Validation of a method for estimating peak skin dose from CT‐guided procedures

Abstract A method for estimating peak skin dose (PSD) from CTDIvol has been published but not validated. The objective of this study was to validate this method during CT‐guided ablation procedures. Radiochromic film was calibrated and used to measure PSD. Sixty‐eight patients were enrolled in this study, and measured PSD were collected for 46 procedures. CTDIvol stratified by axial and helical scanning was used to calculate an estimate of PSD using the method [1.2 × CTDIvol(helical) + 0.6 × CTDIvol(axial)], and both calculated PSD and total CTDIvol were compared to measured PSD using paired t‐tests on the log‐transformed data and Bland‐Altman analysis. Calculated PSD were significantly different from measured PSD (P < 0.0001, bias, 18.3%, 95% limits of agreement, −63.0% to 26.4%). Measured PSD were not significantly different from total CTDIvol (P = 0.27, bias, 3.97%, 95% limits of agreement, −51.6% to 43.7%). Considering that CTDIvol is reported on the console of all CT scanners, is not stratified by axial and helical scanning modes, and is immediately available to the operator during CT‐guided interventional procedures, it may be reasonable to use the scanner‐reported CTDIvol as an indicator of PSD during CT‐guided procedures. However, further validation is required for other models of CT scanner.


| INTRODUCTION
The scope and complexity of computed tomography (CT)-guided interventions is increasing, 1 and CT is now used in combination with angiography in hybrid imaging systems. [2][3][4][5] Patient dose indices vary depending on the complexity of the procedure, 6 and patients can be exposed to high peak skin doses (PSD) during the most complex procedures, as the same skin site is scanned repeatedly. This is very similar to the pattern of irradiation during dynamic imaging, including CT perfusion. The potential effects of high PSD delivered during such procedures were demonstrated in the CT perfusion incidents that occurred in the mid-2000's, with patients experiencing transient and permanent alopecia 7 and an incident during which a young child experienced erythema after repeated CT of the head. Radiation cataractogenesis is a related concern if the area of repeated scanning includes the eye lenses.
Given the potential complications associated with high PSD experienced during CT, it is useful to understand how scanner-reported dose indices relate to patient skin and surface doses, as these indices are available to the physician during the procedure, reported in real-time. Soon after the CT perfusion incidents, studies examining patient doses resulting from CT brain perfusion were published, including a Monte Carlo study of patient lens and skin doses, 8 a phantom study of organ doses, 9 and a phantom study of skin dose. 10 While these studies examined only CT brain perfusion imaging using axial scanning mode, an expression relating patient skin dose to the volume computed tomography dose index (CTDI vol ) for interventional CT procedures was published around the same time, 11 based on previous evaluation of the relationship between CTDI vol and surface dose for different scanning modes. 12 This method was based on a linear weighted sum of CTDI vol (helical) and CTDI vol (axial) 1 . The weighting factors were determined from the definitions of CTDI and CTDI vol , the relationship between peripheral and central CTDI measurements in the 32 cm body phantom, and the relationship between peripheral CTDI and surface dose reported by Bauhs et al. 12 The method is as follows: This method, while incorporated into existing recommendations regarding CT-guided interventional procedures, 1 has never been validated.
The purpose of this study was to validate this existing method for calculating patient PSD during CT-guided interventional procedures. We hypothesized that the method would predict skin doses that were not significantly different from skin doses measured using radiochromic film.

| ME TH ODS
This study was conducted in compliance with the Healthcare Insurance Portability and Accountability Act and received prospective Institutional Review Board approval. Informed consent was obtained from all 68 patients enrolled in this study during a five month period.
All procedures were performed on a single Definition AS128+ CT (Siemens Healthineers, Malvern, PA) located in the interventional radiology department. Procedures performed included radiofrequency, microwave, and cryo-ablation.
PSD were measured using Gafchromic XR-RV3 film (Ashland, New Jersey). The film was cut into 35 cm × 5 cm strips. All film was from a single lot, and the film was calibrated using the same CT scanner used to perform the ablation procedures using standard methods. 13  traceable calibration at 120 kVp (HVL = 7.65 mm Al). Calibration films were exposed to air kerma ranging from 0 mGy to 4,009 mGy at 120 kV. The films were stored in a dark location for 2 weeks prior to scanning, as were films irradiated during patient procedures. All During CT-guided interventions, one strip of film was placed in the same location and orientation as for calibration, immediately under the patient. The technologist was instructed to identify the lesion to be treated during the procedure on prior cross-sectional imaging and then use anatomical landmarks to center the film, oriented along the long axis of the patient, as close as possible to the region of interest.
As PSD was the dosimetric quantity of interest, it mattered only that the film was irradiated during a single rotation during each scanning event, i.e., capturing the entire length of long helical scans was not required. All images from every procedure were reviewed to ensure that the film was irradiated during each scan event. The area of maximum film darkening was identified visually and a region of interest containing at least 6,750 pixels was placed to measure the film darkening with the same methods used for calibration. was adapted to patient size using technique charts. 16 Measured PSD were compared to calculated PSD and to total CTDI vol alone using paired t-tests on the log-transformed data. A simple Bonferroni correction was used to correct for multiple comparisons. Five statistical tests were performed during data analysis, resulting in a corrected alpha of 0.05/5 = 0.01. Bland-Altman analysis was used to examine bias, and correlation was calculated using the Spearman correlation coefficient.

