Phantom design & construction
The phantom dimensions were consistent with the effective diameter of a 5-month-old child according to the AAPM lookup chart [18].
The CNC milling machine was chosen as it is a widely available, programable and reproducible tool, commonly used for fabrication of plastics. However, obstacles included limitations on geometric pathways associated with specific tool-bits and Mastercam design file types not being compatible with the available CNC software. This resulted in some deviation from the original design.
The addition of patient arms is of clinical relevance for paediatric CT, since infants are often scanned in a baby cradle with their arms strapped by their sides to reduce movement during imaging. The alternative central PMMA rods can be used to model paediatric arms, which are typically included in the scan range for patients less than 6 months old.
Tissue equivalent materials
PMMA is well documented to be soft tissue equivalent at low diagnostic beam energies [17], but this agreement drifts at higher diagnostic beam energies, and evidently also with added beam filtration (as seen in Table 2 and Fig. 7). Alternative materials have been used to model newborn thoracic tissues, for instance, Jamal et al. produced a low-cost newborn chest phantom for radiography, using beeswax and polyurethane foam [10]. Other studies have reported various plastics and wood that can be 3D printed for soft tissue substitute materials in phantom construction for radiology [19] and nuclear medicine [20].
Eurosil-10 has demonstrated its diversity in modelling bone equivalent materials appropriate for the use of constructing phantoms. The calibration curve of the Eurosil-10 and gypsum solution (Fig. 3) at multiple beam energies allowed the desired paediatric bony structures to be modelled across a range of beam qualities. The data in Fig. 3 could alternatively be used to interpolate the concentration of gypsum required to model adult bone resulting in a higher HU values at the associated beam energies.
Figure 3 clearly demonstrates the increased HU values observed in bony structures when scanning with lower beam energies compared to higher beam energies and heavily filtered beams. This was also confirmed in the results for the clinical survey of paediatric tissues where the maximum HU values observed in bony structures decreased as the beam energy and filtration increased.
Thorough mixing was required when producing the bone substitute materials. Bubbles of dry gypsum would form if not mixed properly. Due to the high concentration of solution, gypsum naturally settled towards the base of the structure as it solidified. This effect was observed as a peak in HU values on the interface of phantom slices in the tissue validation measurements.
Mixing and pouring the bony structures was difficult due to the large volume required, and limited time to work before solution begins to solidify. It was evident after making samples of varying concentrations, that the high concentration solution solidified significantly faster than the base Eurosil-10 solution without any added gypsum. The bone substitute material needed to be mixed and poured in smaller batches to avoid difficulties pouring the thickening solution.
Eurosil-4, the alternative soft tissue substitute material required for constructing the contrast resolution tool (slice 7) displayed significantly higher HU values compared to soft tissue across all beam energies and was not considered equivalent to soft tissue, as was initially desired. The contrast levels were made using added softener to decrease the density, and therefore the HU values but this was unsuccessful at producing distinctive contrast levels. Muir and Laban recently produced a resolution tool, with contrast levels constructed by adding silicone blasting sand to a Pinkysil product, in a phantom designed for dental CBCT [20].
Dosimetry and image quality slices
Figure 7 shows the reconstructed CT images of some of the image quality slices and the following are noted:
7a) Slice 6: was used to measure contrast resolution in the lung equivalent tissue: the iodinated foam produced contrast levels that could be visually distinguished from background in 4 progressive stages. With appropriate window and levelling, the contrast resolution tool can be used as a subjective measure of image contrast.
Not shown) Slice 7: was used for measuring contrast resolution in soft tissue equivalent material. Unfortunately, the Eurosil-4 Pink solution used to make this tool displayed a HU of between 320–370 HU at 70 kVp and was therefore not a desirable soft tissue substitute. The increased amounts of softener were successful in decreasing the measured HU but not significantly enough to be distinguished by human eye.
7b) Slice 8: The MTF was successfully modelled from the point source signal created by the Nichrome wire insert in the EVA foam background. A short script was written and executed via MATLAB Version 2011a (Natick, Massachusetts: The MathWorks Inc.) to generate the MTF plots displayed in Fig. 6.
7c) Slice 9: Five progressively smaller air gaps can be viewed in the lung equivalent material of the contrast detail phantom slice. The resolution of these air gaps progressively worsens in the presence of noise and with decreasing slice thickness. Reversing the pattern in the opposite lung was designed to remove any effects due to anatomical differences.
Phantom validation
The soft tissue substitute material was considered equivalent to paediatric tissue when scanning at 70-80kVp but not at 100 kVp with or without added tin filtration. In this case, the HU values from PMMA (101–123 and 99–139 HU) were higher than the range of HU values clinically observed (30–90 and 20–80 HU) in cardiac muscle of paediatric patients scanned at 100 kVp and Sn100kVp.
The lung tissue substitute material, cork was validated to be equivalent to paediatric lung tissue as it fell well within the clinically observed range of HU values across all beam qualities tested.
The bone tissue substitute materials were also validated as equivalent to paediatric bone for both ribs and spine according to the wide range of HU values observed in paediatric bone tissue in the CSPT.
Clinical survey of paediatric tissues
Measuring & analysing average HU values within paediatric tissues was complex and not pertinent to this study. Instead maximum and minimum values were collected to classify a range of clinically observed HUs within paediatric tissues. It should be considered that all the patients included in this dataset had clinical pathologies or queries and some patient’s bone development could be inaccurate due to premature birth.
As demonstrated in Table 2 and illustrated in Fig. 4, the CSPT identified a range of clinically observed HU values in the thoracic structures of children across the four groups. The range of HU values differed the most in the high-density bony structures, the minimum HU remained relatively consistent, while the maximum HU values observed in bone decreased significantly with increasing kVp & filtration. Very little difference in HU range was observed in lung, fat and soft tissue when comparing against beam qualities. The overall range of HU values taken as the maximum of bone tissue compared to the minimum of lung tissue was the largest in the 70 kVp group and smallest in the Sn100 kVp group.