Impact of patient centering in CT on organ dose and the effect of using a positioning compensation system: Evidence from OSLD measurements in postmortem subjects

Abstract The purpose of this study was to investigate the frequency and impact of vertical mis‐centering on organ doses in computed tomography (CT) exams and evaluate the effect of a commercially available positioning compensation system (PCS). Mis‐centering frequency and magnitude was retrospectively measured in 300 patients examined with chest‐abdomen‐pelvis CT. Organ doses were measured in three postmortem subjects scanned on a CT scanner at nine different vertical table positions (maximum shift ± 4 cm). Organ doses were measured with optically stimulated luminescent dosimeters inserted within organs. Regression analysis was performed to determine the correlation between organ doses and mis‐centering. Methods were repeated using a PCS that automatically detects the table offset to adjust tube current output accordingly. Clinical mis‐centering was >1 cm in 53% and 21% of patients in the vertical and lateral directions, respectively. The 1‐cm table shifts resulted in organ dose differences up to 8%, while 4‐cm shifts resulted in organ dose differences up to 35%. Organ doses increased linearly with superior table shifts for the lung, colon, uterus, ovaries, and skin (R 2 = 0.73–0.99, P < 0.005). When the PCS was utilized, organ doses decreased with superior table shifts and dose differences were lower (average 5%, maximum 18%) than scans performed without PCS (average 9%, maximum 35%) at all table shifts. Mis‐centering occurs frequently in the clinic and has a significant effect on patient dose. While accurate patient positioning remains important for maintaining optimal imaging conditions, a PCS has been shown to reduce the effects of patient mis‐centering.

filters that shape the x-ray beam to compensate for variations in patient attenuation. 6,7 In the thickest, central, areas of the patient, a thinner segment of the bowtie filter is used to allow for maximum beam intensity through the anatomical regions with higher attenuation, and in the peripheral areas of the patient, a thicker segment of the filter is used to reduce the beam intensity through the anatomical regions with lower attenuation.
The optimal function of a bowtie filter and TCM techniques require that the patient is centered appropriately in the CT gantry. [8][9][10][11][12][13][14] When a patient is placed on the CT table, the technologist should attempt to position the patient in the center of the gantry using the gantry-mounted laser system. This includes aligning the midline of the patient (from the nose to the pubic symphysis) with the central laser and changing the table height so that the center of mass of the anatomy to be scanned is in the center of the gantry. Next, the technologist acquires a localizer radiograph, which serves to measure patient attenuation for proper TCM, as well as help verify correct patient positioning. If necessary, the technologist should correct the patient's position and acquire a new localizer radiograph.
However, studies have shown that technologists do not always correctly center the patient within the gantry. [9][10][11] This is because patients are not perfectly cylindrical, and it can be difficult to define the center of the patient, especially those who are not lying flat due to physical constraints or examination requirements. Inaccurate centering of the patient in the gantry affects the attenuation of the xray beam and apparent size of the patient, affecting radiation dose and image noise. [11][12][13][14] The bow-tie filter shapes the intensity of the x-ray beam assuming that the thickest region of the patient is located in the center of the beam. If the patient is correctly placed in the center of the gantry, the center of the patient will receive the maximum x-ray intensity when the x-ray tube is rotating in the gantry while scanning. This is shown in Fig. 1(a) with the x-ray tube in the anterior and lateral position. However, if the patient is shifted away from center, the patient will receive a different dose distribution. If the patient is shifted anteriorly away from center, as shown in Fig. 1(b), the anterior organs will receive higher dose when the x-ray tube is in the anterior position, but they will also receive lower dose when the xray tube is in the posterior position, compared to when the patient was centered. Alternatively, if the patient is shifted posteriorly away from center, as shown in Fig. 1(c), the anterior organs will receive lower dose when the x-ray tube is in the anterior position, and higher dose when the x-ray tube is in the posterior position, compared to when the patient was centered. Furthermore, in either the anterior or posterior mis-centering scenarios, the patient is shifted away from the center and towards the thicker regions of the bowtie filter when the x-ray tube is in the lateral positions. This results in a reduced x-ray intensity to the majority of the patient's anatomy, as shown in the bottom diagrams of Figs. 1(b) and 1(c). As a result, patient centering is necessary for optimal dose management. [8][9][10][11][12][13][14] Furthermore, if a patient is positioned at the center of the gantry, the magnification in the localizer radiograph is accurate, as shown in Fig. 2(a), and therefore the TCM system will function optimally. However, if the patient is positioned too close to the xray source, the localizer radiograph will experience greater magnification, leading to an overestimation of patient size, as shown in Fig. 2 ( b). This causes the TCM system to transmit a higher tube current, resulting in an unnecessary increase in radiation dose. Alternatively, if the patient is positioned too far from the x-ray source, the localizer radiograph will be under-magnified, shown in Fig. 2(c), resulting in underestimation of attenuation and reduced tube current, potentially resulting in noisy images.
In order to compensate for such magnification effects, a commercially available positioning compensation system (PCS; Auto Couch Height Positioning Compensation, Canon Medical Systems, Otawara, Japan) has been integrated to work with TCM (Sure Exposure 3D, Canon Medical Systems, Otawara, Japan). 15 This system is a vendor-proprietary software included in the scanner that automatically detects the offset between the patient's position and the center of the gantry in order to estimate accurate patient size and attenuation. 15 Figure 3 shows that PCS corrects the magnification effects in the localizer radiograph when the patient is positioned too close does not alter the appearance of the patient size in the localizer radiograph, it instead communicates with the TCM system that the patient is actually smaller (or larger) than that represented in the localizer radiograph, and in turn, the TCM system produces new tube current maps in the x, y, and z planes using the corrected patient size provided by the PCS. As a result, using the corrected patient attenuation information, the PCS software can remove the effects of magnification and communicate with the TCM system to deliver proper tube current values, even for mis-centered patients.
The purpose of this study was to investigate the frequency and  considered to be perfectly centered in the gantry, and labeled as having zero shift, if their shift from the center of the gantry was <0.5 mm. All mis-centering distances >0.5 mm were recorded and later binned into 1-cm groups. Results were analyzed to determine the frequency and magnitude of clinical mis-centering. Frequency and magnitude were also compared between females and males to investigate whether breasts in females reduced positioning errors.
The magnitude of clinical mis-centering was also correlated with patient effective diameters to determine whether larger patients were mis-centered by greater magnitudes due to the potential increased difficulty in visually determining the center of a larger patient compared to a smaller patient. The effective diameter was measured for each patient as the square root of the product of their measured AP and LAT diameters, as described in Report 204 of the American Association of Physicists in Medicine. 16  Optically stimulated luminescent dosimeters (OSLDs) were inserted into the lungs, breasts, liver, stomach, colon, uterus, and ovaries and placed on the skin regions included in the scan range utilizing the methodology described by Griglock et al. 17 Organ doses were measured at each table position and corrected for energy and scatter response. 17

