Validation of a Locally Designed Computed Tomography Dose Phantom: A Comparison Study with a Standard Acrylic Phantom in South-South, Nigeria

Purpose: The aim of this study was to determine the mean volume computed tomography dose index (CTDIvol) for the standard head and body phantoms and locally designed head and body phantoms respectively. Similarly, this study determined and compared the displayed mean CTDIvol and Dose Length Product (DLP) for the above phantoms from the CT monitor. In addition, the percentage deviations of both phantoms were compared with the recommended limits from the International Atomic Energy Agency (IAEA) and the American College of Radiologists (ACR).


Introduction
Computed tomography (CT) has been identified as a powerful tool in clinical diagnosis and management [1]. The advancement in the development of CT scanners has given rise to an increase in the application of this medical imaging modality [2]. The use of this application is on a high and continuous increase [3,4]. In Nigeria, Adejoh et al. observed a significant increase in the use of CT scanners [5]. A fall out of this increased usage is a corresponding increase in radiation dose delivery to the patient relative to that from other imaging modalities [6]. For instance, the National Cancer Institute bulletin indicated that the ionizing radiation dose delivered from the use of CT could be 50 -500 times higher than that from an X-ray chest examination [7,8]. There has been concern that such high doses from this observed increase in the application of this diagnostic tool may in the long-term pose a significant cancer risk to the populace. This consequent increase in dose delivery may be attributable to inefficient optimization of scanner radiographic practices or to substandard/poor equipment conditions. It is therefore expected that appropriate examination conditions and procedures need to be optimized so as satisfy the twin purposes of good diagnostic quality and appropriate patient dose. The United Nations Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) has documented that a number of evaluations on the doses associated with CT scans have been carried out and that these investigations were as a result of some observed incidences of overexposure to radiation [9]. In our study, the mean volume computed tomography dose index (CTDI vol ) was determined for the standard head and body phantom as well as the locally designed head and body phantom. Also, evaluation and comparison of console displayed CTDI vol and DLP values were made for the above phantoms. This is to verify the dose delivery accuracies in head and body CT scans in some available scanners in the South-South region of Nigeria.

Materials and Methods
Before the commencement of this study, ethical clearance was obtained from the Health Research and Ethics Committee from the four CT centers within the South-South region of Nigeria. Two were government-based and the other two were privately owned centers, located in Edo and Delta States, and coded as A, B, C, and D respectively. Some details of the scanners are given in Table 1.
The PMMA phantom ( Figure 1) was sourced from the National Institute of Radiation Protection and Research (NIRPR) in Ibadan, Nigeria. It was made up of two PMMA cylinders with a diameter of 16 and 32cm for the head and body respectively. Each of the cylinders had a length of 15cm. The cylinders had inserts large enough to accommodate the TLDs. The designed phantom was made in line with the design in Akpochafor et al. [10] (Figure 2). A PMMA sheet of thickness of 3 x 10 -3 m and density of 1185 kgm -3 was manufactured to give the desired cylindrical shape and was made to meet the requirements as specified in the standard phantom. The constructed phantom was filled with water before inserting the TLD chips for measurement [11,12]. Measurement was done using the   same protocol for both the standard and locally constructed phantom. Each center used different protocols for head and body. Comparison of dose measurements was made between the standard and constructed phantoms in centers A and B only. This is due to the downtime of the CT scanners in centers C and D when the standard phantoms were available for use. The displayed console doses were also compared using the constructed phantom on all the centers (A, B, C and D).
The average weighted computed tomography dose index CTDI w was determined by inserting the TLD chips in the center and peripheries. The CTDI w values were obtained from the relationship [13]: Where CTDI center represents the mean dose measurements at the center of the standard head and body of a PMMA phantom.
CTDI periphery represents the mean dose of the measurements at four locations around the periphery of the standard head and body of a PMMA phantom.
The CTDI vol values were estimated using the expression in European Union EU given as [14]: Where the "pitch" denotes the ratio between increment per rotation and beam width.
In addition, the Dose Length Product (DLP) was computed as [14]: CTDI vol is primarily useful as a quality assurance tool to compare doses from different protocols and to compare scanner outputs from different manufacturers. This is to help us estimate the doses delivered to the 32 and 16cm head and body phantom [15]. The measured values from the constructed phantom were then validated against those of the standard phantom using the formula: ∆D = D SH/SB -D CH/CB is the deviation of the TLD dose reading in the standard head (SH)/ body (SB) and the constructed head (CH)/ body (CB) phantoms.
A correction factor was determined for the designed head and body phantom in relation to the Standard PMMA phantoms. This is given as:

Statistical analysis
The data analysis was performed using SPSS for Windows, Version 20.0 (SPSS Inc., Chicago, IL, USA). Descriptive and independent sample t-test was used at a 95% level of significance. P<0.05 was considered statistically significant.

