minimum detectable contrast remains unknown. This is true even if the general power law form is valid, so we continue to investigate whether the patient size dependency is in the power indices or in the proportionality factor. Furthermore, it remains unclear whether there exists a generic or scanner design — dependent relationship with which to gauge the dose modulation for different patient sizes.
In this study, we used two approaches to address these questions. We first used a simplified theoretical model to relate the minimum detectable contrast to the patient size, dose and lesion size. The result of this model also indicated a connection between the minimum detectable contrast and the bow-tie filter design. We extended the conclusion to a general equation that could be experimentally validated. The validation was then carried out using abdomen phantoms of various sizes, scanned with two different scanners. The answers can be used to obtain the doses for different patient sizes to achieve consistent low-contrast detectability, which therefore may benefit Image Gently and Image Wisely.