The continued increased usage of Positron Emission Tomography/Computed Tomography (PET/CT) imaging in preclinical research has fostered efforts to develop protocol standardization and to evaluate optimal acquisition and reconstruction protocols. MicroPET/CT is a key non-invasive imaging tool in biomedical research, because it allows for reduction of the number of animals used, while simultaneously improving statistical power by allowing each animal to work as its own control in longitudinal studies, with consequential scientific benefit [1]. Moreover, diagnostic and therapeutic agents can be developed and tested on identical platforms, accelerating the translational process to humans [2]. Hence, efforts toward optimization of microPET/CT acquisition and reconstruction protocols will have a widespread impact across several biomedical fields, including: biomarker research [1, 3–5], and research in neurodegenerative diseases, cardiovascular diseases [6–10], metabolic diseases, musculoskeletal disorders and oncology [11–14].

Literature reporting clinical PET/CT protocol optimization, image co-registration and CT attenuation-based corrections of PET data is vast, e.g., [15–17]. Conversely, the majority of studies addressing microPET/CT optimization of protocols have been performed primarily using the manufacturers default protocol or slight variations thereof [18, 19]. Therefore, the impact of varying microCT parameters on microPET/CT data and preclinical research results is poorly defined and not fully understood. Furthermore, the significance of different microCT parameters on radiation exposure to small laboratory animals is scarcely described in the literature [20], in particular cumulative severity for longitudinal studies. Given that cumulative radiation doses can become increasingly toxic [21], accurate microCT absorbed dose measurements are essential when undertaking longitudinal preclinical research. This is most relevant and obvious when working with rodent tumor models in order to assure a clear distinction between diagnostic radiation imaging and radiotherapy [22].

The imaging community is currently undertaking efforts toward developing guidelines on microPET/CT acquisition and reconstruction methods [18, 23, 24]. While these guidelines admirably address animal handling and preparation, National Electric Manufactures Association (NEMA) testing and scanner quality control testing, there still remains a knowledge gap on evaluation and selection of preclinical PET/CT acquisition and reconstruction for routine use locally and across sites. Specifically, research aiming to identify the optimal protocols in order to improve not only image quality and quantification, but also the impact of absorbed microCT dose to small laboratory animals.

In this paper, we evaluate all the microCT acquisition protocols available on a microPET/CT scanner and how this impacts image quality and quantitation for the CT and PET components, as well as, the CT radiation exposure to small laboratory animals. This is done with the intent to optimize microCT imaging protocols for the avoidance of high radiation exposure while addressing any impact that varying microCT protocols has on microPET imaging.

All images were acquired and reconstructed on a second generation Mediso microPET/CT scanner (nanoPET/CT, Mediso, Hungary) at the Edinburgh Preclinical Imaging (EPI) Laboratory. This scanner is a two-ring microPET system with a flat screen microCT component (configuration based on previously described methods [25, 26]). The scanner's microCT x-ray source (tube and high voltage generator) is the Source-ray, Inc. (Ronkonkoma, NY, USA) SB-80-1K. Quality assurance was performed daily on the Mediso nanoPET/CT for both the PET and CT components. Daily quality assurance testing was also conducted on the Capintec CRC-25R (New Jersey, USA) dose calibrator. Mediso's scanner software generates the attenuation maps, per pixel value, from the CT in which HU values are converted to the linear attenuation coefficients for application to the PET data.

CT acquisition protocols available on the Mediso nanoPET/CT scanner were evaluated; and the impact each protocol had on image quality, quantitation and the CT radiation exposure to small laboratory animals was assessed. Additionally, the impact of different microCT protocols on microPET reconstructed data was investigated.

Using the air/water phantom, a wide spread of measured HUs was observed, depending on the acquisition protocol used. This informed on the best and worst microCT protocols based on bias and image noise. Based on collected data, protocol 4 provided a good compromise between air and water quantification and image noise, based on HU bias. This protocol also resulted in low levels of noise using the TEM phantom, but delivered a relatively high absorbed dose of about 100 mGy. In comparison with the default settings (i.e., protocol 7), protocol 4 performed marginally better in air, water, muscle and low density bone; but resulted in higher absorbed dose. These differences are likely due to scanning duration being higher for protocol 4 compared with 7, consequence of higher number of projections (480 vs. 720 projections); distance of x-ray source relative to object (medium zoom and maximum FOV) and differences in voltage where high voltage (protocol 7) performed better than low voltage (protocol 4) in high density tissues, such as bone. Protocols 3 and 9 performed worst in terms of bias and image noise. Notably, protocol 3 resulted in low absorbed doses (approximately 15 mGy), while protocol 9 yielded relatively high absorbed doses (about 85 mGy). The highest absorbed dose was measured for protocol 5 (about 700 mGy) and the lowest was measured for protocol 2 (about 4 mGy). Although these values are well below therapeutic thresholds for rodents x-ray irradiation (typically > 6 Gy), the appearance of chromosomal aberrations in mammals has been reported with doses of 250–300 mGy [28–30].