Results
Scan protocols for both adult head and abdomen phantom were set at potential tube voltage of 120kvp for all centers while the tube current varied amongst the scanners.
A total of four (4) models were used in this study, with three (3) from the same manufacturer. The number of slices of the CT ranged from 4-64 slices with axial and helical modes [ Table 1]. The parameters for determining dose (mGy) were kVp, mAs, slice thickness, pitch, scan length and rotation time. The parameters for this study were statistically different from a one-way ANOVA (P < 0.001) [ Table 2]. respectively. An independent sample t-test shows that there were no differences between the mean dose for centers A and B for both head (P = 0.870) and body (P = 0.766) phantoms [ Table 3].
The percent deviation in CTDI vol for the standard phantom (SP) and constructed phantom (CP) for the head in center A and B were 4.6 and 5.5% respectively and the body were 9.20 and 0.00% respectively [ Table 4]. The correction factor (k) between the PMMA and constructed phantom for the head and body was around one.
There was no difference in the estimated dose to the head between the SP and CP (P = 0.948), and the estimated dose to the body between the SP and CP (P = 0.901) [ Table 4].
The percent deviation in CTDI vol in the control console for the standard phantom (SP) and constructed phantom (CP) for the head in centers A and B were 4.6 and 4.3% respectively and the body were 0.00 and 2.65% respectively. The percent deviation in DLP in the control console for the standard phantom (SP) and constructed phantom (CP) for the head in center A and B were 20.4 and 10.25% respectively and the body were -8.55 and 2.72% respectively. In addition, the CTDI vol in the control console for both phantoms for the head in centers A and B was statistically the same (P = 0.955). The CTDI vol in the control console for both phantoms for the body in centers A and B was statistically the same (P = 0.993). The DLP in the control console for both phantoms for the head in centers A and B was statistically the same (P = 0.857). The DLP in the control console for both phantoms for the body in centers A and B was statistically the same (P = 0.950) [ Table 5].     Table 5. Comparison of the uncertainty CTDI vol between the standard phantom with that of the locally constructed phantom for head and body.
SP= standard phantom, CP= constructed phantom The console displayed CTDI vol values and estimated CTDI vol for the standard head phantoms for center A were 72 and 66.97 mGy respectively [ Table 6]. The console displayed CTDI vol values and estimated CTDI vol for the standard head phantoms for center B were 26.82 and 23.39 mGy respectively [ Table 6]. The console displayed CTDI vol and estimated CTDI vol values for the standard body phantoms for center A were 22.8 mGy and 18.75 mGy. The console displayed CTDI vol and estimated CTDI vol values for the standard body phantoms for center B were 7.55 and 6.29 mGy. There was no difference between the CTDI vol from the console and that estimated from the TLD chips for the head of the standard phantom (P = 0.905). Similarly, there was no difference between the CTDI vol from the console and measured dose from the TLD chips for the body standard phantom (P = 0.813) [ Table 6]. There was no difference between the CTDI vol from the console and the estimated values from the TLD chips for the head of the designed phantom (P = 0.380). Similarly, there was no difference between the CTDI vol from the console and measured dose from the TLD chips for the body designed phantom (P = 0.774) [ Table 7].

Discussion
The uncertainty between the SP and CP with the head and body for centers A and B was within ±20%. Similarly, the console uncertainty between the SP and CP with the head and body for centers A and B was also within ±20%. A selftest of the constructed phantom for head and body shows that the maximum console CTDI vol and estimated CTDI vol with the TLD chips was <±20%. The uncertainty of the displayed console CTDI vol of the standard and constructed phantom for the head and body of centers A and B was < ±5%, while the TLD measurements for the head and body was < ±6%. Generally, the displayed console doses were higher with the constructed phantom compared to the standard phantom. Also, the estimated dose values were higher for the body of the CP compared to that of the SP. A major reason for this could be in the design and accuracy of the locally made phantom compared to the generally accepted PMMA phantom. Nevertheless, the results obtained from the locally made phantom proved useful alongside those of the PMMA standard phantom.
The results from this study were in line with the American College of Radiology (ACR) quality control manual CTAP reference value of ±20%. The ACR recommendation also states that percent deviation could increase to ±30-40% based on the manufacturer's specified tolerance limit [16,17].
Similarly, the obtained percent deviation in this study was within the IAEA acceptable limit of ±20% [19]. In addition, this study met the achievable criteria of ±10% set by the IAEA [19].
A study by Akpochafor et al., which developed a local phantom for dose verification, shows that the variation in doses between the standard and constructed head was ±15.2% [18]. This was higher than the variation in our study, which averaged to ±5%. Also, the percent deviation between the standard and constructed body phantom from Akpochafor's study was 5.3%, which was lower than that of our study, which averaged to ±9%. In most cases the variation between both results is largely dependent on the calibration factors of the TLDs, temperature conditions of the chips, uncertainty of the reader, and many other factors [10,19]. It is worthwhile to note that the displayed console dose is only an estimate from a cylindrical phantom of the CT algorithm, which assumes the patient size. In reality, the displayed console dose is not the same as the real patient dose. The mathematical computation from the displayed console dose (mGy) and TLD measured dose (mGy) with bot head and body phantom were seen to be within ±20% [19].

Conclusion
Our study has successfully verified the accuracy of the dose delivered with both the standard and locally constructed phantoms. The findings revealed that the uncertainty between both phantoms with TLD measurements was within the ±10% achievable criteria limit. The displayed console CTDI vol was within the ±20% acceptable criteria limit, proving the validity of the new phantom. Correction factors for the head and body were 0.998 and 1.05 respectively, which makes the phantom valid. With the above validation, the locally designed phantom can be used for CT dose assessment and for dosimetry measurements.