2.C | Statistical analysis
A student's t-test was used to assess differences in mis-centering distances between the male and female patients in the study. Pearson correlation coefficients (R 2 ) were used to evaluate the relation- The confidence levels of 95% were calculated, and a two-tailed P < 0.05 was considered to indicate statistically significant differences.   F I G . 9. Tube current values along the zaxis of Subject 3 for a centered scan and scans shifted 4 cm anteriorly acquired with and without the positioning compensation system. Subject 3 for the centered scan and the scan shifted 4 cm anteriorly both with and without the PCS. The PCS brought the tube current values closer to those used when the subject was centered, reducing the absolute percentage difference between the shifted and centered scans (mean 6%, maximum 19%) compared to when the PCS was not used (mean 9%, maximum 34%).

3.C | Organ dosimetry
All organ dose measurements are presented in milligray (mGy) in Table 1. When the PCS was not used, significant correlations were found between organ doses and vertical table shifts for all subjects in five out of eight organs including the lung, colon, uterus, ovary, and skin, demonstrating an increase in organ dose with increasing vertical shift (R 2 = 0.73-0.99, P < 0.005), shown in Fig. 10(a). For these five organs, organ dose differences relative to the centered position ranged from −35.0% to 22.0%. Figure 11 displays organ dose percent differences in these five organs, averaged across the three postmortem subjects (Standard Deviation 0.008-0.099) measured at each table shift position ranging from 4 cm posterior to 4 cm anterior (R 2 = 0.90-0.99, P < 0.005). The liver, stomach, and breast organ doses had smaller relative organ dose differences, ranging from −13.0% to 15.0%, compared to the other five organs mentioned above. Strong linear correlations were observed for the liver (R 2 = 0.71) and stomach (R 2 = 0.94) in Subject 1 and for the breast (R 2 = 0.59) in Subject 3, but correlations were weak for these organs in the other subjects (R 2 = 0.01-0.50), as shown in Fig. 12.
When the PCS was utilized, organ doses were observed to decrease as a function of vertical shift, as the PCS attempted to compensate for the table shift, as shown in Fig. 10(b). Significant correlations were observed between vertical shift and organ dose differences relative to organ doses measured at the center of the gantry in all three subjects for the skin (R 2 = 0.73-0.86, P < 0.005) and colon (R 2 = 0.90-0.98, P < 0.005) but did not have a strong correlation for the lungs, uterus, or ovaries (R 2 = 0.01-0.30, P < 0.005).  | 147