Using computational simulation, Taschereau et al. and Boone et al. investigated the effects microCT parameters on absorbed doses [31, 32]. They predicted that for a whole body scan, the dose could vary between 80 and 133 mGy. Recently, Vande Velde et al. [33], using a SkyScan Bruker microCT scanner (parameters: 50 kVp and 120 ms; 50 kVp and 450 ms) and adult mice, reported absence of radiation induced lung damage even at a high dose protocol of 450 ms, despite measured skin doses as high as 1,104 mGy and lung doses of 813 mGy. They suggested that high dose microCT is safe for longitudinal lung studies, but recognized the current limited understanding among the scientific community on the impact of high absorbed doses in small animal longitudinal studies. Unfortunately this study didn't report measures of image quality or quantitative bias, which would have assisted in establishing whether high dose scans effectively, translate into high quality and quantitatively more accurate scans. In 2011, Kersemans et al. reported DNA damage as a result of high absorbed doses CT scanning of >150 mGy. They also reported absence of damaging effects when using low dose CT protocol (8.3 mGy), without compromising image quality. Notwithstanding, the impact high dose CT has on longitudinal experimental results remains controversial. Studies often report opposing findings from no effects regarding dose delivered to tumors during longitudinal studies to results indicating high doses can induce tumor inhibition [19, 20, 34–36]. Using a Mediso nanoPET/CT scanner, we found that for both the air/water and the TEM phantoms, decreasing CT dose resulted in visual image quality degradation and high noise, but had a relatively minimal impact on HU quantification, while adversely increasing small animal absorbed dose. Furthermore, this highlights the importance of assessing the trade-off between image resolution and absorbed dose per application and scanning platform. This current work also demonstrated that combinations of microCT parameters, not simply exposure time or tube voltage, but also scanner trajectory, impacted the absorbed dose. This new data can help informing future scanner developments.

Here we reported the use of a prototype 3D-printed phantom for absorbed dose measurements and compared the ionization chamber results with NanoDots. Collected data showed strong correlations with the absorbed dose measured using NanoDots in a rat cadaver, albeit biased by 4–27%. Bias in commercial phantom materials can also be observed, therefore our phantom results provide encouraging proof of concept data for further development of the prototype model with improved printing materials and polymers. Ultimately, the development of 3D-printed phantoms has potential for replacement of the use of animals for preclinical PET/CT protocol development.

Although different microCT acquisition protocols resulted in visually different attenuation maps, this negligibly impacted microPET image uniformity, RC, SOR and quantitative accuracy. Attenuation correction errors in small animal PET imaging are likely minimal, which may explain the obtained results. The 9% average measured difference for protocol 5 between microPET images reconstructed when the PET phantom was radioactive vs. non-radioactive (hot/cold phantom) could be due to the scatter correction used or count rate. This data can help with deciding protocol sequence (i.e., PET first then CT or vice-versa) when conducting dynamic PET/CT imaging. Moreover knowing this bias can support potential data corrections of PET data for CT attenuation map bias.

The measured RCs were consistently overestimated (average 1.12), suggesting overcorrection during the process of PET image reconstruction using the iterative proprietary default methods implemented in this dual-ring PET/CT system. Previous microPET/CT studies have reported RC values > 1.0 in the 5 mm cylinder. Using the NEMA NU 4-2008 methodology, a TriFoil system RC for the 5 mm cylinder was 1.02 RC [37]. RCs of 0.93 and 0.99 have been reported for the Siemens Iveon [38] and the first generation Mediso nanoPET/CT scanners [25], respectively. Here we report, for the first time, the RCs for the new Mediso nanoPET/CT system with two-ring configuration vs. first single ring systems [25], which could explain differences in measured RCs (1.12 vs. 0.99). However, RCs can also vary depending on the reconstruction algorithm applied and the radionuclide used for testing [23]. This data supports further development and improvement of available reconstruction tools.

This seminal work conducted on a single PET/CT scanner platform clearly demonstrates the importance of assessing available parameters for image acquisition. It provides valuable information for the community of imagers using the Mediso PET/CT scanning platforms, as well as scanner developers, and highlights the importance of extending this analysis to other platforms.

Protocols with increased CT scanning durations, consequently leading to an increase in small animal absorbed doses from X-rays, do not directly translate onto improved image noise nor impacts onto measured HUs. These findings can be particularly pertinent information when conducting longitudinal microPET/CT studies. Different CT protocols minimally impact PET data quantification. Our results support the need to investigate imaging outcomes from different protocols and scanners prior to selection of PET/CT methods for routine use, in order to avoid quantitative spurious comparisons and minimize radiation exposure to small laboratory animals.

This study was carried out in accordance with the recommendations of the Home Office under the Animals (Scientific Procedures) Act 1986 and the local University of Edinburgh animal ethics committee. The protocol was approved by the local University of Edinburgh animal ethics committee.

WM collected all data, conducted data analysis and drafted the manuscript. AT assisted in the interpretation of the data and the revising of the manuscript. RC assisted in the design, development and printing of the 3D-printed rat phantom prototype. MG assisted in the collection and measurements of the ion chamber microCT data. All authors participated in the conception of the study and design. All authors read and approved the final manuscript.