| RESULTS
The film calibration data are plotted in Figure 1. One procedure was cancelled, the technologist forgot to place the film for 11 JONES ET AL.

| 225
procedures, and no exposure was recorded on the film (i.e., the film was placed outside the imaged field of view) for 10 procedures, leaving 46 procedures available for analysis. Measured PSD ranged from 30.5 mGy to 1,303.9 mGy, with a median of 263.1 mGy.
The fraction of CTDI vol contributed by axial scanning varied widely (Fig. 2). Calculated PSD were significantly different from measured PSD (P < 0.0001). Calculated PSD underestimated the measured PSD by 18.3% (Fig. 3a, 95% limits of agreement, −63.0% to 26.4%). The standard deviation of the differences between calculated PSD and measured PSD across all procedures was 20.5 mGy.
Total CTDI vol was not significantly different from measured PSD (P = 0.27) and underestimated the measured PSD by 3.97% (Fig. 3b, 95% limits of agreement, −51.6% to 43.7%). The standard deviation of the differences between CTDI vol and measured PSD across all procedures was 24.1 mGy.

| DISCUSSION
In the specific context of this study, CTDI vol alone was a better predictor of PSD during CT-guided interventions than PSD calculated using an existing method. The 95% limits of agreement for the difference between CTDI vol and measured PSD indicated that CTDI vol can predict the PSD within 50%, a reasonable benchmark for patient dosimetry with indirect dose indices. 17 While the existing method also, in general, predicted PSD within 50%, bias was greater, and use of the existing method required that CTDI vol be stratified into contributions from axial and helical scanning, which is somewhat cumbersome and difficult to do during the procedure at the scanner console.
It has been reported previously that surface dose as a percentage of CTDI vol for axial scanning increases with decreasing phantom size, from 49% to 65% for phantom lateral widths ranging from 50 cm to 25 cm. 18 There was a weak, non-significant correlation (ρ= 0.30, P = 0.041) between percent error in calculated PSD compared to measured PSD and patient effective diameter (Fig. 4a).
There was no correlation (ρ= 0.25, P = 0.093) between percent error in CTDI vol compared to measured PSD and patient effective diameter (Fig. 4b). There was a weak, significant correlation (ρ= 0.44, P = 0.0023) between percent error in CTDI vol compared to measured PSD and the fraction of CTDI vol contributed by axial scanning (Fig. 4c). These findings likely reflect the complex interactions among the factors that affect the relationship between patient surface dose and CTDIvol, including scanning mode, patient size, and the position of the patient relative to isocenter, as patient access and probe placement considerations often mean that the patient is not centered in the gantry.
There are several uncertainties associated with radiochromic film dosimetry, these have been discussed extensively in a review by McCabe et al 13 and summarized elsewhere. 19 Considering that calibration of the film for the present study was done using backscatter conditions, the accuracy limits on measured PSD in this study were approximately AE10%, as calculated using the summary information in

| CONCLUSION
Peak skin doses (PSD) resulting from CT-guided procedures, calculated using an existing method, were significantly different from measured PSD for procedures performed using a single make and model of CT scanner. Total CTDI vol was not significantly different from measured PSD in the specific context of this study.
F I G . 3. Bland-Altman plots of (a) percent difference between peak skin dose (PSD) calculated using an existing method and measured PSD and (b) percent difference between total CTDI vol and measured PSD. The dashed line indicates bias and the dotted lines indicate the 95% limits of agreement (LoA), and the x-axis title "Average (mGy)" is the average of the estimates produced by the two measurement methods.
F I G . 4. (a) Percent error in PSD was calculated using an existing method compared to measured PSD versus patient effective diameter; (b) percent error in CTDI vol compared to measured PSD versus patient effective diameter; and (c) percent error in CTDI vol compared to measured PSD versus fraction of CTDI vol contributed by axial scanning.