| DISCUSSION
The frequency of clinical mis-centering of patients undergoing CT examinations in our institution was slightly less than those found in other studies. For example, one study reported that 81% of patients were vertically mis-centered by >0.5 cm, 10 or 8% greater than our findings, while another study reported that 74% of patients were mis-centered by >1 cm, 6 or 21% greater than our findings. The frequency of clinical mis-centering is likely dependent on the technologist training methods and years of experience, so it is not expected that one hospital would have the same frequency of clinical patient mis-centering as another. In addition, it was concerning to find that 23 patients (7.7%) were mis-centered vertically by >3 cm, as one may expect this to be detected by the technologists. However, similar findings were also reported in other studies, where one group found that 17% of their patients were mis-centered by >3 cm, 10 and another group reported that 22% of their patients were mis-centered by >3 cm with maximum errors ranging from 6.6 cm posterior to 3.4 cm anteriorly. 6 Our findings demonstrated that mis-centering was more pronounced in the vertical direction than in the lateral direction, and showed greater prevalence for posterior mis-centering (54% of patients) rather than anterior mis-centering (46% of patients), in agreement with other studies. 10 The CTDI vol increased linearly with anterior table shift without PCS because magnification caused the subject to appear larger at table positions closer to the x-ray tube, as shown in Fig. 7(a), and therefore the scanner responded by increasing tube current output.
A similar trend in increasing CTDI vol with increasing anterior shift was reported by the manufacturer when they scanned a phantom at different  It can be argued that we would expect for the PCS system to correct the CTDI vol at all table shifts to match the CTDI vol produced at center. However, neither we nor the manufacturer observed this.
In fact, it appears that rather than normalizing the values to be consistent, the system slightly overcompensated the output in the opposite direction, resulting in a decreasing trend of CTDI vol with table height. However, both this study and the manufacturer found that the change in CTDI vol was greater without PCS (range 7%-10%) than with PCS (range 2.8%-5.7%).
Although the PCS did not normalize CTDI vol and organ doses to be equivalent to the scan acquired at the center of the gantry, it was able to provide smaller organ dose differences. For example, the PCS reduced the average dose difference at 4 cm posterior shift from 11% to 2%, shown in Fig. 14(a), and also reduced the maximum dose difference among all subjects and organs from 35% to 18%, shown in Fig. 14(b).
In addition to the magnification in the localizer radiograph, there are other complex mechanisms in the scanner that affect patient dose, such as the bowtie filter. When the center of mass is shifted away from the gantry, some organs will be shifted towards the center of the gantry, and therefore will experience an increase in dose due to the bowtie filter allowing maximum beam intensity at the central region of the beam, as shown in Fig. 1.
While the PCS alters the tube current to correct for patient magnification, it does not modify the shape or position of the bowtie fil-  There are limitations to this study. The first is that this study did not measure organ doses for lateral shifts. Mis-centering in the lateral direction will also produce magnification in the lateral localizer radiograph, affecting the TCM output. Therefore, it is expected that a lateral shift would also affect organ doses, with effects depending on whether the patient was shifted towards or away from the stationary x-ray tube during the acquisition of the lateral localizer radiograph, as well as distribution of organs within the anatomy. We chose to focus on reporting the effects of vertical mis-centering as these table shifts are more pronounced and prevalent, as found in the first part of this work as well as in other studies. Furthermore, shifting bilateral organs towards or away from center would introduce additional complex effects due to the bowtie filter, with one side of the organ experiencing a different effect than the other side. Second, this study did not evaluate the effect of mis-centering on image quality. Other studies have shown that mis-centering has an effect on both patient dose and image quality. For example, a study by Toth et al. 6 reported that a vertical shift of 6 cm increased image noise by up to 30% and a study by Kaasalainen et al. 9 reported that a vertical shift of 6 cm increased image noise by up to 28%. However, both of these studies conducted noise measurements in phantoms, and have not evaluated the effect of patient mis-centering on diagnostic image quality.

| CONCLUSION
In conclusion, this study has shown that patient mis-centering occurs frequently in the clinic and impacts organ doses. It remains essential for technologists to strive for accurate patient positioning at the center of the CT gantry. In the events where positioning is not performed correctly, the commercially available position compensation system can automatically detect mis-centering and modify the scan techniques to improve acquisition techniques for optimal scanner performance.

CONFLICT OF INTEREST
Co-author Manuel Arreola has been a recipient of research support from Canon Medical Systems, USA. All other authors have no disclosures.