An emerging method of limiting radiation exposure in populations requiring frequent imaging is development of disease-specific low-dose CT protocols [20
]. Low-dose CT protocols have been successfully developed which preserve diagnostic yield at significantly reduced radiation exposure, and this has been particularly effective in CT scanning of the thorax [21
]. Achieving diagnostic quality low-dose CT in patients with IBD is particularly challenging as abdominal and pelvic imaging does not lend itself as well to low radiation dose scanning as the thorax. CT in the abdomen and pelvis requires good image contrast to resolve the pathological changes which occur in the liver, spleen, and kidneys. The slight increases and decreases in attenuation value that represent pathology can be obscured by increased image noise more so than in the thorax, where tissues have greater inherent contrast due to the large difference in their densities [22
In the past, fixed tube kilovoltage and amperage settings were used in CT of the abdomen and pelvis; this resulted in areas such as the mid-abdomen receiving the same exposure as regions such as the pelvis. This was an inefficient method of acquisition, and some regions were overirradiated, without any benefit in terms of image quality while other regions were potentially underexposed, increasing image noise and reducing image quality [23
Automatic tube current modulation (ATCM) in CT was a major development of the last decade [20
]. This technology adjusts tube current during a CT scan depending on X-ray attenuation in that anatomic location, and tailors the dose output of the CT tube to patient size and shape. This method ensures that thicker regions of the body are imaged using higher tube currents than thinner, less attenuating areas. Within each imaged section, user-specified noise indices are chosen, which predict an acceptable level of image noise, thereby preserving diagnostic quality while minimising radiation exposure [24
]. The higher the noise index chosen by the radiologist, the greater the noise in the CT image and the lower the radiation dose required to acquire the image. Studies which investigated the effectiveness of ATCM for optimising radiation exposure in CT have reported that a reduction in dose can be achieved in 87% of examinations using ATCM, with an average tube-current time product reduction of 32% [24
Reductions in CT dose often have a negative impact on the diagnostic quality of the images due to increasing noise. Noise is a statistical variation in the attenuation values not reflecting the underlying anatomy which can result in a lack of clear contrast between two adjacent tissues and blurring of the anatomic features of the image. As previously stated, noise can be particularly problematic in the solid upper abdominal organs such as the liver, spleen, and kidneys [22
]. In patients with IBD where CT findings are most commonly limited to the small and large bowel, a larger amount of image noise has been found to be diagnostically acceptable. In a single series, elevating the noise index to 18–25 yielded a 31–64% reduction in radiation dose while preserving diagnostic accuracy [26
At present many noise reduction strategies are being developed, with varied success, to maintain image quality while significantly reducing radiation dose. The earliest strategies included the use of noise reduction filter (NRF) software applications which are used to postprocess CT images acquired at significantly reduced radiation dose, improving their diagnostic quality by reducing image noise [22
]. NRFs were shown to be effective at reducing image noise, but there is potential for negative impact on diagnostic quality of images by reduced lesion conspicuity in organs such as the liver by “over-smoothening” [27
More promising noise and, therefore, dose reduction strategies involve improvements in the image reconstruction process at the time of CT acquisition. The choice of image reconstruction algorithm is critical to the quality and appearance of CT images [28
Although iterative image reconstruction algorithms were used to generate images in the first commercial clinical CT scanner [29
], filtered back projection (FBP) became ubiquitous as it is a more rapid and more computationally efficient method with relatively low mathematical demands [30
]. However, FBP is not well suited to low-dose CT where data within the image is limited and noise is high [31
]; this has led more recently to renewed research and commercial interest in refining methods of iterative reconstruction for reducing noise in images acquired at lower radiation dose. Methods of iterative reconstruction currently represent the most exciting dose optimizing developments in CT [32
]. Various modifications of iterative reconstruction are being developed and refined by different CT manufacturers including Adaptive Statistical Iterative Reconstruction (ASIR) (General Electric Healthcare, Milwaukee, WI), Iterative Reconstruction in Image Space (IRIS) (Siemens Healthcare, Erlangen, Germany), Adaptive Iterative Dose Reduction (AIDR) (Toshiba Medical Systems, Tustin, CA), and iDose (Phillips Healthcare, Best, The Netherlands).
Adaptive statistical iterative reconstruction (ASIR) is a noise-efficient reconstruction algorithm [33
] which is computationally fast and is proven to result in images with good low-contrast detail, preserved image quality, and with typical radiation dose reductions of greater than 30% [35
]. Pilot studies with ASIR found that radiation dose can successfully be reduced by 50% in CT colonography [36
], 44% in coronary CT angiography [37
], and approximately 50% in CT abdomen and pelvis [38
] without significantly affecting image quality.
The next step in optimising image quality in studies acquired at significantly reduced radiation dose is the ongoing development of advanced generations of iterative reconstruction such as Model-Based Iterative Reconstruction (MBIR), which is being developed by GE Healthcare. MBIR can incorporate a physical model of the CT system into the reconstruction process to characterize the data acquisition process, including noise, beam hardening, and scatter [39
]. MBIR offers the potential to further enhance image quality at even lower radiation doses than ASIR. However, due to limitations in computing power and reconstruction technology, model-based iterative approaches have not been practical for commercial CT scanners until recently, though studies refining this technology with the aim of introducing it into clinical practice are now in progress.
Despite emerging advanced reconstruction and processing techniques, it is imperative that dose reduction should also be achieved by optimizing basic acquisition parameters during the day to day practice of CT. Human errors in CT planning have recently been publically highlighted resulting in gross dose increases [40
]. These errors emphasize the importance of the role of the CT technologist in the careful planning and design of CT protocols that prioritize dose optimisation in day to day practice. As an example, scanning beyond the anatomical limits of the imaging examination in the context of CT of abdomen and pelvis has been demonstrated to be common practice, with 97% of cases having extra images above the diaphragm and 94% having extra images below the symphysis pubis in a published study [41
]. In the above study, the additional images added little in terms of extra diagnostic information but added significantly to the radiation dose imparted. This study therefore emphasises that the goal of radiation dose optimisation requires a multidisciplinary approach and radiologists, CT technologists, medical physicists, and referring physicians must be alerted to the fact that attention to detail, and fine adjustments in practise can have a major impact in this area.